Key Run Blocks - let's figure this thing out once and for all...

[QuikSand]

Okay, I have heard a variety fo comments in various places about key run blocks. We all basiclaly understand (or think we understand) the nature of the stat... but there are some oddities that seem to happen in there that make me wonder what's up.

So... let's work through this thread, and figure out, as completely as we can, what's going on with the KRB phenomenon in FOF 2007.

If you want to contribute knowledge here from what you have studies, seen, or recorded from FOF 2004, that's fine - but be sure to mark it clearly as such. Just in case there are some changes in the game engine in this respect from 2004 to 2007, I don't want to get lost on that account.

With that... what do we know?

[QuikSand]

A few questions to be answered (and shouldn't be tough):

When is a KRO generated for a play?

-It seems that there is exactly one player assigned a key run block opporunity on most runnning plays.

-I am quite certain that there are ever more than one assigned on a play.

-I think there are some exceptions to the 1:1 ratio -- things like QB scrambles don't seem to trigger a KRO. I suspect there may be more exceptions, but they are a pretty distinct minority.

-Is it all QB carries that fail to create a KRO, but all other carries that do? Or more complex than that? What about draws? Reverses? Called QB rushes? Kneel-downs?


What sort of plays always trace back to a successful key run block?

I think this is more at the heart of my thining here. I believe, though am not 100% certain, that every run that nets a first down or a touchdown will result in a KRB being credited to someone on the offense.

Do we have any real insight here? Can anyone refute the above? It just takes one counterexample.

A related question, which I will place under here, is: what is the potential universe of results from a KRB in FOF 2007? I seem to recall some notion that a KRB in FOF 2004 resulted in a running play of at least X yards (5?), but in this game, I'm pretty certain that converting a 3rd and 1 for exactly the 1 yard needed will generate a successful KRB. Is there some layered, but definable, way to sort out what happens when the KRB is successful?


I am also open to the results of this diggins sending us in a rather counter-intuitive direction... i.e. that the KRB is actually an in-game sult of a good running play, rather than the other way around.


I suspect there are more factual questions to be tackled here... but all input is welcome.

[Vinatieri for Prez]

For 2K7, I can only add anecdotally that TDs and 1st downs generate KRBs every time for me (even 1-yard runs). As for particular yardage in other instances, I can't peg it down. I have seen a run as long as 9 yards occur without a KRB, and many times a 6-yarder has failed to register a KRB. I also see a KRO registered for each run, except QB scrambles, as you noted; this has occurred even when some of my runs have included finesse/counterplays. Again, this is all looked at in a very elemental review of box scores.

[MrBigglesworth]

For 2K7, I can only add anecdotally that TDs and 1st downs generate KRBs every time for me (even 1-yard runs). As for particular yardage in other instances, I can't peg it down. I have seen a run as long as 9 yards occur without a KRB, and many times a 6-yarder has failed to register a KRB. I also see a KRO registered for each run, except QB scrambles, as you noted; this has occurred even when some of my runs have included finesse/counterplays. Again, this is all looked at in a very elemental review of box scores.
I've always assumed this but never researched it, but is every KRB mentioned in Solevision?

[Vinatieri for Prez]

I've always assumed this but never researched it, but is every KRB mentioned in Solevision?

Again, my limited review of 3 game logs/solevision show this to be true.

I also point out that when you get the message in Solevision like "he knocked him flat" or "threw him to the ground," the lineman gets credited also with a pancake block. (i.e. the lineman gets credited with a KRB and a pancake, not just the pancake). Or in other words, all pancake blocks are KRBs.

Again, this is with limited anecdotal review, no testing.

[Eaglesfan27]

A few questions to be answered (and shouldn't be tough):

When is a KRO generated for a play?


-Is it all QB carries that fail to create a KRO, but all other carries that do? Or more complex than that? What about draws? Reverses? Called QB rushes? Kneel-downs?



I'm positive that I have seen a few QB carries that did result in a KRO. It's very rare and it always says QB draw on the play, but it does happen. I'll try to find the game log where that occurred, but it's been a few years in my current SP game since I last saw it.

[Vinatieri for Prez]

I should note that I have seen many 5-yard runs have a KRB credited, even when the run does not go for a 1st down or TD. However, as I stated earlier, I have seen runs of more than 5 yards (often with 6-yard runs) not have a KRB credited.

So, this would seem to show there is more at work than just a basic yardage requirement for KRBs (when there is no TD or 1st down).

[Vinatieri for Prez]

I'm positive that I have seen a few QB carries that did result in a KRO. It's very rare and it always says QB draw on the play, but it does happen. I'll try to find the game log where that occurred, but it's been a few years in my current SP game since I last saw it.

This would seem to suggest that designed QB carries get KRO, but impromptu scrambles do not.

[Eaglesfan27]

This would seem to suggest that designed QB carries get KRO, but impromptu scrambles do not.


Agreed.

[Vinatieri for Prez]

One issue I am interested in too, is what decides which player gets the KRO. I think it is clear from cursory glance of the game logs that the side of the run does not entirely determine who gets the KRO. I've seen the LT get a KRO for a run inside the LG. I've seen a RG get a KRO for a run outside the RT. I've seen a LG get a KRO for a run inside the RG. I've even seen a LG get credit for the KRO for a run around the right end.

Obviously, in real life, you have pulling guards, etc. which explain this. I've always wondered if there are actual play calls being made in-game that determine this or whether it's just a simple die roll along the line with the odds of each player getting the KRO determined somewhat from the direction of the run.

I'd also add that I'm not so sure the run direction has a very strong bearing. For instance, a recent game had:

runs left: 5
runs middle: 14
runs right: 2

Using the game log, I charted them with more specificity as:

left end: 3
outside left tackle: 2
inside left tackle: 4
inside left guard: 4
inside right guard: 1
inside right tackle: 5
outside right tackle: 0
right end: 2

Total: 21

However, my starting LT (who does not play backup at any position and participated in 19 of 21 run plays) was credited with 10 of 21 KRO. Even if you give him all of the left end and outside left tackle KROs for 5, he still gets 5 of the remaining 16 KROs on which he was on the field for only 14 of the run plays. Now, I know my LG got the KRO for one of the inside left tackle runs (because he got a KRB for it). This means that you have to credit the LT with all of the remaining 3 inside left tackle runs AND 2 of the KRO for the runs starting at LG moving to the right.

This is purely anectodal of course off a limited sample of 1 game log in 2K7. But, it buttresses something I have always thought, that while run direction determines KROs somewhat, linemen with better run blocking get a higher proportion of the KROs. You see, my LT is rated 100 in run blocking (while my LG is 76, C is 55, RG is 63, and RT is 42).

Anyways, just food for thought.

[Narcizo]

I seem to recall some notion that a KRB in FOF 2004 resulted in a running play of at least X yards (5?), but in this game, I'm pretty certain that converting a 3rd and 1 for exactly the 1 yard needed will generate a successful KRB.

In FOF2004 a key run block was awarded if the run went for 5 yards or over and/or it achieved a first down or touchdown. As Vinny has noted runs can go for over 5 yards and not get a key run block awarded in FOF2007. However I have never seen a run for less than 5 yards (which didn't result in a 1st down/touchdown) get a key run block awarded.

[Vinatieri for Prez]

So, I decided to chart the previous game as well for comparison:

run left: 9
run middle: 6
run right: 4

Via the game log, charted specifically as:

left end: 6
outside left tackle: 3
inside left tackle: 2
inside left guard: 2
inside right guard: 1
inside right tackle: 1
outside right tackle: 2
right end: 2

Total: 19

This time, my starting stud LT was credited with only 4 of 19 KROs (he participated in 18 of the 19 run plays). This is the case even though there were 6 runs around left end and 3 around left tackle. So, in this game, we can't even credit him with half of the 9 runs going LE/Out LT, but yet in the other game, we had to credit him with all 5 of the LE/Out LT runs and then numerous others just to have it add up.

I would also add that my LG alone registered KRBs for runs all the way from LE to RE.

It seems there is a complete disparity of how KROs are calculated. One I have yet to figure out. Of course, this particular game log completely blows out of the water my theory that the higher the run blocking rating, the more KROs you get.

[Ben E Lou]

I should note that I have seen many 5-yard runs have a KRB credited, even when the run does not go for a 1st down or TD. However, as I stated earlier, I have seen runs of more than 5 yards (often with 6-yard runs) not have a KRB credited.

So, this would seem to show there is more at work than just a basic yardage requirement for KRBs (when there is no TD or 1st down).My observation is that it is like this:

Five-yard run on first and 10: KRB awarded
6-10 yard run on 2nd or 3rd and Long that does NOT result in first down or TD: No KRB awarded.

In the OSFL sim last night, I definitely had a 10-yard run by a RB on a 2nd and very long situation that did *not* result in a KRB awarded. However, I have seen several 5-yard runs on 1st and 10 that result in KRBs.

[Vinatieri for Prez]

My observation is that it is like this:

Five-yard run on first and 10: KRB awarded
6-10 yard run on 2nd or 3rd and Long that does NOT result in first down or TD: No KRB awarded.

In the OSFL sim last night, I definitely had a 10-yard run by a RB on a 2nd and very long situation that did *not* result in a KRB awarded. However, I have seen several 5-yard runs on 1st and 10 that result in KRBs.

Yep, I think you're onto something. Looking at those two same game logs, the 6-yard runs that did not have a KRB were on 2nd and 10; 2nd and 10; and 2nd and 7.

Note that even a 6-yard run with 7 yards to go did not register a KRB. So, 7 yards to go crosses the threshold of 2nd and long. I would imagine then that the remaining question is the 10+ yard rule you have for 2nd or 3rd and long, or if it doesn't matter at all how long the run is on 2nd or 3rd and long, if it doesn't get a 1st down or TD, it will be no KRB.

Finally, I had an 8-yard run on 1st and 16 register as a key run block, so perhaps distance to go on 1st down does not matter.

[Vinatieri for Prez]

I would imagine then that the remaining question is the 10+ yard rule you have for 2nd or 3rd and long, or if it doesn't matter at all how long the run is on 2nd or 3rd and long, if it doesn't get a 1st down or TD, it will be no KRB.

SD is right, it's likely the former. I had a KRB registered for a 17-yard run on 2nd and 19.

[QuikSand]

My observation is that it is like this:

Five-yard run on first and 10: KRB awarded
6-10 yard run on 2nd or 3rd and Long that does NOT result in first down or TD: No KRB awarded.

In the OSFL sim last night, I definitely had a 10-yard run by a RB on a 2nd and very long situation that did *not* result in a KRB awarded. However, I have seen several 5-yard runs on 1st and 10 that result in KRBs.

The guys over at Football Outsiders have a grading for running plays that result in "success" which comes generally close to this sort of measure. Maybe Jim's game decides that if the run was a "success" (either by the FO metric, or by something similar of his own) then the blocking is rewarded with a KRB, and if not, not.

Once again, pointing toward the KRB stat being an outcome of a running play, rather than being an input into its result.

[QuikSand]

So, a hypothetical model for how this might work...

On any given run play:

-an aggregate "run blocking" rating for your offense and the defense is determined, perhaps with added weight to the players in the immediate zone where the run is to take place

-dice are rolled to determine the net effect of the blkocking scheme

-the ball carrier's skills are checked to determine how well he takes advantage of the setup

-the play nets a certain result

-the game goes back and assigns a tackle to somebody on the defense

-the game goes back and assigns a key run block opportunity to somebody, presumably weighted toward the immediate zone where the run took place (but not exclusively so)

-if the run met a certain "success" threshhold, then the KRO for that player is considered to have succeeded, and the KRB is awarded and tracked



I don't yet have a notion of where to fit in the "pancake" blocks, so I have left them out for now. Is this, essentially, how we think the engine works?

And if so... anyone have a philosophical problem with that?

[MIJB#19]

I think the same order of things happens as you posted QS, yet I'll provide a bit of an alternate theory:

-an aggregate "run blocking" rating for your offense and the defense is determined, perhaps with added weight to the players in the immediate zone where the run is to take place

-dice are rolled to determine the net effect of the blkocking scheme

-the ball carrier's skills are checked to determine how well he takes advantage of the setup

-the game assigns a key run block opportunity to somebody, presumably weighted toward the immediate zone where the run took place (but not exclusively so)

-the game assigns a key run block defender, presumably weighted toward the immediate zone where the run took place (but not exclusively so)

-the game assigns a potential key run block (and pancake block?) if the run blocker beats the run defender

-the play nets a certain result

-the game assigns a tackle to somebody on the defense

-if the run met a certain "success" threshhold, then the KRO for that player is considered to have succeeded, and the KRB is awarded and tracked

[Ben E Lou]

So, a hypothetical model for how this might work...

On any given run play:

-an aggregate "run blocking" rating for your offense and the defense is determined, perhaps with added weight to the players in the immediate zone where the run is to take place

-dice are rolled to determine the net effect of the blkocking scheme

-the ball carrier's skills are checked to determine how well he takes advantage of the setup

-the play nets a certain result

-the game goes back and assigns a tackle to somebody on the defense

-the game goes back and assigns a key run block opportunity to somebody, presumably weighted toward the immediate zone where the run took place (but not exclusively so)

-if the run met a certain "success" threshhold, then the KRO for that player is considered to have succeeded, and the KRB is awarded and tracked



I don't yet have a notion of where to fit in the "pancake" blocks, so I have left them out for now. Is this, essentially, how we think the engine works?

And if so... anyone have a philosophical problem with that?If that's how it works, then yes, I would. KRBs and tackles shouldn't be assigned after the fact. They should be first causes.


The flow should be something along these lines:


1. check key blocker(s) vs. key defender(s) at point of attack for strength vs. strength ("dissolves play" or pancake). If it's a run outside LT, for example, and the strong side is the right, and the FB is in the backfield, then the key matchups at the point of attack would probably be FB vs.WLB and LT vs. RDE.
1a. If playdissolved=Yes, then playdissolver vs. RB is checked for possible tackle in backfield. In FOF, where there's no tackling or breaking tackle rating, I guess that would need to be a matchup of strength versus elusiveness. If tackle is broken, assign penalty to blocking scheme overall impact, and continue wiht play resolution. If tackle is made, play ends here.
1b. If pancake=Yes, then assign KRB to pancaker and add bonus to blocking scheme overall impact, and continue with play resolution.
1c. If both=No, then continue with play resolution.
2. Check key matchups for KRB. If KRB=Yes, then bonus to overall impact. If KRB=No, then penalty to overall impact.
3. check blocking scheme overall impact, weighted more heavily for key blockers/defenders, applying bonus/penalties from above
4a. Check hole recognition, and if KRB=Yes or Pancake=Yes, then assign very high bonus. (Since the KRB/Pancake is taking place where the ball carrier was supposed to go in the first place, it should be pretty rare for him to miss THAT hole.). If recognition result is positive set creasefound=Yes.
4b. If KRB and Pancake both are No, then carrier still has a chance to find a little crease. Higher chance creasefound=Yes in these situations for those with high hole recognition.
5. If creasefound=No, then it's a low gain run.
6. If creasefound=Yes, then start looking at power inside, breakaway running, and elusiveness for potential big gains.

This could get very drawn out, but the point of it all is that if we're going to properly evaluate our offensive linemen for run blocking, and KRB is the primary stat we have to do so, then the KRB should be assigned on the front end of the play, based on whether or not he did his job, rather than on the back end, partially based on whether or not the RB did his job.

[Ben E Lou]

Dola:

There should be cases (besides 3rd and 1) where there's a KRB credited, but the RB only gets a yard or two, because he missed the whole. There should also be cases where there's no KRB credited, but the RB gets yardage on his own. As far as I can tell, neither is possible in FOF at the moment.

[QuikSand]

-the game assigns a potential key run block (and pancake block?) if the run blocker beats the run defender

-the play nets a certain result

-the game assigns a tackle to somebody on the defense

-if the run met a certain "success" threshhold, then the KRO for that player is considered to have succeeded, and the KRB is awarded and tracked

So, you're suggesting that there are two decisions for the KRB stat... one that would happen before the play is resolved, and would have effect on the play outcome, and then another after the play is resolved, to ex post facto assign credit for a successful play even if the KRB wasn't generated beforehand?

I guess that is also consistent with the notion that certain successful plays *always* generate a KRB. I hadn't thought of this being a multi-step process, but we can't rule it out, as best I can tell.

[QuikSand]

There should also be cases where there's no KRB credited, but the RB gets yardage on his own.

To me, this is the most frustrating piece of it. As it is, I'm starting to wonder how much you can use an individual player's KRB stats as a measure of his own performance... all indications point to this being very much a team stat, with individual credit apportioned in some rough way back to the players, but very much a function of a lot more than that one guy. An excellent back, or excellent OL colleagues, would certainly seem to boost any OL player's KRB numbers, it seems to me, and the converse is true as well.

[QuikSand]

One other thought... it's at least possible, within the model that I am using above, that the KRO-selected player is determined before the play happens, and that it's only the success of the KRO that is determined afterwards. It's at least possible. More to come... I'm expanding on that notion in a moment.

[QuikSand]

So, let's put numbers to this, even. All hypothetical, not based on anything empirical, just some intuition.

First, let's say that every run blocker has some consolidated rating I'll call Run Blocking Effectiveness, of RBE. RBE would presumably be some sort of weighted and situation-specific result of combining his skill in run blocking, along with other relevant skills like blocking strength and who knows what else, modified for the game situation (fatigue at that point might reduce his effectiveness), player readiness/rustiness and experience at the position he's playing, and so forth. But at ant point in any game, a given player has one number that encapsulates all this stuff, for now we'll call it RBE.

As the play is selected, and it's a designed run play, the game comes up with a formula to determine the overall RBE of the offense for this play. Maybe it looks something like this:

Offensive RBE=
sum of each OL's individual RBE
plus 1/2 the TE's and FB's RBE (for now I'll assume there's just one of each)
plus the "key" player selected for a KRO's RBE x2

So, if each player's RBE is on a scale of 0-100, then this is essentially the sum of eight such ratings -- which would either put it onto a scale of 0-800, or perhaps it would be rescaled back to 0-100 by dividing by 8. (Fair to question here whether adding additional run blockers who are less effective than the average of those around them ought to help or hurt the calculation here --I'll leave this note as a placeholder for an open question on that point)

Anyway -- so, let's assume that this calculation is made in something like this fashion -- every blocker counts, but one guy (the KRO guy for the play) is selected to have extra weight for determining the outcome of the play. It all gets bundled up into one consolidated rating, that gets used against a comparably-calculated Defensive RBE (or RDE) in determining the base outcome for the play itself.

Presumably, these factors actually end up creating (or adjusting an already-created) a mean and standard deviation for the play outcome. Then the game rolls dice to determine what happend on *this* particular play.


So, back to our matters of philosophy. If the system works this way, more or less, it would suggest that the selected "key" player actually does have a disproportionately large effect on this particular play (I gave him triple weight, but that was arbitrary -- it could be 2X, 3X, 10X or anything else, I suppose). And if that is so, then going back and giving him credit for the outcome of the play doesn't seem too absurd ot me... even though the assigning of after-the-fact KRB credit seems absurd on its face. It might just be another case of a FOF stat (or rating) that might have some intuitive explanation, it's just that it's not the explanation that you or I might draw form the title.

Rather than "key run blocks," perhaps this stat truly represents something more like "successful runs behind this player as key blocker." Which, itself, wouldn't be a terrible stat (in my view, at least) but just isn't the same thing that I think of with the label it's given in the game.


Just a bit more to chew on.

[Ben E Lou]

Rather than "key run blocks," perhaps this stat truly represents something more like "successful runs behind this player as key blocker." Which, itself, wouldn't be a terrible stat (in my view, at least) but just isn't the same thing that I think of with the label it's given in the game.Your overall postulation makes sense, and the bolded section here is consistent with several other things in FOF. Maybe you're on to something. {shurg}

[MIJB#19]

So, you're suggesting that there are two decisions for the KRB stat... one that would happen before the play is resolved, and would have effect on the play outcome, and then another after the play is resolved, to ex post facto assign credit for a successful play even if the KRB wasn't generated beforehand?

I guess that is also consistent with the notion that certain successful plays *always* generate a KRB. I hadn't thought of this being a multi-step process, but we can't rule it out, as best I can tell.Yes. I don't specifically believe this is true (at least didn't think so in the past) but with the Solevision view, it seems more like that there are some additional checks involved to see who is in position to make a key run block and then makes the room for the RB, maybe checks for run defenders who knocfk down their blockers, followed by the RB having the speed and eye to use the hole, before the key run block and tackle get assigned to players. Again, this is probably how I hope it works, more than anything. Given how (special teams) tackles used to be assigned in FOF2004 (kickers tackling a kickoff returner at the 20-yard line), I'm still afraid it's a bit more random, with the better players having a better chance to rack up stats.

[Warhammer]

What if you have a flow chart like this:

Run 23 Smash

Aggregate run blocking and defensive run defense generates modifier for play.

Check run technique among OL for who gets the KRO. In this case, LT gets it.

LT checks for run block technique, if successful then compares blocking strength to RDE run defense. If successful, check on KRB chart. If unsuccessful, check on normal play chart.

If unsuccessful, check RDE run defense to "blow up" play. If successful, check RB elusiveness, if unsuccessful, check play result on normal chart.

Or, maybe what happens is after the KRB is determined, you check the hole recognition for the RB. If successful, he finds the hole, goes through it, then you make a breakaway speed check. If you get this far, a KRB is granted.

[PiemasterUK]

Or, maybe what happens is after the KRB is determined, you check the hole recognition for the RB. If successful, he finds the hole, goes through it, then you make a breakaway speed check. If you get this far, a KRB is granted.

That sounds reasonable to me. Basically for the KRB to happen, the Lineman has to succeed in his task and the Running Back has to suceed in his. In real life, linemen get very little public credit for a great block that the back doesn't take advantage of.

[Vinatieri for Prez]

After reading all of this, the biggest stumbling block I come across is the fact that an O-lineman who springs a RB for a 1-yard gain (for 1st down) gets a KRB, but an O-lineman who springs a RB for a 9-yard gain (on 2nd and long) does not. In real life, you would want the O-lineman who can consistently help his RB break off longer runs (bigger holes) than smaller ones, no? Yet in the FOF universe, with only the stats to look at, you may be going after the wrong guy in FA.

While I don't like it, I can't seem to get around the fact that the KRB has to be assigned post facto based on the "success" of the run. I would much prefer SD's mapping of the formula, but I don't think it stands up to the post-facto requirement.

As an example with SD's map:

4b. If KRB and Pancake both are No, then carrier still has a chance to find a little crease. Higher chance creasefound=Yes in these situations for those with high hole recognition.
5. If creasefound=No, then it's a low gain run.
6. If creasefound=Yes, then start looking at power inside, breakaway running, and elusiveness for potential big gains.

#6 roll for big gain could never happen in FOF. This is because once 4b has determined there is no KRB, and you cannot have a long gain without a KRB. In other words, if you got the big gain in #6, you then have to have a KRB, but it has already been decided there was no KRB in 4b.

Based on the post-facto requirement, MIJB and QS's maps still work. But I do like where QS is going with Rather than "key run blocks," perhaps this stat truly represents something more like "successful runs behind this player as key blocker."

[MrBigglesworth]

After reading all of this, the biggest stumbling block I come across is the fact that an O-lineman who springs a RB for a 1-yard gain (for 1st down) gets a KRB, but an O-lineman who springs a RB for a 9-yard gain (on 2nd and long) does not. In real life, you would want the O-lineman who can consistently help his RB break off longer runs (bigger holes) than smaller ones, no? Yet in the FOF universe, with only the stats to look at, you may be going after the wrong guy in FA.
I don't know if I think this is true, but it's a possibility that things are more complex than we might initially think. For example, what's more important, the lineman that drives his DT into the end zone on third and goal for a TD, or the lineman that blocks the dime back for an 8 yard gain on third and 20? If the stats were recorded the way you are suggesting they be recorded, the long run would be more important, but the short run would be a much more impressive block.

[Vinatieri for Prez]

I don't know if I think this is true, but it's a possibility that things are more complex than we might initially think. For example, what's more important, the lineman that drives his DT into the end zone on third and goal for a TD, or the lineman that blocks the dime back for an 8 yard gain on third and 20? If the stats were recorded the way you are suggesting they be recorded, the long run would be more important, but the short run would be a much more impressive block.

I'd agree with that somewhat, but what about the guy who blocks for a 5-yard gain on 1st and 10 (or an 8-yard gain on 1st and 16 like what I saw) and a guy who blocks for a 6-yard gain on 2nd and 7? In FOF, the first guy gets a KRB, the second guy doesn't. That would seem unfair in real life to say the former guy did a better job of blocking.

As for the shorter runs for 1st downs, the problem is that even a 1-yard run on 4th and 1 against a dime or prevent personnel will get a KRB. You assumed that the 8-yard gain on 3rd and 20 was against a dimeback, but as far as we've seen, the personnel and run expectancy does not factor into the KRB success chart in FOF. Thus in FOF, a 6-yard gain on 2nd and 7 against a 4-3 defense aggressively expecting the run is no KRB (although it would be an impressive job by the O-lineman), while a 1-yard gain on 2nd and 1 against a nickle defense expecting the pass would get a KRB (not all that impressive). As a result, I just don't see it as complex as you've suggested.

[dj_morton]

Good info!

[Celeval]

I've got no empirical data to back it up, but I've always had it as much more simplistic... similar, maybe, to the pass attempt. An offensive blocker is selected to have an opportunity, against one (or two) defenders. If the OL wins his check, KRB kicks in, and the run result comes up on a sliding chart, with the minimum being (5 yards or the first down), with no maximum. If the OL loses his check, the run result comes up on a sliding chart with the maximum of (first down yardage - 1). If the OL loses his check by a lot, the play is severely broken up with some other sliding chart of yardage. The breakway speed, hole rec, etc. all come into play in some sense as well, perhaps adjusting the curve of possible yardage within the restriction.

[Warhammer]

I'd agree with that somewhat, but what about the guy who blocks for a 5-yard gain on 1st and 10 (or an 8-yard gain on 1st and 16 like what I saw) and a guy who blocks for a 6-yard gain on 2nd and 7? In FOF, the first guy gets a KRB, the second guy doesn't. That would seem unfair in real life to say the former guy did a better job of blocking.

But given the situations, would you rather have 2nd and 5 or 3rd and 1? I'd rather have 2nd and 5.

[Ben E Lou]

10 yards isn't the threshold. Here's a 3rd and 22 situation in IHOF today:

Tucker comes out in the I formation with two tight ends, the right side is strong.
The defense comes out in a 34 alignment and 2-deep bump-and-run coverage.
They are expecting the pass.
Shaun Duncan takes the snap from center.
He pitches the ball to R.J. Ellard.
Ellard is running inside the left tackle.
Ellard breaks a pair of tackles and continues.
He rushes for 11 yards.
Tackled by Outside Linebacker Hunter Northcott, and an assist from Defensive End Thomas Barnes.


11 yard gain, but no KRB.

[Vinatieri for Prez]

10 yards isn't the threshold. Here's a 3rd and 22 situation in IHOF today:



11 yard gain, but no KRB.

Somewhere from 12-17 then (17 yards on 2nd and 19 gave me a KRB). We know it's not percentage of distance to go because 6 yards on 2nd and 7 did not get a KRB.

[Vinatieri for Prez]

But given the situations, would you rather have 2nd and 5 or 3rd and 1? I'd rather have 2nd and 5.

Actually, I'd rather have a guy who can get me 6 tough yards on 2nd and 7. And since I have 4 downs, I am fine with 3rd and 1. But the key here is I think it's a close call. They're both valuable, which is why they shouldn't be treated differently.

[Vinatieri for Prez]

I've got no empirical data to back it up, but I've always had it as much more simplistic... similar, maybe, to the pass attempt. An offensive blocker is selected to have an opportunity, against one (or two) defenders. If the OL wins his check, KRB kicks in, and the run result comes up on a sliding chart, with the minimum being (5 yards or the first down), with no maximum. If the OL loses his check, the run result comes up on a sliding chart with the maximum of (first down yardage - 1). If the OL loses his check by a lot, the play is severely broken up with some other sliding chart of yardage. The breakway speed, hole rec, etc. all come into play in some sense as well, perhaps adjusting the curve of possible yardage within the restriction.

This would fit into what we've seen in game. And I like it since it gets rid of the post facto KRB, which I really don't want to be true. In essence the KRB is more goal or sitatuation oriented. It essentially asks the player, can you get me that first down, that TD, or that 2nd and short? The O-lineman either hits the goal or not. Then the RB performance determines what happens in those limited parameters. If this is the case, even a stud RB cannot spring an important run without the help of a KRB from his O-line. Makes sense. As part of this, I would imagine that the probability of making a KRB goes up the easier the goal is.

[Warhammer]

Thinking about this, I really think that there is at some point a global calculation. Let's face it, there are plenty of teams that have absolutely great run defenses, even though they have individual players who aren't that great. Conversely, you have defenses that are mediocre against the run, but have individually great run stoppers. I have seen this in both RL and in FOF, I tend to have good run defenses, but not have any absolute studs in Run D.

Also, in regards to KRO, let's face it, each time there is a run, there are a number of blocks made. It isn't always the lead block that springs the run, sometimes it is a block made on the backside of a play on a player that is coming down the LOS from the backside of the play.

Something else we need to take into consideration, what about playcalling? I am sure playcalling factors into this. We also need to consider aggressiveness and play diagnosis.

[Warhammer]

In regards to play calling, my guess is that there is a general chart for each type of play call. Ag. pass vs. run. Pass vs. run run vs. run Ag. run vs. run, etc. This gives you the chart(s) to use.

The defense makes a reaction based upon the play diagnosis of the players, this can affect whether or not the incorrect play call is mitigated at all. Basically you have 11 guys, if the play is called correctly, the play diagnosis of the players gets a big bonus. Players make the check, and those that pass add their run defense to the play, etc.

Then you go through SkyDog's routine. The results of which can either lead to a DRM (die roll modifier) for the charts or the use of another chart.

[MrBigglesworth]

I've got no empirical data to back it up, but I've always had it as much more simplistic... similar, maybe, to the pass attempt. An offensive blocker is selected to have an opportunity, against one (or two) defenders. If the OL wins his check, KRB kicks in, and the run result comes up on a sliding chart, with the minimum being (5 yards or the first down), with no maximum. If the OL loses his check, the run result comes up on a sliding chart with the maximum of (first down yardage - 1). If the OL loses his check by a lot, the play is severely broken up with some other sliding chart of yardage. The breakway speed, hole rec, etc. all come into play in some sense as well, perhaps adjusting the curve of possible yardage within the restriction.
I also have no data, but my feeling from watching hundreds of FOF games is that when teams run the ball on third and 1, they get the first down a lot more than the average rate of key run blocks (which I think is like 25-30%), which doesn't fit with this theory.

[QuikSand]

For those struggling with this whole notion of running "success" here's the FO link to same:

http://www.footballoutsiders.com/ramblings.php?p=254&cat=11

A snippet:

* In general, a play counts as a "hit" if it gains 40% of yards on first down, 60% of yards on second down, and 100% of yards on third down.
* If the team is behind by more than a touchdown in the fourth quarter, the benchmarks switch to 50%/65%/100%.
* If the team is ahead by any amount in the fourth quarter, the benchmarks switch to 30%/50%/100%.

[QuikSand]

I also have no data, but my feeling from watching hundreds of FOF games is that when teams run the ball on third and 1, they get the first down a lot more than the average rate of key run blocks (which I think is like 25-30%), which doesn't fit with this theory.

Completely agreed there.

[primelord]

To me, this is the most frustrating piece of it. As it is, I'm starting to wonder how much you can use an individual player's KRB stats as a measure of his own performance... all indications point to this being very much a team stat, with individual credit apportioned in some rough way back to the players, but very much a function of a lot more than that one guy. An excellent back, or excellent OL colleagues, would certainly seem to boost any OL player's KRB numbers, it seems to me, and the converse is true as well.

I haven't made it all the way through this, but I kinf od think this should be a team stat. For example in real life football, on a run off the tackle a good RB is often going to be partially responsible for a "KRB" that gets credited to the T by knowing when to cut back inside of the tackle or bounce outside etc. In addition a good guard keeps the DT from pentrating and stopping the play in the backfield before the T even gets a shot at a KRB.

I know FO uses stats like ALY to try and seperate the RB from the line, but in general I usually view most run stats as the RB and his line produced these stats.

[Vinatieri for Prez]

I also have no data, but my feeling from watching hundreds of FOF games is that when teams run the ball on third and 1, they get the first down a lot more than the average rate of key run blocks (which I think is like 25-30%), which doesn't fit with this theory.

Although there is likely a higher success rate of KRBs for 3rd and 1, Celeval's model may still work if you look at it in terms of the KRB probability changing with each down and distance situation (say, 4-5 general types). That is, it is easier under the formula for the O-lineman to get a KRB in 3rd and 1 instead of 1st and 10.

[MrBigglesworth]

Although there is likely a higher success rate of KRBs for 3rd and 1, Celeval's model may still work if you look at it in terms of the KRB probability changing with each down and distance situation (say, 4-5 general types). That is, it is easier under the formula for the O-lineman to get a KRB in 3rd and 1 instead of 1st and 10.
I think that may work. Ok, this thread is kind of long and confusing, but what I think we have are the 'Pre Facto model' (where the KRB check is done first, and then the result of the play is determined) and the 'Post Facto Model' (where the result of the play is determined, and then the game looks to see if the result matches the conditions necessary for a KRB). How can we either disprove one or the other with info we have, or how can we design an experiment where one model will predict the correct results and the other will not? If we can't, ie if our observation of the game wouldn't change no matter what the correct theory was, then it really doesn't matter which one is correct.

For the team vs. individual theories, someone can run a test where one team has one good run blocker and four other crappy linemen, and another test where the team has five good run blockers. The skill of the RB could also be another variable, giving four separate trials.

[Vinatieri for Prez]

I agree that the 2nd part could be tested easily. Unfortunately, I don't see an easy way to disprove/prove the post facto or prefacto model.

[Vinatieri for Prez]

I'll interject another thought with the RBs 3rd down rushing ability.

The help file says that the skill measures "their third down running."
The csv file says that the skill is the "ability to gain yardage in third- and fourth-down and short situations."

If a KRB results in a 3rd down and short conversion everytime, what does that say about this skill. It would seem if the pre facto theory applied, this skill is not very useful at all. Again, bad news as far as I am concerned.

And what about the elusiveness rating?

The help file says it means the "ability to occasionally break long runs at the expense of sometimes getting caught behind the line of scrimmage."
The csv file says that it means "the ability to avoid getting caught behind the line of scrimmage."

Again, it would seem this rating alone could be the difference between a long run (requiring a KRB) and getting caught behind the line of scrimmage (not a KRB). So, this also would seem to cut against the pre facto theory.

Any thoughts about this?

[Vinatieri for Prez]

Ok, re-thinking the RB ability thoughts. The only way I see the pre facto theory apply is that the game must do a check on elusiveness and/or 3rd down rushing ability before the blocking check with the O-line. That would seem pretty counterintuitive and would require involving defender ratings before the blocking occurs. I'm not so sure the game would do it that way.

I am more and more leaning towards QS's theory that a KRB simply represents something more like "successful runs behind this player as key blocker."

[MIJB#19]

But I do like where QS is going with Rather than "key run blocks," perhaps this stat truly represents something more like "successful runs behind this player as key blocker."
I'm starting to think this is exactly what's going on.
Go into the game and look at OL stats, the default view is key run block opportunities, not key run block, block%, sacks allowed, sacked% or OL%.

At least that would fit in with my "potential key run blocks" idea. Just that the potential key run block is in fact the "key run block opportunity".

[OldGiants]

My take on this has always been a bit different from what I’ve read here, although not by much. It is more in line with Skydog’s postulates then Quicksand’s, but not far from either. Like Skydog, my guess (for that is all any of this is) comes from a knowledge of board games like Strat and APBA and how they derive results.
<!--[if !supportEmptyParas]--> <!--[endif]--><o></o>
I agree with everyone that an overall calculation of the strength of the offense and defense is made (think of this as the Offensive Index used in APBA football, if that helps).
<o></o>>
Next the play, position on the field (again, like APBA) and defensive formation and call determine a base column to roll on, and a preliminary result is arrived at. (Note: here is where the number of formations your QB knows helps or hinders you).
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>

The base result is not one number, but two. For example, +4 or +1. As I’ll show below, if the KRBO is met, the best number is the result, if missed, the lower is the result.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Now the skills of the RB come into play, as do things like ‘play diagnosis’ on the defense. Each of these adjusts either the preliminary result itself (say from the set +4,+1 to +3,0) or the chances of make the better result. Here’s how I think it works, tying what we know of the ratings to this system.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Breakaway only comes into play on those occasions where a long run is possible. Plodders don’t get long runs except by exceptional chance. Breakaway is likely a cap on the result, not an increase in chance to ‘go all the way.’ To quote LeRoy Hoard “If you need three yards, I’ll get you three yards. If you need six yards, I’ll get you three yards.” He had a zero breakaway rating.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Inside running or Outside running adjust the possible results up or down based on where the play goes. Third down also works this way when applicable.
<!--[if !supportEmptyParas]--> <!--[endif]--><o></o>
Elusiveness only impacts those runs where a loss (or losses in some extreme cases) could be the result. It decreases the loss, say from –3 to –2, but never to better than ‘no gain.’ I think this is consistent with the FOF Help files on this point.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Now a blocker and a defender are selected. The run ratings of both are compared (although sometimes this changes to a Strength comparison) and the probability of rolling the best result is calculated.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Now comes the final adjustment before the die roll. The probability of rolling the better result is adjusted upward by the RB’s ‘find the hole’ skill. That’s why this is such a valuable skill.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Now the result comes. Most of the time, the block is missed and lower run is the result. This is a significant part of the run result routine because a blocker who ‘misses’ the block gets a KRB if the team makes a yard and a FD or TD.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Anyway, the exact order can be swapped about, but I think this is essentially how the running game works.

[Celeval]

I'm starting to think this is exactly what's going on.
Go into the game and look at OL stats, the default view is key run block opportunities, not key run block, block%, sacks allowed, sacked% or OL%.

At least that would fit in with my "potential key run blocks" idea. Just that the potential key run block is in fact the "key run block opportunity".

This should be relatively provable.

Same offensive lines, multiple trials with a stud RB and a nobody RB. If KRB is a result stat and not a cause stat, the percentage of successful KRBs should be much higher with a stud RB than with a nobody RB.

[Warhammer]

My take on this has always been a bit different from what I’ve read here, although not by much. It is more in line with Skydog’s postulates then Quicksand’s, but not far from either. Like Skydog, my guess (for that is all any of this is) comes from a knowledge of board games like Strat and APBA and how they derive results.
<!--[if !supportEmptyParas]--> <!--[endif]--><o></o>
I agree with everyone that an overall calculation of the strength of the offense and defense is made (think of this as the Offensive Index used in APBA football, if that helps).
<o></o>>
Next the play, position on the field (again, like APBA) and defensive formation and call determine a base column to roll on, and a preliminary result is arrived at. (Note: here is where the number of formations your QB knows helps or hinders you).
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>

The base result is not one number, but two. For example, +4 or +1. As I’ll show below, if the KRBO is met, the best number is the result, if missed, the lower is the result.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Now the skills of the RB come into play, as do things like ‘play diagnosis’ on the defense. Each of these adjusts either the preliminary result itself (say from the set +4,+1 to +3,0) or the chances of make the better result. Here’s how I think it works, tying what we know of the ratings to this system.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Breakaway only comes into play on those occasions where a long run is possible. Plodders don’t get long runs except by exceptional chance. Breakaway is likely a cap on the result, not an increase in chance to ‘go all the way.’ To quote LeRoy Hoard “If you need three yards, I’ll get you three yards. If you need six yards, I’ll get you three yards.” He had a zero breakaway rating.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Inside running or Outside running adjust the possible results up or down based on where the play goes. Third down also works this way when applicable.
<!--[if !supportEmptyParas]--> <!--[endif]--><o></o>
Elusiveness only impacts those runs where a loss (or losses in some extreme cases) could be the result. It decreases the loss, say from –3 to –2, but never to better than ‘no gain.’ I think this is consistent with the FOF Help files on this point.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Now a blocker and a defender are selected. The run ratings of both are compared (although sometimes this changes to a Strength comparison) and the probability of rolling the best result is calculated.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Now comes the final adjustment before the die roll. The probability of rolling the better result is adjusted upward by the RB’s ‘find the hole’ skill. That’s why this is such a valuable skill.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Now the result comes. Most of the time, the block is missed and lower run is the result. This is a significant part of the run result routine because a blocker who ‘misses’ the block gets a KRB if the team makes a yard and a FD or TD.
<!--[if !supportEmptyParas]--> <!--[endif]--><o>:p></o>:p>
Anyway, the exact order can be swapped about, but I think this is essentially how the running game works.

Agreed. Based upon the ratings in game, and what we know, I think this makes the most sense.

[jzicc]

So everyone is agreement that elusiveness is merely the ability to not have a loss ? with no factor in every single tackle? -I would think it would really be the last modifier prior to the dice roll ---- thoughts?

[cuervo72]

Interesting discussion.

After looking at OldGiants' post, you all should hope Jim does a board game, just to gain insight into how Jim resolves outcomes.

Right now, I'm a little bit in the camp of treating the line as a whole, then assigning the glory/blame after the fact (in which case, any weak link along the line could really hurt...) I don't have any evidence here to offer, and I don't particularly like the idea that it's assigned that way, but that's where I'm leaning.

As for guys who don't seem like they'd be in the area getting tallies, it could be from traps, etc. Maybe test with a plan where finesse running is eliminated?

The help and csv definitions of elusiveness seem to contradict each other too, IMO. But maybe I'm reading it wrongly.

[QuikSand]

I think there's an Occam's razor component here... absent testing that tells us to look elsewhere, I'm basically looking for the simplest explanation for the things that we see. The fact that the KRB allocations seem to match up, at least pretty well, with the same conceptual notion of "rushing success" that has been used by Football Outsiders (an organization with whom Solecismic obviously has a contractual arrangement) make me think that this is, essentially, what the stat now reflects.

The more I think about it, the more comfortable I am thinking that the stat is really best interpreted as mentioned earlier: "successful runs behind this player as the key blocker." And whether that means that it's dependent in part on the skill of the ball carrier (for certain) or perhaps all the supporting cast of blockers (speculated here, but pretty sensibly so), it's pretty clear that it is not a solo result. In my view, that is one of the essential things to try to understand on this front.


I think that the notion of doing some extended testing and comparing KRB stats for the same team employing an excellent RB and a lousy one would certainly help. I'm not sure what else could easily be done... maybe sub out most of the quality OL starters with scrub reserves, and see how the remaining one or two guys do over the long haul? Both of those, I suspect, would yield evidence that KRB performance is a function of a supporting cast, not just the individual player. (Which flies in the face of what I had always expected this stat to be, but is much mroe in keeping with the results tree we're starting to reveal)

[QuikSand]

My take on this has always been a bit different from what I’ve read here, although not by much. It is more in line with Skydog’s postulates then Quicksand’s, but not far from either. Like Skydog, my guess (for that is all any of this is) comes from a knowledge of board games like Strat and APBA and how they derive results.

I think your "guess" here is similar to what SkyDog posted earlier for a simple reason -- SkyDog (by his own account) wasn't offering his thoughts on how FOF actually works, he was offering his thoughts on how he would prefer it to work. My sense is that you're essentially doing the same thing here -- offering conjectuire based on other games or systems, but not really based on FOF, perhaps with some rounding-off of the edges to account for the assignment factors that have already been discussed.

In one sense, I think I agree with you. Specifics aside, I think I would prefer the sort of models that SkyDog adn you are offering up. But I don't think there's much evidence that FOF is making decisions in that sort of order, even if it makes intuitive sense to us that it ought to do so. I don't think intuition, or even the plain meaning of the words used in the FOF game, are our best guides here.

[Ben E Lou]

SkyDog (by his own account) wasn't offering his thoughts on how FOF actually works, he was offering his thoughts on how he would prefer it to work.Correct. I don't see much evidence that it does work that way. I just wish that it generally *did* work that way.
I don't think intuition, or even the plain meaning of the words used in the FOF game, are our best guides here.Unfortunately, I think you're correct again here. http://www.fof-gefl.com/ihof/phpBB2/images/smiles/icon_sigh.gif (javascript:emoticon(':sigh:'))

[QuikSand]

Okay then... if the concept of KRBs is indeed something like my statement above (successes on runs behind this player) then what the heck does the RB's rating for "hole recognition" mean in that world?

I used to think that it was a sequential process -- KRB is checked, and if that succeeded, then the back's HR was checked to see how well he takes advanatge of it. If KRB is really an after-the-fact stat, then how does Hole Rec fit in at all? *shurg*

[MIJB#19]

Okay then... if the concept of KRBs is indeed something like my statement above (successes on runs behind this player) then what the heck does the RB's rating for "hole recognition" mean in that world?

I used to think that it was a sequential process -- KRB is checked, and if that succeeded, then the back's HR was checked to see how well he takes advanatge of it. If KRB is really an after-the-fact stat, then how does Hole Rec fit in at all? *shurg*
By my anecdotal experience, I think hole recognition helps OL guys to get their key run blocks. When I see green rookies and 0/0 run blockers post 40-45% key run blocking stats, it can't be that the run blocking skill determines whether the key run block opportunity becomes a succesful key run block.

[OldGiants]

Okay then... if the concept of KRBs is indeed something like my statement above (successes on runs behind this player) then what the heck does the RB's rating for "hole recognition" mean in that world?

I used to think that it was a sequential process -- KRB is checked, and if that succeeded, then the back's HR was checked to see how well he takes advanatge of it. If KRB is really an after-the-fact stat, then how does Hole Rec fit in at all? *shurg*

I don't know why you're going down the 'after the fact' route if you are a professed Occam's Razor sort of guy. If after the fact doesn't explain HR, then shouldn't you toss out after the fact as a possibility? Clearly KRB isn't an after the fact assignment but HAS to work much like the model I proposed. Wouldn't that be the Occam's Razor conclusion?

First you find out if the KRB was made, that ratches up your chances and then the RB's Hole Recognition helps determine if he takes advantage of the KRB to its fullest extent. If the KRBO is missed, no need for the HR rating of the RB for there is no 'bonus' on this play.

As Cuervo72 points out, this is why I want to see Jim's boardgame.

[QuikSand]

Clearly KRB isn't an after the fact assignment but HAS to work much like the model I proposed. Wouldn't that be the Occam's Razor conclusion?

How then would you square what we see on short-yardage situations? It's perfectly clear that if the play nets the first down, a KRB is assigned. If the play fails to net the first down, the KRO fails.

But teams succeed in these situations FAR more often than KRBs succeed generally. That completely flies in the face of the notion that the KRB is an input into the play determination. I guess the only way that would make any sense would be to add a good deal of situational complexity into the calculations of when a KRO is successful (higher chance in situations where a rushing "success" is likely, and so forth).

So, I'm back to the simplest explanation. It would be really simple, I think, to program this as a "run success ---> key block gets awarded" system. Having a deeply complex structure of calculations for all manner of circumstances that wouuld guide the likelihood of the KBR g=being awarded, and only then generate a result of the play where there's a perfect IFF correlation to the outcome just seems way too complicated here, to explain what we see.

That's why I'm invoking Occam. I htink your model is by far the more complicated of the two. (Again, not that I dislike what you suggest as a way to hypothetically make a game work... but that I just don't suspect FOF works that way)

[QuikSand]

By my anecdotal experience, I think hole recognition helps OL guys to get their key run blocks.

Which, under my current thinking, seems to generally mean that "hole recognition --> better rushing otucomes" (since it's good rushing outcomes that themselves determine the assignment of key run blocks). That's bordering on tautology, I fear. I assume that the rating is used somehow, and that a higher rating is a good thing to have, all else equal. What I'm searching for is a fair way to articulate what it means as a calculation in the game.

I used to think that it was a check after a KRB is produced. Now, that model makes no sense. So I'm basically at a loss, unless we would substitute in some other generic measure of "the blocking opens up a hole for the runner" that might get calculated in the game, even if it isn't what is being shown as a KRB. (If that makes any sense)

[Vinatieri for Prez]

Something to add with respect to the KRBs and Pancake Blocks. I have no idea what it means.

In the box score, my LG is credited with 1 KRB and 1 Pancake. However, in the game log, the commentary only describes a single KRB and no pancake (i.e. ran over * to open the hole). However, in Solevision, there is one play on 2nd and 22 where the run went for 8 yards. Solevision said the LG ran over * to open the hole. On that same play in the game log, there is no mention of a KRB or him running over someone to open up the hole. But earlier in the game on a 3rd and 4 that went for 12 yards, my LT was credited with a KRB and pancake in the game log and Solevision.

So, apparently it is possible to get a pancake block even though it is not a KRB. On the 3rd and 22, we got 8 yards. Normally, this is not good enough to get a KRB, and my LG didn't get one. But he did get a pancake block for it. The weird thing is that although it showed up as a pancake in the box score and Solevision it did not show up in the game log. We know pancakes can show up in the game log because we've seen them (like for my LT above), but apparently the game log did not show the pancake when it was not accompanied by a KRB -- I'm not sure if this is an every time kind of thing or just a glitch in this one game.

Either way, because you can get a pancake block without a KRB, does it shed any light on pre or post facto theories or how the blocking formula may work? I don't believe I have ever noticed a pancake not be a KRB before. How about you guys? I'm too tired to analyze this in great detail right now, but just wanted to make the observation now while it is fresh. I'll mull it over some more tomorrow.

For now, while I have been a big fan of the post facto KRB theory, it would seem at some point there has to be a pre facto strength test between the blocker and the defender. Now, if most pancakes are KRBs, this would cut against the post facto theory, no?

[Vinatieri for Prez]

Again, my limited review of 3 game logs/solevision show this to be true.

I also point out that when you get the message in Solevision like "he knocked him flat" or "threw him to the ground," the lineman gets credited also with a pancake block. (i.e. the lineman gets credited with a KRB and a pancake, not just the pancake). Or in other words, all pancake blocks are KRBs.

Again, this is with limited anecdotal review, no testing.

Following up from the above post, it appears that my speculation is incorrect -- that while most pancake blocks are KRBs, not all pancake blocks are KRBs.

[Sgran]

I just have a few small things to add. First, elusiveness is for open field, one-on-one situations, wherever they occur. I belive this because sometimes in the game you get something along the lines of "James is through the line. He fakes out Palimino and keeps going."
Although i might not be adding much to the discussion, i always think of running plays in terms of waves. First there is the point of attack, i.e. the hole, where there can be a) a stalemate and the runner gets a few yards (perhaps calling on the "running inside" factor); b) the defensive player "blows up" the play (maybe elusiveness enters in here and Barry Sanders gets 25 yards); c) there is a "key run block" that opens up a hole for the runner. At this point, the postion of the other defensive players comes into play and their speed is factored against the "breakaway" and "speed" of the runner. My guess is that the blockers disappear from the equation at this point, but there might be a small allowance for WRs that block well. Elusiveness is now called into play in the open field and a back high in this quality can juke a poor tackling safety.
As for key run blocks by lineman away from the hole, it could be that they're pulling (did Jim ever say what the first number of RB 24 Smash means? In my day that meant the "2" lineman was pulling). It could also be that the hole wasn't there and the back cut back to a hole opened up by a lineman on the other side.
I don't think that KRBs are assigned retroactively. If the block gets the back into the secondary, that's a KRB. If it's a short-yardage situation, the defense is keying the run, and therefore any success on the line is key.

[Warhammer]

1 - QB
2 - HB
3 - FB

So 24 Smash is the HB going through the 4 gap.
34 Smash would be the same thing with the FB getting the carry.

[Warhammer]

Based on how things work, could it be possible for the check to take place, the RB fails his HR check, and thus the KRB is not awarded because the RB failed to take advantage of the hole?

[Sgran]

Based on how things work, could it be possible for the check to take place, the RB fails his HR check, and thus the KRB is not awarded because the RB failed to take advantage of the hole?

If you think about it, that's the way it works in real life. Seattle and San Diego have acclaimed lineman because they have star half-backs hitting the holes and having 1500 yard seasons. Football is a team game, so all of these things are interdependent. What i like about FOF as opposed to Madden is that players don't improve simply because they have great stats. If you think about Madden, it's a rich-get-richer system where players with great ratings have great seasons, and their ratings improve, so they have great seasons...
One thing Jim did say was that he wanted us to look at stats, and not just at ratings, for what that's worth. In one of my MP leagues we had a college import draft and you could see the player's stats. I weighed them heavily in my selection process.

[Vinatieri for Prez]

Based on how things work, could it be possible for the check to take place, the RB fails his HR check, and thus the KRB is not awarded because the RB failed to take advantage of the hole?

Hmm, I'm warming up to this one. This would explain away some of the stumbling blocks that had me in favor of the post facto theory. In other words, when the blocking check is done to start with, the lineman may get a Run Block, but it does not become a Key Run Block unless the running back does something good with it.

This would also explain my pancake block discussion above. There, my LG got a Run Block on the blocking check (and such a good one that it was recorded as a pancake), but because my RB could only spring it for 8 yards on 2nd and 22, my LG does not credit for a Key Run Block.

[Vinatieri for Prez]

I think the same order of things happens as you posted QS, yet I'll provide a bit of an alternate theory:

-an aggregate "run blocking" rating for your offense and the defense is determined, perhaps with added weight to the players in the immediate zone where the run is to take place

-dice are rolled to determine the net effect of the blkocking scheme

-the ball carrier's skills are checked to determine how well he takes advantage of the setup

-the game assigns a key run block opportunity to somebody, presumably weighted toward the immediate zone where the run took place (but not exclusively so)

-the game assigns a key run block defender, presumably weighted toward the immediate zone where the run took place (but not exclusively so)

-the game assigns a potential key run block (and pancake block?) if the run blocker beats the run defender

-the play nets a certain result

-the game assigns a tackle to somebody on the defense

-if the run met a certain "success" threshhold, then the KRO for that player is considered to have succeeded, and the KRB is awarded and tracked



I now believe that MIJB's formula is probably how it is done, including how the pancake fits into it.

[QuikSand]

This would also explain my pancake block discussion above. There, my LG got a Run Block on the blocking check (and such a good one that it was recorded as a pancake), but because my RB could only spring it for 8 yards on 2nd and 22, my LG does not credit for a Key Run Block.

I don't see how that example undermines my theory from above at all.

Every play gets one key blocker, who is credited with a KRO. Iff run = "success," award KRB to the selected key blocker.

I still think that's the simplest and most elegeant way to understand this, based on every single example we're seeing.

[QuikSand]

I now believe that MIJB's formula is probably how it is done, including how the pancake fits into it.

Agreed, except that there appears to be exactly one Key Run Block Opportunity for every run play, other than QB scrambles.

[Vinatieri for Prez]

Yes, QS, I think we're saying the same thing. That absolutely the KRB is awarded post facto depending on success of the run. That means that a bad RB will reduce a great lineman's KRBs and vice versa. Moreover, cohesion and other teammate lineman's skills play into it as well, as you previously suggested. And further that a KRB simply means successful team runs behind a certain player.

I just think with my new example a little nuance has been added in the form of "potential KRBs." I am now certain that at a minimum, a pancake is a "potential KRB," and that there are many other "potential KRBs" that we will never see in the stats because the run was not a success and there was no pancake. It also warms my hear that it must mean there is at least some sort of pre-result blocking check with the individual lineman. Again, I think we knew that already (at least in terms of hidden "potential KRBs" involving the whole team concept formula), but now I believe at least that there is an individual check going on pre-run result that is wholly independent of, and whose success is independent from, how the final run result turns out.

I think this means that if you look at a linemans' stats, the pancacke block is probably the one true independent stat that can shed light on his purely individual skill. While this may have seemed obvious to begin with, I think it becomes more important now that we know (or at least my opinion is) the KRB is more of a team stat.

And as for KRO, I use to think these were linked to lineman's individual skills, but since there is exactly one for each designed run, I don't think this is the case. It may be slightly related to it, but I think it's just more a function of run direction (not a complete match but weighted that way). Of course, since most people run behind their best blockers (or at least perceived as such by their ratings), you should see more KRO for those guys.

[QuikSand]

Fair enough... I guess I got a little lost with the notion of "potential KRB" as something separate from the state we actually see, the "key run block opportunity" or KRO. I'm not at all persuaded that there has to be any such stat or consideration buried into all this logic to make sense of the ouctomes we clearly are seeing.

I just think this is a case of the stat itself meaning something very different than what a reaosnable person would expect it to mean.

[Sgran]

has anyone tracked KROs?

[QuikSand]

Key Run Block Testing

General Hypothesis: The Key Run Block stat is essentially a measure of “rushing play effectiveness” and not of something that happens to set up a given play prior to determining its outcome. Every designed rushing play causes a key run blocker to be selected for a KRO, and if the play meets a certain threshold for success, that blocker is awarded a KRB, and gets a positive notation in the Solevision report, the extended game log, and will have that stat recorded.

Testing Plan: If KRBs are indeed an individual measure, then they should be independent of the running back and his skills. I will run a given team, in a controlled environment, several times with a highly skilled RB, and then several times with a far lesser-skilled RB, and compare the KRB performance of the teams. If the hypothesis above is correct, we should expect the KRB performance of the same offensive linemen and other blockers be better with a superior running back.

Specifics: I am using a test team, created by a preference draft with emphasis on offensive skill players, and playing the 2006 season, setting injuries to zero, and using a staff-recommended gameplan selected at the start of the season. My nominal starting RB is Tiki Barber, rated 74/74 overall, and strong on most areas (except elusiveness). My alternate starter will be rookie free agent Monty Blake, who is rated 16/26 overall, has good enough endurance to take most of the starting snaps, a weakness in elusiveness, and generally pretty shaky skills. My two backup RBs have similar splits of skills – we’ll use a decent backup behind Barber, and a hobo behind Blake.

- - -

Standard Roster Data

Washington Chitterlings Team Rank
Rushes per Game 37.4 1
Rushing Yards 159.3 1
Yards Per Carry 4.25 4

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin193 572 33.7 1001 57.1 13 81 980 1.3 54

Washington Chitterlings Team Rank
Rushes per Game 38.1 1
Rushing Yards 152.4 1
Yards Per Carry 4.00 11

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin208 579 35.9 955 60.6 20 132 1013 2.0 51


Washington Chitterlings Team Rank
Rushes per Game 37.6 1
Rushing Yards 146.4 1
Yards Per Carry 3.90 17

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin194 570 34.0 969 58.8 13 77 987 1.3 51

Washington Chitterlings Team Rank
Rushes per Game 36.0 2
Rushing Yards 148.5 2
Yards Per Carry 4.13 15

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin183 531 34.5 990 53.6 22 140 1029 2.1 45

Washington Chitterlings Team Rank
Rushes per Game 37.9 1
Rushing Yards 146.9 1
Yards Per Carry 3.88 21

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin191 574 33.3 951 60.4 30 203 1105 2.7 42

So – with five trials, this actually looks pretty stable. The overall results vary a bit, no susprise, but the blocking data, as a team, seems pretty consistent:


Yards Per Carry 4.20 4.00 3.90 4.13 3.88 Avg = 4.02
Key Run Block % 33.7 35.9 34.0 34.5 33.3 Avg = 34.1


Aside - I find it somewhat interesting to see that the KRB% doesn’t seem to correlate very well with overall YPC – but I’m not confident that five seasons is enough to draw any meaningful conclusion there. With more data reinforcing this weak or even opposite correlation, I think this could be argued to further support the hypothesis.


Alternative (deeply weakened RB position) Roster Data

Washington Chitterlings Team Rank
Rushes per Game 39.5 1
Rushing Yards 132.6 4
Yards Per Carry 3.36 31

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin166 605 27.4 972 62.2 17 102 1030 1.7 40

Washington Chitterlings Team Rank
Rushes per Game 38.3 1
Rushing Yards 136.0 3
Yards Per Carry 3.56 26

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin142 575 24.7 985 58.4 19 130 986 1.9 37

Washington Chitterlings Team Rank
Rushes per Game 37.1 1
Rushing Yards 130.8 4
Yards Per Carry 3.53 31

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin153 546 28.0 983 55.5 21 150 1007 2.1 51

Washington Chitterlings Team Rank
Rushes per Game 36.1 2
Rushing Yards 121.9 8
Yards Per Carry 3.38 32

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
Denver M180 553 32.5 981 56.4 19 119 1009 1.9 47
**Washin146 550 26.5 953 57.7 14 92 1084 1.3 44

Washington Chitterlings Team Rank
Rushes per Game 39.2 1
Rushing Yards 147.1 2
Yards Per Carry 3.75 24

Team KRB KRBO BPct RPly OPct SckA Yards PPly SPct OpHur OpBlk OpPRPct
**Washin169 601 28.1 978 61.5 16 101 983 1.6 35


Yards Per Carry 3.36 3.56 3.53 3.38 3.75 Avg = 3.52
Key Run Block % 27.4 24.7 28.0 26.5 28.1 Avg = 26.9


Well… there you have it. If Key Run Blocks were indeed purely a function of the blocker who gets the opportunity and credit, then we would have expected to see KRB% stay more or less constant here, but the team just wounl’t be running as effectively despite the good blocking.

Instead, the initial hypothesis predicts that your Key Run Block results will get worse if you have lousy running backs – the kind of guys who can’t take advantage of good line play. And that is quite obviously what happened here. Five trials isn’t a huge sample size, but to be candid – it’s awfully easy to tease out this trend, as the OL performances are just miles apart.


Obvious Conclusion

Key Run Blocks are a measure of how often the team “succeeds” with running plays. For individual players, the Key Run Block Opportunity (KRO) stat measures the number of times he was selected as the key blocker, and the Key Run Block (KRB) stat measures the number of times the run play met a certain defined success threshold. This is clearly a function of more than how well the lineman does his job, or even how well the entire blocking unit works.

[KWhit]

Great work QS (and everyone else). Thanks for sharing the results. 34.1% versus 26.9% couldn't be much more clear. I am quite surprised that the discrepancy is so large.

[Sgran]

This is consistent with other stats, such as sacks allowed. A good QB will "sense rush" and get rid of the ball. No sack allowed.

[SFL Cat]

Did a fast and dirty test using the Detroit Lions

RB is rated 66/66

Set run direction to 100% around left end and 0 in all other fields. In all formations screens, I set finesse to 0 (hopefully reducing or eliminating pulling guards, centers and whatnot).

Results:

My team rushed 394 times for 820 yards and a 2.08 average.


Attempts AVG
Left 367 2.13
Middle 16 3.56
Right 0 0.00


A quick perusal of game logs and box scores indicate a majority of the middle runs can be attributed to QB scrambles

Here are the KRB stats for my starters (ratings and run ratings included).

G/GS OVERALL RATING RUN BLOCK KRO KRB %

LT 16/16 66/66 97/97 100 19 19.0
LG 14/14 65/65 44/44 84 15 17.9
C 16/16 78/78 80/80 71 8 11.3
RG 16/16 45/52 52/65 37 12 32.4
RT* 16/16 31/53 28/63 19 5 26.3
TE 16/16 63/63 55/55 16 3 18.8


*combined two players who started in the RT spot and averaged their ratings.

Based on this there does seem to be at least some correlation between where you run and who gets selected for a KRO.

When I have time I go more into the game logs and see if the KRBs make sense in a football context (i.e. I can see a center nicking a lineman and then releasing to go find a linebacker or safety downfield, so if he's getting credit for a KRB on a 2 or 3 yard gain on a left end around, something is screwy there.)

[QuikSand]

This is consistent with other stats, such as sacks allowed. A good QB will "sense rush" and get rid of the ball. No sack allowed.

I susppose this isn't worth arguing over -- but two things...

(1) Sacks allowed is an unambiguous event. The QB gets sacked, and someone gets assigned the blame. Any intuitive understanding of that explanation would tell us immediately that a very elusive QB would escape pressure more than a flatfoot, and therefore reduce the number of sacks the team would yield -- regardless of whether the blockers are doing a good job.

(2) I'm also perfectly open to an explanation that "sacks allowed" works on a teamwide basis first, with assignment only after the fact, just like KRBs clearly do. That is -- if you are under the impression that your RT being picked as the guy who "should have prevented the sack" means that he lost out on some sort of an individual dice roll against the pass rusher he directly faced on that play... maybe it doesn't work like that at all. Maybe it's a team-wide dice roll, based on everyone's skills in some weighted average, and if it triggers a pressure resulting in a sack, only then is individual credit for the sack and blame for the sack allowed back to some guy drawn off the list of contenders by a dice roll of his own. Heck, maybe there's even a "key pass blocker" selected for every pass play, just like the KRO, and we just don't see that stat anywhere. *shurg*

[OldGiants]

How then would you square what we see on short-yardage situations? It's perfectly clear that if the play nets the first down, a KRB is assigned. If the play fails to net the first down, the KRO fails.

But teams succeed in these situations FAR more often than KRBs succeed generally. That completely flies in the face of the notion that the KRB is an input into the play determination. I guess the only way that would make any sense would be to add a good deal of situational complexity into the calculations of when a KRO is successful (higher chance in situations where a rushing "success" is likely, and so forth).

So, I'm back to the simplest explanation. It would be really simple, I think, to program this as a "run success ---> key block gets awarded" system. Having a deeply complex structure of calculations for all manner of circumstances that wouuld guide the likelihood of the KBR g=being awarded, and only then generate a result of the play where there's a perfect IFF correlation to the outcome just seems way too complicated here, to explain what we see.

That's why I'm invoking Occam. I htink your model is by far the more complicated of the two. (Again, not that I dislike what you suggest as a way to hypothetically make a game work... but that I just don't suspect FOF works that way)

My explanation for KRBs in short yardage covered this point. If the KRO is missed, but the team gets a yard anyway and makes the first down, a KRB is awarded. An example would be KRB made= +4 KRB missed = +1. When this happens on third and one, the first down is made and a KRB is awarded even though the KRO was missed.

I like your test results. I think they bear out what I've outlined. Running is an overall team result and bad runners give you worse results. The tricky bit is to come up with the cheapest/lowest rated RB who will still get the job done.

Again, I don't pretend to have the exact order of the calculations or the formulas themselves, but the general idea is how the system must work.

[QuikSand]

My explanation for KRBs in short yardage covered this point. If the KRO is missed, but the team gets a yard anyway and makes the first down, a KRB is awarded. An example would be KRB made= +4 KRB missed = +1. When this happens on third and one, the first down is made and a KRB is awarded even though the KRO was missed.

I'm still looking for the simplest answer here. Your idea that the game reaches back and awards or takes away KRB credit based on the final result of the play makes the initial calculation totally irrelevant. So, why do you insist it's still being calculated as an input into the play result? That's what makes no sense to me.

I like your test results. I think they bear out what I've outlined. Running is an overall team result and bad runners give you worse results. The tricky bit is to come up with the cheapest/lowest rated RB who will still get the job done.

Again, I don't pretend to have the exact order of the calculations or the formulas themselves, but the general idea is how the system must work.

Of course a team's overall rushing effectiveness is a function of all the players involved, including the RB as well as the blockers. I can't see how that was ever in question -- but if it was in question, then by all means award yourself full credit, you're dead on there.

I think the meaningful point of the study is that the skill of the RB makes a huge difference in the stats that would appear to be all about the performance of the individual lineman/blocker. If you still believe that the KRB% for a given lineman is a measure of that one guy's effectiveness in a vacuum, then I think you're missing the point.

[QuikSand]

My explanation for KRBs in short yardage covered this point. If the KRO is missed, but the team gets a yard anyway and makes the first down, a KRB is awarded. An example would be KRB made= +4 KRB missed = +1. When this happens on third and one, the first down is made and a KRB is awarded[I] even though the KRO was missed.

Okay, a second try. I assume that you also concede the converse, that if the run play doesn't yield "success" then no KRB is awarded. There's ample evidence to support this, just in case you're unsure. So, presumably, you also would argue that any successful KRB in your preliminary calculation then gets wiped out after the fact.

So, essentially you're arguing that there really is a phantom KRB calculation going on the way that we intuitively think of it, and that it's meaningful in determining the outcome of the play... but since it gets wiped out by this after-the-fact stuff, it's not what we see reported in the player's stat sheet.

I guess there's no possible way to refute that. I don't see what possible value it adds to the debate about what the stat that we actually see means, but I think you've created an unassailable theory that cannot be disproven.

[Vinatieri for Prez]

Nice test Quik. I think we have pretty good proof there.

I just discovered another interesting twist on KRBs. And it makes the KRB stat even less of an individual stat.

In my last game, there were 3 lost fumbles by the running backs on run plays. The all occurred on 1st and 10, and involved runs of 7, 8, and 9 yards. Ordinarily, these type of runs would be awarded KRBs. But . . . you guessed it -- there were no KRBs awarded when fumbles were lost on the run. Not in Solevision, the game logs, or the box scores.

Thus, no matter how good the run, no matter how good the blocking, a high roll on a player's punishing hitter rating on the play and the random roll on the fumble recovery would be enough to prevent a KRB from being recorded.

[QuikSand]

More good input V4P - never occured to me to watch for fumble plays, but on that point...

...a high roll on a player's punishing hitter rating on the play...

...but let's not start seeding more things as givens. We certainly have enough slipperiness in this thread already.

[Vinatieri for Prez]

More good input V4P - never occured to me to watch for fumble plays, but on that point...



...but let's not start seeding more things as givens. We certainly have enough slipperiness in this thread already.

Well, it's either that or the fumble rating, so take your pick.

[QuikSand]

Probably a whole separate thread about what exactly the "Punishing Hitter" rating accomplishes in-game. Maybe it's causing fumbles, maybe it's causing injuries (I think so), maybe it's something else. But that's good for another thread.

[Vinatieri for Prez]

I agree, but my main point is that something basically unrelated to blocking and running ability can determine a KRB.

[CU Tiger]

Ok let me take a shot here.

#1 I agree 100% that KRB is not a measure of lineman success in a vacuum

#2 It does appear that without question a KRB is awarded for a positive rushing result

#3 With a poor RB, a great Olineman wont get as many KRBs

BUT these do not have to be completely independent events.
Any code string is purely a guess here but what if

-A KRO opportunity is awarded to lineman 1-5 (LT,LG,C,RG,RT)
- The first check is Does the KRO succeed (IF LT is selected for KRO does he win his individual battle)
If no then no KRB will be awarded and play will not result in KRB result.
In other words if the KRO fails there is a maximum limit placed on how successful the run can be. I.E. play result = < 5yards (this may be field position specific) The remainder of the play and dice rolls continue and determine where in that range it falls but we KNOW it wont exceed 5 yards.
If KRo opportunity succeeds, more code is played but the POTENTIAL for a succcessful KRB is present, but the potential for a fail is also present. Basically this doesnt eliminate any possible results.
At this point no value is determined for play result
- The remaining calcualtion for the play determine whether play is a success or a failure. This will include RB value match ups AS WELL as possibly the other linemen's match ups. A sum total is awarded to all these mini battles and a numerical score given based on the number of these battles won.

For example LT is KRO LT = success
LG "wins"
C "wins"
RG" loses"
RT "wins"
(I would think each of these would also be weighted based on run direction)
So we have a 4-1 victory
maybe there is a chart that looks like
0 victories = -15<play result<0 yard
1 victories = -5<play result<2 yard
2 victories = -1<play result<4 yard
3 victories = 0<play result<10 yard
4 victories = 4<play result<20 yard
5 victories = 8<play result<100 yard

In our example with a 4-1 result AND a successful KRO we now know that the run will result in a 4-20 yard gain. the RB skills now determine a final set of dice rolls, based on who on D won their " battle" to determine where in the range the result falls.

If the play result falls above the KRB minimum a KRB is awarded if not then no KRB is awarded.

In this review a given linemen could win his Key opportunity, but NOT get a KRB

But if he lost the opportnity, not only would he not get a KRB but the play would also be doomed to a poor result.

There are so many variable I have no clue how you set up a test. But I feel very strongly that there is more than 1 dice roll at play. There may even be a margin of victory in those individual weighted match-ups that determines play succes.

In this example the RB plays into every possible outcome. Every linemen could win their individuals, but if the RB loses the play would still get a 4 yard gain, but NO KRB would be awarded. So the Lineman HAS to win to qualify for a KRB but a win does NOT guarantee a KRB.

So the awarding of a KRB iss a posst event, but its result ALSO plays into the outcome

Thoughts?
Tear it apart, where are my flaws?

[Vinatieri for Prez]

I think this what I was saying with my hidden "potential KRB" theory in post 74. How pancakes are calculated bolster this possibility.

[QuikSand]

Ok let me take a shot here.

#1 I agree 100% that KRB is not a measure of lineman success in a vacuum


#2 It does appear that without question a KRB is awarded for a positive rushing result

#3 With a poor RB, a great Olineman wont get as many KRBs


The key observation is that your item #2 above also has a flip side, I believe. I have yet to see any compelling example to refute the notion that on a non-successful run, a blocker is still awarded a KRB.

So far, there are a few key definitions of a successful run that seem pretty clear:

-on 3rd down, the run must reach the 1st down to be successful

-on 1st and 10, the run must get 5 yards or more (maybe this could be expanded to on 1st down, must get 1/2 of the yards toward another first down?)

I'm guessing that there's a magic fraction at work with a 2nd down run -- that it needs to get something like...say, 2/3 of the remaining yardage toward the first down to count as successful. I don't know what the perfect fraction is, but I'm confident there is one, and 2/3 is always a worthy guess.

Anyway... until someone can show me *a single case* where the running play nets a first down or some other clear case of "success" and yet there is no KRB awarded... I'm sticking with the obvious conclusion: that the KRB is an award exclusively based on the "success" of the rushing play.

In any event -- this concept of a successful rushing play traces back rather neatly, if not perfectly, to the Football Outsiders stat.


And once we concede that runplay=success ==> KRB awarded, and runplay=notsuccess ==> no KRB awarded... then there's not reason at all to cling to this stat meaning *anything else* but a measure of whether the run play met success. You can dream up any sort of intermediate calculations you like, you can draw parallels to games played with dice, or whatever you want-- but in terms of the KRB stats that we actually see, they are irrelevent fantasies.

Until someone can prove -- with a single case, from any single game, at any time -- that the KRB does not simply follow some reasonable notion of "success" for a running play, there's nothing else to debate here. You, or OldGiants, or anyone else might theoretically be absolutely right about how all the dice-rolling might happen under the hood, and you can concoct whatever complicated theories you like... but in terms of the stuff we actually see (just talking KRBs here, not including pancakes which could actually be informative) it doesn't make one whit of difference. A hidden process that we can neither track nor record nor test might indeed be true, or might be false, but it doesn't matter either way.

As long as the perfect relationship exists between "run success" and "KRB awarded," then the KRB simply cannot be a measure of anything else. It's a rock-solid impossibility.

[QuikSand]

As long as the perfect relationship exists between "run success" and "KRB awarded," then the KRB simply cannot be a measure of anything else. It's a rock-solid impossibility.

While I think the above statement is actually the simplest way of putting all this, I will (again) try another rephrasing.

If you are inventing some sort of dice roll system to explain how you think blocking might work inside the game, but is not actually reported -- that's fine. But there's no reason to suggest that those internal calculations have anything to do with the reported stat of "key run blocks" -- they are just something else, which we cannot record or measure directly. Nothing wrong with those theories (well, perhaps that they can neither be proven nore disproven, and that they have no practical implications) except they have nothing at all to do with the original question.

A theory about what other calculations besides KRBs are doing something important in the game might be entertaining, but is unrelated to the search that this thread posed -- which is to understand what the stat that we actually see is really measuring. I'm very confident now that we have answered that question - that the KRB stat is simply a measure of overall running play success, and then apportioned back to individual blockers in some way, with some game elements suggesting a causative relationship that simply isn't there.

[CU Tiger]

QS,
I follow what you are saying, and from a stand point of "How do I maximize my KRB stat" I agree 100%.
Run success = KRB
Run Failure = No KRB

I think the danger in your wording is someone reading this for the first time, knowing your extensive testing history, and concluding (wrongly) that since KRBs are awarded after the fact, that in fact, OL play no role in running success.

I think this is absolutely not accurate and have test data to back up my thoughts. On a basic level this is a very easy test to set up and run. All other variable are equal test 1 with ALL PRO linemen test 2 with undrafted rookie scrubs. Net results was a 13% increase in YPC and a 14% increase in KRB while KRO remained static.

[gstelmack]

I think the danger in your wording is someone reading this for the first time, knowing your extensive testing history, and concluding (wrongly) that since KRBs are awarded after the fact, that in fact, OL play no role in running success.

That is simply not true, and anyone reading that into this is not paying close attention.

What is clearly (to me at least) being stated is that KRB/KRO does not tell you how good the O-Lineman is. And KRB% is even MORE misleading.

If you want someone to add:

Blocking(all OLine + FB + TE) + RB - Defense => run success chance
run success = KRB

Then fine.

To find out how good your O-lineman is, you need to find out what percentage of his team's KRBs he's getting. And combine it with the ALY for the team to see how good the team's O-Line as a whole is.

[CU Tiger]

Greg,
I follow exactly what you are saying.
The only thing I think we vary on is ALL OTHER FACTORS HELD CONSTANT a better Olineman will have more KRBs.

Since linemates have to be equal to hold ALL OTHER VARIABLES I think this stats does a FAIR (not good,very good or certainly not excellent :) )job analyzing whicch lineman is the best on a given team. But is just about useless when comparing linemen between teams.

It is a fun stat. Can add slightly to immersion, I suppose. But just going out and signing the OT with the most KRBs from last year will NOT fix your running game.

[digamma]

Thanks for all the work done here.

Slight error to report on the "Linemen" page in "Team Statistics."

It seems the "RPly" statistic and "OPct" columns factor in the run plays for both your team and against your team. For example, in WOOF, my Jeffersonians have 897 Running Plays and a 49.6 OPct. According to our rushing stats, we only have 484 rushes, but our opponents have 413 rushes on the season----897.

Only significance here is that the OPct column can help us with one of the first assumptions QS made--that there is a KRO assigned for each run. Going with this, I have 484 Running Plays and 445 KRO's. My quarterback has 48 carries on the season. Subtract those and we get 445 KROs and 436 RB carries. Allowing for a few designed QB runs, we're pretty much dead on the 1:1 ratio of one running play = one key run block opportunity.

Didn't see this reported elsewhere (but that's entirely possible).

[gstelmack]

Greg,
I follow exactly what you are saying.
The only thing I think we vary on is ALL OTHER FACTORS HELD CONSTANT a better Olineman will have more KRBs.

Since linemates have to be equal to hold ALL OTHER VARIABLES I think this stats does a FAIR (not good,very good or certainly not excellent :) )job analyzing whicch lineman is the best on a given team. But is just about useless when comparing linemen between teams.

It is a fun stat. Can add slightly to immersion, I suppose. But just going out and signing the OT with the most KRBs from last year will NOT fix your running game.

This we agree with, and I'm not sure that what is being said above contradicts any of this. Note my comment about having to also factor in Adjusted Line Yards to determine your O-Lineman's worth. In fact, by stating that a KRB is awarded only after the fact, you REINFORCE that this stat can't be compared across teams. And we're stuck with the fact that a great Lineman paired with a sucky RB leads to the lineman looking sucky.

I view this situation kind of like QB completion percentage. By itself, this stat is not a worthy comparison across QBs, since WRs factor so heavily in. But at least you are only dealing with 2 players, not 6+ in the running case. And it is possible with a QB to look up how many incompletions were bad throws, drops, etc. No such comparison is possible with O-Linemen. You have no idea when looking at any given play whether or not the Lineman did his job. That's why I think people are so hopeful for alternate explanations; if a KRO picks the key blocker, then you determine if it was successful, THEN determine if the RB hits the hole, the stats have a lot more meaning. But as it stands now figuring out how good your lineman really is turns into a giant puzzle involving divination, with his sacks allowed being the only reasonable cross-team stat (even that is not perfect depending on the QB, but it is reasonable), and OPct being comparable vs Adjusted Line Yards to figure out his contribution on a team.

When you combine this with the increased scouting error in 2k7 pushing people to use stats more heavily, it seems like a bad situation.

[QuikSand]

I think this stats does a FAIR (not good,very good or certainly not excellent) job analyzing whicch lineman is the best on a given team. But is just about useless when comparing linemen between teams.

Yup.

[Vinatieri for Prez]

But as it stands now figuring out how good your lineman really is turns into a giant puzzle involving divination, with his sacks allowed being the only reasonable cross-team stat (even that is not perfect depending on the QB, but it is reasonable), and OPct being comparable vs Adjusted Line Yards to figure out his contribution on a team.

Don't forget pancakes too. I think they may be very helpful cross-team stat (and related to running, not passing like sacks) and may even have some value between different teams. Of course, that is educated speculation based on my discussion of pancakes above.

[yabanci]

....
A theory about what other calculations besides KRBs are doing something important in the game might be entertaining, but is unrelated to the search that this thread posed -- which is to understand what the stat that we actually see is really measuring. I'm very confident now that we have answered that question - that the KRB stat is simply a measure of overall running play success, and then apportioned back to individual blockers in some way, with some game elements suggesting a causative relationship that simply isn't there.

I haven't really followed this thread so some of this surely is repetitive, but I totally agree with you on this conclusion, although I wouldn't say it necessarily measures "success" (I know it's not really your word). Sometimes it does measure success, but not always.

A KRB is a simple stat that is counted when certain conditions are satisfied, not some special ability. Here are the basic rules:

As already noted, the obvious is that KRBs are always counted (a) on 1st and 10 runs for 5 or more yards and (b) on any run that results in a first down or touchdown (minor exceptions noted below). It's fair to say those are successful runs.

But there are other situations where KRBs are counted that hardly can be considered successful.

For example, KRBs are counted for runs of 5 or more on 1st and long:

1-20-KCY30 (2Q: 00:17) Corey Peck ran around the left tackle for 5 yards. Tackled by OLB Bryan Hayes, assisted by DE Pete Hatcher. Key block delivered by Renaldo Riddols.

A 5-yard run on 1st and 20 is as "successful" as getting sacked for a loss of 5 yards on 1st and 10. They both result in a terrible 2nd and 15.

Some other examples of the same thing, KRBs being awarded on 1st and long:

1-15-BUF28 (2Q: 07:09) Shandrel Hogan ran inside the left tackle for 5 yards. Tackled by OLB Kenny Orta. Key block delivered by Nicky Hutton.

1-15-PIT35 (2Q: 13:06) Allen Barrett ran inside the left guard for 6 yards. Tackled by S Broderick Garner. Key block delivered by Leslie Harbaugh.

1-20-LAX16 (2Q: 09:33) Terry Burnett ran around left end for 13 yards. Tackled by DE Warren Sexton. Key block delivered by Marcus Duran.

KRBs are also counted on 2nd and long runs that come within 1 or 2 yards of a first down. Some examples of this:

2-10-LAX46 (1Q: 06:50) Jack Sanderson ran outside the right tackle for 8 yards. Tackled by OLB George Kennedy. Key block delivered by Reggie Sweeney.

2-10-LAX20 (1Q: 06:09) Terry Burnett ran around left end for 9 yards. Tackled by S Chad Dennis, assisted by ILB Lester Palmer. Key block delivered by Daniel Cook.

2-9-LAX21 (1Q: 10:03) Clarence Thompson ran inside the right tackle for 8 yards. Tackled by DT De'Shane Ryan, assisted by S Niko Prescott. Key block delivered by Marcus Duran.

2-10-LAX46 (2Q: 05:17) Shandrel Hogan ran a trap outside the right tackle for 8 yards. Tackled by S Les Forbes. Key block delivered by Marcus Duran.

2-9-LAX09 (2Q: 02:44) Marshall Massa ran around left end for 8 yards. Tackled by ILB James Lee, assisted by DE Angelo Cross. Key block delivered by Randall Springer.

The same is true of 3rd and long runs that come within 1 or 2 yards of a first down. Running for 9 yards on 3rd and 11 in your own territory isn't successful by any means, but a KRB is counted anyway. Some examples:

3-11-LAX37 (4Q: 07:42) Shandrel Hogan ran inside the left tackle for 9 yards. Tackled by CB Shannon Duncan, assisted by OLB Arnie Armstrong. Key block delivered by Daniel Cook.

3-10-MIN46 (4Q: 02:10) Jimmy McDaniel ran around right end for 8 yards. Tackled by DT Rondell Chandler. Key block delivered by Hunter Cortez.

3-9-LAX21 (3Q: 02:28) Frank Reitz ran a draw inside the left guard for 8 yards. Tackled by DT Alex Burton, assisted by ILB James Lee. Key block delivered by Donnie Lechler.

(note: this is where you see the strange results on 2nd and 3rd and long. A guy can run for 17 yards on 3rd and 20 and no KRB, but a guy runs for 7 yards on 3rd and 9 and a KRB is counted)

The main exceptions to the rules are that KRBs are not counted on (a) QB scrambles that satisfy the criteria and (b) runs that satisfy the criteria but where the ball carrier fumbles after gaining the first down, even when his own team recovers (example):

Los Angeles: I formation, strength is right. The defense is in a 34 and 3-deep zone coverage, keying aggressively on the pass. Martin is blitzing.
2-4-PIT27 (4Q: 02:08) Terry Burnett ran around the left tackle for 14 yards. Tackled by OLB Kyle Barlow. The ball was fumbled and recovered by LAX Daniel Cook to the PIT13.

In my view, although KRBs are interesting to look at, they don't tell you a whole lot and can be deceiving, as already has been stated. In multiplayer, you often see teams with a RB rated very highly in power inside and first down running, and the team runs 100% of the time on 2nd, 3rd, or 4th and short. Another team that passes in those situation or plays it evenly might not rack up the KRBs but might have a higher conversion rate and gain more yards in those situations. What is better, completing a 25-yard pass on 2nd and 1 or rushing for 2 yards and a KRB? The team that runs all the time in these short yardage situations and/or run in long yardage situations against pass defenses is going to have a higher percentage of KRBs, but that doesn't necessarily tell you much about the value of their running back or offensive linemen, just like the QB on a 3-13 team who racks up passing yardage against 4th quarter prevent defenses doesn't tell you much about his value.

[QuikSand]

Thankss for the examples, yabanci. Tough to say what is goin on in the 1st-and-long situations, but it sure seems like (for better or for worse) that the "success" is determined by 5 yards or more, regardless of how many yards to go for the first down. I don't like that, but it does still seem to fit the general concept -- just a poor execution.

For the 2nd and even 3rd down situations, it does seem to be a percentage -- so getting at least close to the first down is worth a "success" somehow. I confess the more detail we see here, the more the implementation seems to be straying away from the cleaner definition used by Football Oustiders (all based on % of yardds to go, and adjusted based on score situation) but so far, it does still hold up to some notion of "success" broadly defined.

Thanks for the useful examples.

[Sgran]

Okay, let's test what we know then with an example:
You have 2 similarly rated lineman at the same position. After your last preseason game, one player is 1 of 5 on KROs, the other is 0 of 1. Who do you like better?

[Ben E Lou]

Okay, let's test what we know then with an example:
You have 2 similarly rated lineman at the same position. After your last preseason game, one player is 1 of 5 on KROs, the other is 0 of 1. Who do you like better?The one with the better ratings.