GRASSY KNOLL: Significant, career-long ratings masking.

[Ben E Lou]

A story told in this thread (http://www.operationsports.com/fofc/showthread.php?t=59566) makes one suspect that significant career-long masking is in FOF2K7. Here's another story, of a guy who crept a relatively small amount, but who consistently outperformed his ratings, and significantly outperformed other QBs on the same team, using the same game plan and same receivers, who had similar ratings.



2009 PRE-DRAFT

http://www.younglifenorthdekalb.com/oliveras/predraft2009.jpg




2009 END OF SEASON

http://www.younglifenorthdekalb.com/oliveras/endseason2009.jpg




2015 END OF SEASON (First time he went all red.)

http://www.younglifenorthdekalb.com/oliveras/endseason2015.jpg



2022 FA1 (Current bars, and likely the max that we'll ever see)

http://www.younglifenorthdekalb.com/oliveras/current2022.jpg



CAREER-LONG RATINGS PROGRESSION
yyyyc=Precamp, year (yyyy-2008)
yyyyd=Postcamp, year (yyyy-2008)
yyyye=EndSeason, year (yyyy-2008)

<table align="left" border="1"> <tbody><tr> <td>Name</td> <td>Pos</td> <td>Team</td> <td>2009c Current</td> <td>2009c Future</td> <td>2009d Current</td> <td>2009d Future</td> <td>2010c Current</td> <td>2010c Future</td> <td>2010d Current</td> <td>2010d Future</td> <td>2010e Current</td> <td>2010e Future</td> <td>2011d Current</td> <td>2011d Future</td> <td>2012e Current</td> <td>2012e Future</td> <td>2013e Current</td> <td>2013e Future</td> <td>2014d Current</td> <td>2014d Future</td> <td>2015e Current</td> <td>2015e Future</td> <td>2016e Current</td> <td>2016e Future</td> <td>2017e Current</td> <td>2017e Future</td> <td>2018e Current</td> <td>2018e Future</td> <td>2019d Current</td> <td>2019d Future</td> <td>2020e Current</td> <td>2020e Future</td> <td>2021e Current</td> <td>2021e Future</td> <td>2022b Current</td> <td>2022b Future</td> <td>Overall Current Change</td> <td>Overall Future Change</td> </tr> <tr> <td>Oliveros, C.J.</td> <td>QB</td> <td>Atlanta</td> <td>11</td> <td>35</td> <td>13 (2)</td> <td>38 (3)</td> <td>17 (4)</td> <td>39 (1)</td> <td>18 (1)</td> <td>40 (1)</td> <td>21 (3)</td> <td>40 (0)</td> <td>24 (3)</td> <td>42 (2)</td> <td>31 (7)</td> <td>45 (3)</td> <td>35 (4)</td> <td>46 (1)</td> <td>36 (1)</td> <td>47 (1)</td> <td>44 (8)</td> <td>44 (-3)</td> <td>45 (1)</td> <td>45 (1)</td> <td>45 (0)</td> <td>45 (0)</td> <td>46 (1)</td> <td>46 (1)</td> <td>47 (1)</td> <td>47 (1)</td> <td>47 (0)</td> <td>47 (0)</td> <td>48 (1)</td> <td>48 (1)</td> <td>48 (0)</td> <td>48 (0)</td> <td>37</td> <td>13</td> </tr> </tbody> </table>




STATISTICAL PERFORMANCE

http://www.younglifenorthdekalb.com/oliveras/careerstats.jpg


And, if you're wondering if he's a "system" player, I had a backup QB who had bars suited to the same game plan, and for most of the time he was with us, bars as good or better than those of Oliveras. He maxed out at 44/44, but looked like he would have been further unmasked with more playing time. It's not a huge sample size, but it's worth noting that in 281 attempts, he had a 83.4 QB rating, using the same game plan and same receivers. His bars at the top of his game (comparison to Oliveras's max is in parenthesis)

Screen: 32 (36)
Short: 61/64 (0)
Medium: 67/72 (73)
Long: 66/74 (72)
Deep: 75 (55)
Third Down: 58/64 (56)
Accuracy: 18/39 (51)
Timing: 100 (50)
Sense Rush: 8 (54)
Read Defense: 0/11 (18)
Two-Minute: 45/60 (65)
Scramble Freq: 78 (82)

As I mentioned in the other thread, it is particularly interesting to note that with Oliveras--you know, that guy with the ZERO rating in short passing--at the helm, we have consistently been near the top of the league in ypa on short passes. It could be argued that the backup had ratings that were as well suited, if not *better*, than those of Oliveras to run my offense, yet Oliveras left him in the dust.

Thoughts? Specifically, do you believe that there's significant, career-long masking in FOF2K7?

[beargrowlz]

Thoughts? Specifically, do you believe that there's significant, career-long masking in FOF2K7?


Yes, I do. In fact, I think it's prevalent enough that I significantly take into account a player's past statistical history when making transactions.

I've seen too many players perform far above their "bars" to think otherwise.



Cheers.

[revrew]

I believe the opposite is also true, that players can have a tendency toward career-long UNDERperformance as well. No matter how good their bars are, they just don't live up to the billing.

Admittedly, this is purely anecdotal conjecture.

[QuikSand]

As I mentioned in the other thread, it is particularly interesting to note that with Oliveras--you know, that guy with the ZERO rating in short passing--at the helm, we have consistently been near the top of the league in ypa on short passes.

Well, I'll toss in a partial explanation for this limited obeservation.

I believe that:

-the ratings for the various lenght of passes correlate with the likelihood of the passer throwing a "bad pass" in the absence of a hurry or a pass defensed by the defender (these are noted in the game logs in various ways, and calculated in the detailed passer stats in the box scores)

-the likelihood of a "bad pass" on short passes is lower than with longer passes

So... I'm guessing that by putting these two together, it's probably true that (all else equal -- same blockers, same receivers, etc) the difference between a good short passer and a bad short passer isn't likely to be all that much, in terms of completion percentage.

So... in yards per attempt, if completion percentage isn't likely to vary all that much in this situation anyway, what else is there?

-YAC by the receivers
-distance of passes selected that qualify as "short" in that particular area of the game (remember, this term varies wildly in use form place to place in FOF)

I'll basically dismiss YAC, thought it could be a factor to some degree - I'm guessing that a SkyDog team doesn't focus too heavily on "getting downfield" unless he mistakenly lapses into a common sense interpretation of that rating, rather than its FOF meaning.

So... I'm guessing the main culprit here is play selection. An AI team (bnasically everyone but your team) is probably splitting their short passes fairly equally between screens, 0-4 yards, and 5-8 yards... and a typical completion might go for, say, 5 yards. I would also guess that a SkyDog team likely throws more of the 5-8 yard variety, and fewer of the screen variety when throwing short... and therefore, probably has a typical yardage for a completion of something like 7 yards.

If the typical team, or even one that's slightly better, is completing their short passes at a 70% clip for 5 yards each, they will end up with 3.5 ypa. SkyDog team might only connect on 60% (less because of a lousy short passing QB, and also maybe the effect of some slightly longer passes), but at 7 yards each, he'll still average 4.2 yards per attempt -- better than the better skilled, but less innovative play caller on the typical AI team.

I am not by any means throwing out your boy's accomplishments, and I think creeper players like this have real merit (and tend to really skew the whole game in ways that less advanced managers don't realize)... but in this limited case, I can see how just about any QB in a SkyDog (or, to be candid, in a QuikSand) offense is likely to be among the league leaders in ypa on short passes, due mostly just to play selection.

[Ben E Lou]

Quik:

The problem with that theory is that I'm using short pass distances quite close to what the AI recommends for at least one of my teams. On the flip side, though, is the fact that while I don't value Getting Downfield, having receivers rated highly in it *is* a residual effect of the way I draft. And all of my receivers were acquired via the draft. Right now my best four have 88, 77, 74 and 61 Getting Downfield ratings, so I suppose that could have something to do with it over time.

[Anthony]

i think this is good and bad.

Good: one really can have a "Tom Brady" type player, taken in later rounds, maybe undrafted, go on to have a great career. that really mirrors the NFL's "diamond in the rough" scenario, where any guy taken at any spot in the draft has the potential to be great, it's all about timing and the right situation (the whole "for every Peyton Manning there's a Ryan Leaf").

Bad: this seriously negates the importance of ratings. i mean, really, what is the point behind Oliveros being rated so low in all but Scrambling if he's gonna drop 40+ td's? it doesn't correspond to what the game suggests his ceiling is. to me, ratings are all about ceilings - the system and players surrounding a guy will determine how successful he can be, his ratings will determine his peak/maximum output. i'm not expecting 40+ td's from a guy rated like Oliveros like i am from someone rated Very Good/Excellent overall. i'm not expecting Chris Sims to get 40+ td's, but I am expecting that from Peyton Manning.

one thing this does clearly illustrate is if you're willing to give playing time to anybody - he will realize his potential and often surpass it, moreso than the guy rated similar but is trapped on the bench. but then again - who would even consider letting Oliveros start for anything other than testing the game? i look at Oliveros and i immediately say "roster filler".

[revrew]

This is one of the things that I might like to see changed about the game: so many players only show their potential after they've been starting. And even then, it shows up as red bars, not green.

I'd rather see a scenario where players in practice "show the coaches something." This would be indicated by a jump in green bars. It could be in TC or the preseason. It's the sudden re-evaluation of a player. Whether they can turn those green bars into red would still be a question, and one that can only be verified by playing time. But to HAVE to start a player to see if he's better than advertised or not really creates a quandry for GMs. What? I have to start every loser just to find out who isn't? Isn't that what training camp and preseason is for?? Instead of players having their redlines adjusted based on "creeping" "breakouts" etc, I'd like to see their green lines adjusted based on "practice performance." If the guy sucks in practice, he's not going to suddenly play well when you make him the starter. But that's what guys like the one in this thread are doing.

[Anthony]

you are completely correct, and i agree with your idea. a jump in green bars just to give some indication that there's "more underneath the hood" that warrants a closer look/increased playing time. if you have Mike Vick you're not gonna all of a sudden make DJ Shockley your starter just to see if he's a diamond in the rough.

something like this almost makes you wish for an injury to your starter, just so you have justification to start scrubs like this (Tom Brady becoming the starter when Bledsoe went down).

furthermore - something like this doesn't seem right. so throughout all of Oliveros' college career, combine, and first couple years he had no talent for long passes. then by 2015 he somehow miraculously is rated in the 70's. that means for a better part of his career he fooled people into thinking he didn't have an arm and all of a sudden can throw deep? that doesn't make sense. i can see a guy showing he's more accurate than the league has believed him to be. or through increased playing time his timing and sense rush improves (that makes a ton of sense). but to have one year you can't throw long and then the next you can? that completely makes the ratings system moot.

i like this addition to the game, it makes it deeper and forces you to take every player seriously, but as far as it correlates to an actual real life scenario - no, i don't get it.

[Ben E Lou]

This is one of the things that I might like to see changed about the game: so many players only show their potential after they've been starting. And even then, it shows up as red bars, not green.

I'd rather see a scenario where players in practice "show the coaches something." This would be indicated by a jump in green bars. It could be in TC or the preseason. It's the sudden re-evaluation of a player. Whether they can turn those green bars into red would still be a question, and one that can only be verified by playing time. But to HAVE to start a player to see if he's better than advertised or not really creates a quandry for GMs. What? I have to start every loser just to find out who isn't? Isn't that what training camp and preseason is for?? Instead of players having their redlines adjusted based on "creeping" "breakouts" etc, I'd like to see their green lines adjusted based on "practice performance." If the guy sucks in practice, he's not going to suddenly play well when you make him the starter. But that's what guys like the one in this thread are doing.

Well, in this case, his green *did* increase. Notice that he didn't get any playing time worth talking about until 2013, but by then his green had already increased by 11 of the 13 points that we'd ever see it go up. Yes, there are guys who continue to creep after they've hit all red, but I get the impression that it's rare/never that the creeping we see then is more than a very small number of points.

I'm not saying it's always the case, but that was an indication that he might be better than advertised. If there's a concern here, it's not so much that there was no indication, but that after eight years of dominating performances, and eight years of backups not doing nearly as well with the same game plan/receivers, the league's scouts can still say with a straight face that he's no better than an average starting QB.

[cuervo72]

Could be a mask, could be that he has an avoid int that is off the charts (though I think that *does* still factor into the ratings calculation). I still think that avoid int can make otherwise pedestrian QB's look pretty darned good.

[Coder]

All these discussions lead me to wonder "What kind of impact does a coach have"... I mean, could it be that these guys' ratings are actually sucketysuck when drafted, but a certain coach influences their ratings-development?

If there was a third set of ratings (current, visible-future, and masked-future), I'm pretty sure Greg would have discovered these when working with the fileformats.

[Ben E Lou]

All these discussions lead me to wonder "What kind of impact does a coach have"... I mean, could it be that these guys' ratings are actually sucketysuck when drafted, but a certain coach influences their ratings-development?I strongly doubt this. Real (unmasked) ratings have been static for a long time in FOF (probably forever), only changing with aging and random volatility-based booms/breakouts. Development can be influenced by outside factors, but I get the strong impression that ratings can't.

If there was a third set of ratings (current, visible-future, and masked-future), I'm pretty sure Greg would have discovered these when working with the fileformats.Actually, I'm pretty sure this is a "no," too. Greg's working with the unencrypted data. The ratings were encrypted in FOF2K7.

[Coder]

It does seem as if the general consensus here right now is that everything is pre-destined to happen a certain way.. there's absolutely no chance for a player to have sucketysuck ratings and then "boom" out of those ratings due to volatility, without having a masked set of bars "hidden away".. am I correct in saying that's the general belief?

[Ben E Lou]

It does seem as if the general consensus here right now is that everything is pre-destined to happen a certain way.. there's absolutely no chance for a player to have sucketysuck ratings and then "boom" out of those ratings due to volatility, without having a masked set of bars "hidden away".. am I correct in saying that's the general belief?I don't think that's the case. I haven't had any on my teams yet, but I believe that there are still volatility-based "booms" in this version (maybe they're more rare now?) Basically, I'm saying that there are the following types of ratings changes:

UNMASKING (also known as "creeping")
Player x has a true future potential of 60, but he shows up on your team after the draft as rated 15/40, but his future potential slowly climbs, and eventually he ends up 60/60. The reality in these situations is that he has always been 60 in future potential, but the league scouts had him underrated. He's not a 15/40; he's a something/60.

(Note that the purpose of this thread is to explore the possibility that sometimes the mask is never fully taken away, and you could have a player who is, say, 15/35 at the draft, and climbs to, say, 45/45, but is really, say, 60/60 or better.)

AGING
Player x is in his 9th year now, showing 60/60 ratings, but after training camp, he drops to 52/52. That's an actual drop. The ratings change. He's no longer a 60 player. He's a 52 player.

BOOMS/BUSTS
These are random, based on volatility. Player x is 60/60, and in his prime. He suddenly jumps up to 85/85 or down to 35/35 in one fell swoop. That's not unmasking, but an actual change in his real core ratings. He's no longer a 60 player. He's a 35 player or an 85 player. I can personally attest to having seen a couple of massive busts that I believe were more than just aging. For example, I had a starting LT drop 32 points in one training camp recently. As I said above, in my detailed tracking I haven't had any big random booms on my team, but that doesn't mean that they don't happen. Having seen a handful of star players who were late-round picks or undrafted, I tend to believe that some of them were booms, not just creepers.

[gstelmack]

Maybe this is a special "unmasking" case, but it seems like a 4th development path:

<TABLE border=1><TBODY><TR><TD>Name</TD><TD>Pos</TD><TD>Team</TD><TD>Hayes a Current</TD><TD>Hayes a Future</TD><TD>Hayes b Current</TD><TD>Hayes b Future</TD><TD>Hayes c Current</TD><TD>Hayes c Future</TD><TD>Hayes d Current</TD><TD>Hayes d Future</TD><TD>Hayes e Current</TD><TD>Hayes e Future</TD><TD>Overall Current Change</TD><TD>Overall Future Change</TD></TR><TR><TD>Hayes, Chester</TD><TD>LCB</TD><TD>New</TD><TD>34</TD><TD>66</TD><TD>36 (2)</TD><TD>60 (-6)</TD><TD>38 (2)</TD><TD>60 (0)</TD><TD>49 (11)</TD><TD>49 (-11)</TD><TD>46 (-3)</TD><TD>46 (-3)</TD><TD>12</TD><TD>-20</TD></TR></TBODY></TABLE>

a = draft year 1
b = post-TC year 1
c = post-Preseason year 1
d = late in season year 1
e = post-TC year 2

This is a guy whose current crept up during the year until he hit "10" experience (he started every single game), then his future immediately tanked. Between c & d was at least one sim where he was something like 48/60, then next sim he was 49/49.

Based on year 2 and subsequent, he was slowly creeping back down. But he clearly immediately hit some sort of scout "ceiling" in the middle of the year, rather than being one of those "10" guys that still show green and still creep up on both current and future.

So this may have all been masked, or my scout was completely wrong, but he hit that 49/49 in year one then slowly sank back down after that, and his future topped out once he hit "10" experience.

FWIW, he was a 42/42 after year 3 TC.

[QuikSand]

Only thing I see about Hayes that's a bit unusual is the dropoff at stage e... and if he was only a part time player that really isn't too unusual there, I don't think. But that overall pattern, where the current rating grows gradually, the future rating drops in spurts, and once he reaches full positional experience he's maxed out well short of the original forecast -- that's really pretty standard, especialy for players who didn't have great combines to support their initial projections.

[gstelmack]

Only thing I see about Hayes that's a bit unusual is the dropoff at stage e... and if he was only a part time player that really isn't too unusual there, I don't think. But that overall pattern, where the current rating grows gradually, the future rating drops in spurts, and once he reaches full positional experience he's maxed out well short of the original forecast -- that's really pretty standard, especialy for players who didn't have great combines to support their initial projections.

I'm not necessarily saying it's unusual. Just saying it's not covered by SkyDog's 3 categories. FWIW, he was not a part-time player, he started every game at CB for me as a rookie.

As for his combines, this was my first 2k7 season, so I did not have Extractor or anything to save them off. He looked reasonable, though.

[QuikSand]

Ok, fair enough... I guess if we had to classify your guy Hayes, then he'd be in the "standard" group -- the one that basically applies to the majority of players in any draft class. I think SD was trying to group together the various classes of ratings development that fall outside the "standard" progression (red up, green down, eventually he guy finishes, and eventually he fades).

[Julio Riddols]

I think the way the ratings work is perfect, because in all reality, I don't want to "know" anything about exactly how skilled my guys are. I've seen players jump in ratings but still have trouble performing, I've seen guys in bad shape barswise become stars with playing time, even if their bars didn't really show it... I've seen one year wonders, etc.

I think there are so many variables, and one of the major ones we might be overlooking is cohesion and how well a player meshes with a team on that level. I think players develop a sort of underachieving or overachieving persona in some cases, where you know the kid has the talent to perform at a very high level, but just cant seem to do it or is inconsistent. In other cases, it just may be that a guy who doesn't have the physical tools to really excel at his position still outplays those skills when you put him on the field.

Ultimately, in this way, each player could be overrated or underrated throughout his career. Intelligence, Loyalty, Leadership, Personality and Popularity may also have a lot to do with it. Weather conditions play a smaller role in player performance, but could be a factor too. I'm sure there are also a couple hidden ratings that help determine whether a player overachieves or underachieves.

To sum it up, I think maybe the issue is how we interpret the bars, rather than the bars being eternally hidden. The mystery of it all is what keeps this game challenging. I'd rather it not be an excercise in logic based on knowing what to look for and what not to look for in a player. I want to mess up and draft a bust every now and then, or watch RB Sucky Suckerson have a 1,000 yard season just because he has that special something that helps him get that job done. This game needs its Chad Penningtons and its Joe Montanas, and Its Harold Greens and Hank Fraleys, to balance out with the rare-talent superstars, the average joes, and the total scrubs.

[Ben E Lou]

Ok, fair enough... I guess if we had to classify your guy Hayes, then he'd be in the "standard" group -- the one that basically applies to the majority of players in any draft class. I think SD was trying to group together the various classes of ratings development that fall outside the "standard" progression (red up, green down, eventually he guy finishes, and eventually he fades).

True. And actually, when you think about it, "red up, green down" is just an example of unmasking that I didn't specifically mention. It's the same thing, though: the guy comes out of the draft at 15/60, but really he's a 35. He doesn't "get worse." He never was any better than a 35. The league scouts just had him all wrong.

[Ben E Lou]

I think the way the ratings work is perfect...
Whenever I see someone use that word about any aspect of a computer simulation, I tend to pretty much discount the rest of what they have to say. You don't really mean that, do you?

[tarcone]

Im testing a theory. Do you have a screenshot of him post draft, pretraining camp?

[Ben E Lou]

Im testing a theory. Do you have a screenshot of him post draft, pretraining camp?

No, but I guess I can go back and create one. What's your theory?

[ddrrbb]

Need ratings based on performance (with potential and physical attributes as the guideline) and performance based on ratings.

I remember dropping a fortune on a FA RB (80+). My backup RB (45, maybe) had twice as many ypc and TDs at the end of the season. I guess the good back softened up the D for the scrub. ;)


db

[gstelmack]

I remember dropping a fortune on a FA RB (80+). My backup RB (45, maybe) had twice as many ypc and TDs at the end of the season. I guess the good back softened up the D for the scrub. ;)


That's a common theme. In WOOF I've got a solid guy starting who does okay, but his veteran backup can reel off those long runs like they're going out of style and has almost as many total yards as the starter. I let him start one week and it was a total disaster. Put him back at backup, and he starts reeling off 100-yard games again.

[Noop]

If we had some sort of training camp with scrimmages then it would be easier to identify the guys who may not be stars bar wise but stats wise they are great. For instance lets just say your starting QB has retired and your left with only your third string QB lets call me Diaz (the second string guy left for greener pastures) So you draft a rookie lets call him Henry and sign a career backup lets call him Kellen who barely played at his last team.

Lets say their rating is as follows...

Diaz 27/42
Henry 25/55
Kellen 23/23


Without seeing actual stats from the scrimmage I would be likely to play both Diaz and Henry over the first 8 games; allowing the one who is performing better start the final 8 games. Kellen would be nothing more then a roster spot to me...

Now if during training camp we had scrimmages then I can see something like this...

First Scrimmage
Diaz - 07/23 100 0tds 1 int 2 sacks
Henry 09/21 121 0tds 2 int 0 sacks
Kellen 10/15 97 1td 0 int 4 sacks


Second Scrimmage
Diaz - 11/20 111 0tds 2 ints 3 sacks
Henry - 10/30 90 3tds 0 ints 0 sacks
Kellen - 14/19 91 1tds 0 ints 4 sacks

----------
Based on those stats I would let Diaz sniff the field, I instead would let Kellen be my starter because he performs at a steady level while Henry shows promise but isn't there yet.

This is just a small example but what would really help us find out who the gems are in the league. Perhaps Jim can really expand on this to make it more sensible then I have but it is just a rough idea of mine. This no doubt will help the immersion alot in my opinion.

[Celeval]

Whenever I see someone use that word about any aspect of a computer simulation, I tend to pretty much discount the rest of what they have to say. You don't really mean that, do you?

That one particular increment function is perfect. :-P

[Ben E Lou]

That one particular increment function is perfect. :-P

When it comes to being a smart-ass, Kevin is PERFECT!

:D

[Julio Riddols]

Whenever I see someone use that word about any aspect of a computer simulation, I tend to pretty much discount the rest of what they have to say. You don't really mean that, do you?

Yes, I do. Perfect in that it is imperfect, and therefore a better simulation of reality. Maybe I should have said "just fine" instead?

Sometimes players just don't work out on teams for whatever reason. Sometimes bad players play well on certain teams, whether it be their system, or whatever. To be an accurate simulation of reality, I think it makes sense that our scouts don't always see how good or bad someone is. It is up to you as a GM to do your research and plug in the guys you think can do the job. If we could rely wholly on scout interpretation to determine who would be good and bad, there wouldn't be anything left to chance.

What I am saying is, why change it? Does the game need to be easier? Jeff Blake would have never been a starter in the NFL if the 2 QBs in front of him didn't get hurt. What about Samkon Gado? I'm positive there are hundreds of real-life examples of players who would likely have "nice bars" after theyve gotten the chance to start games. Why? Because they're finally put in a game situation, and they perform. It just happens that way. There are also plenty of examples of guys who were outstanding on one team, then went on to terrible seasons elsewhere.

If the game is intended to simulate managing a football team, I think its done about as well as I can expect.

If the game is intended to test our ability to put numbers in order from highest to lowest, then it needs to be fixed.

[Ben E Lou]

Yes, I do. Perfect in that it is imperfect, and therefore a better simulation of reality. Maybe I should have said "just fine" instead?

Sometimes players just don't work out on teams for whatever reason. Sometimes bad players play well on certain teams, whether it be their system, or whatever. To be an accurate simulation of reality, I think it makes sense that our scouts don't always see how good or bad someone is. It is up to you as a GM to do your research and plug in the guys you think can do the job. If we could rely wholly on scout interpretation to determine who would be good and bad, there wouldn't be anything left to chance.

What I am saying is, why change it? Does the game need to be easier? Jeff Blake would have never been a starter in the NFL if the 2 QBs in front of him didn't get hurt. What about Samkon Gado? I'm positive there are hundreds of real-life examples of players who would likely have "nice bars" after theyve gotten the chance to start games. Why? Because they're finally put in a game situation, and they perform. It just happens that way. There are also plenty of examples of guys who were outstanding on one team, then went on to terrible seasons elsewhere.

If the game is intended to simulate managing a football team, I think its done about as well as I can expect.

If the game is intended to test our ability to put numbers in order from highest to lowest, then it needs to be fixed.

You've clearly missed my point. I'm not talking about guys who have one or two good seasons somewhere and don't play well elsewhere. I'm talking about star players. Obvious star players. Guys who run faster, lift more weight, and outperform their peers on the field. Wherever they go, they perform well.

Those are the guys who shouldn't remain rated as mediocre players. Everyone and their cousin knows they're not mediocre players. I'm drafting *OBVIOUS* star players again and again and again, and they're rated as mediocre. I'm talking about a guy like WR Melvin Hutchinson. The consensus of the scouts is that he's a 5.9/6.1 player. He runs a 4.35. He puts up 24 reps in the bench press. He completes the agility drill in 6.85 seconds...

...yet EVERY SCOUT IN THE LEAGUE (despite the consensus) thinks he's a 57 coming out of the draft. That just plain doesn't make any sense to me. Surprise, surprise, he's really an 84 (or more), and is the best receiver in league history. Shocker. I'm not talking about Peerless Price; I'm talking about Randy Moss or T.O. or Calvin Johnson. Obvious nonpareil talent.

C.J. Oliveras is no Blake. He's a Manning. His best receiver dropped 15 points this year, and Oliveras didn't miss a beat--another 100+ QB rating.

Of COURSE there should be guys who show up big all of a sudden. That's why the game has breakouts. Of COURSE there should be guys rated in the 40s and 50s who perform like 60s. But when it's obvious as all get-out that the player is a superstar, the scouts should see him as a superstar, or at least a very good player--not an average starter.

[Ben E Lou]

What I am saying is, why change it? Does the game need to be easier?

And no, it needs to be harder. The ambiguous system is screwing up the AI to the point where there are multiple young players with obvious starter ability being cut every single year (usually in only the second or third year of the contract). They're very easy to spot.

[wade moore]

And no, it needs to be harder. The ambiguous system is screwing up the AI to the point where there are multiple young players with obvious starter ability being cut every single year (usually in only the second or third year of the contract). They're very easy to spot.

So just thinking outloud here - do you think a system where the AI, at least some percentage of the time, identified "creepers" would be a good step forward - or would you prefer some complete revamp of how sleepers are done?

That's what I'm not sure on - do you not like how creepers work, or do you not like that the AI doesn't identify them? For instance, in an MP league (assuming that most owners are knowledgeable enough to identify creepers) does this phenomenon bother you as much?

[Ben E Lou]

So just thinking outloud here - do you think a system where the AI, at least some percentage of the time, identified "creepers" would be a good step forward - or would you prefer some complete revamp of how sleepers are done?

I'm not sure about a revamp, but I'd advocate a system where the AI "remembers" the evaluation it made of the player at draft time. I don't have that much of a qualm with how it evaluates draftees. The problem is that it "forgets" that it thought that players were studs.


That's what I'm not sure on - do you not like how creepers work, or do you not like that the AI doesn't identify them? For instance, in an MP league (assuming that most owners are knowledgeable enough to identify creepers) does this phenomenon bother you as much?Not nearly as much, since the playing field is level, but it still creates situations where we make decisions based on FOF mechanics, rather than football strategy. I think Noop's idea of a practice or scrimmage report makes a lot of sense. I get criticized for "playing the system" in MP, but the fact is that everyone does it to some degree. If given a choice between making a decision on football strategy as opposed to FOF mechanics, I'll always choose football strategy, unless it's clear (as it is in several places in FOF) that the likelihood of success is increased by focusing on the game mechanics.

[gstelmack]

So just thinking outloud here - do you think a system where the AI, at least some percentage of the time, identified "creepers" would be a good step forward - or would you prefer some complete revamp of how sleepers are done?

I would personally vote for revamping sleepers. When a guy plays, you generally find out how good they are pretty quickly. You also find their strengths and weaknesses. Thus, I have no problem with lots of scout error for guys who have never played, but once a guy plays you find out VERY quickly what he is good or bad at. It doesn't take 4 or 5 years of starting to find out what a player does or does not have. His future ceiling should take longer than his current value, but not by a whole lot.

Keep the system of random booms / busts. I have no qualms at all with this system and is very reflective of the myriad issues that can happen to players on and off the field, including factors like lack of confidence, etc.

One thing that FOF does WELL is creating players with strengths and weaknesses. This is where I think the whole concept of system players comes in. People have given myriad examples of "good" players who tanked under a different system, but I think this is just reflective of whether or not they fit the system. Peerless Price was still a good receiver, but he had Mike Vick throwing to him and had sufficiently poor "route running" or whatever FOF stat you want to apply to break double-teams. His rating didn't tank, he just went someplace that did not play to his strengths.

Terrel Davis, on the other hand, pretty much had a very high "Hole Recognition", and that was pretty much all you needed to be a success in the Denver scheme (they want patient backs that hit the holes when they open up). Maybe good breakaway speed as well. But the rest of his ratings may have been poor.

In both cases, FOF as it stands now does a good job of modeling the peaks and valleys of the skillset vs the playcalling of the team (witness some of SkyDog's dynasties and reports and testing done with his gameplans) that allows a 40+ "mediocre" player to perform well in the system.

So what I'd expect is something like:

- Rookie bars continue to have some error in them that firms up with interviews / training camps.
- Everyone's bars firm up once they start playing. This should be based on plays, not starts. It shouldn't take lots of seasons, though, to really hone in.
- Random booms / busts stick around.
- Aging handled by reduction of appropriate skills. RBs lose speed, QBs deep balls, Linemen strength, etc, so you can see that they are physically slowing down. Endurance loss is probably another good factor here.

I'm actually a big fan of switching to a "don't show future ever" scheme (which I know was tried briefly with FOF). Instead, always show physical skills (combine numbers), and show current ratings. Drafting college players is a combo of finding kids with good current ratings, plus an aptitude (intelligence) for picking up more football skills, plus the basic physical skills. Players then develop their skills based on playing time, training camps, weight training, etc.

The reality is scouts have two major roles:

1) Identify FAs with skills that fit your system (do all the filtering for the head coach)
2) Identify college kids that have a good core to develop into future pro players.

They miss on #2 a lot more than they miss on #1. And on #1, a lot of these misses are attributable to age decline or injuries. I could see a game moving to, say, and A-F potential system like FBCB uses that gives a rough idea of how much room the guy has to develop, but showing specific numbers just confuses the issue. FWIW, I think FBCB has my favorite scouting / player rating system of any sim I've played, doing a good job of letting you know how well the player plays right now, his rough potential down the road, and allowing players to exceed or fall short of those expectations based on playing time, quality of opponent, and training camp focus.

[wade moore]

Edit: This is in response to SD, not greg who responded at the same time as me...

Fair enough. I appreciate from Jim's side that he wants (and I want as a player) there to be some grey area where it's not like ratings are clear-cut and the best players are the guys with the best ratings.

I think he's got the right goal in mind, but that it needs work - the AI "remembering" development/performance seems like one way to address it.

The "scouting report" seems like another way to me - for someone with poor ratings to show "looks better than expected in practice" or "he really seems to be learning and improving quickly" without the ratings going up..

[gstelmack]

The problem is that it "forgets" that it thought that players were studs.

I would argue that a fundamental "problem" with FOF is the lack of AI planning. For example, calling a timeout late in the half so you have time to punt rather than just letting the clock run down shows that playcalling is done in a vaccuum, or the problem of signing then immediately releasing guys. Each of these decisions is done autonomously, and simply executing the one action changes what you'd do with the following action.

I put "problem" in quotes because I feel like that is nitpicking what is in general an excellent game. I hope Jim realises that the very fact we're having these kind of debates shows how solid much of the game is. We're at the point where all we have left to do is pick on the weaknesses, but it's important to slip a positive comment in, too, even if it's a rehash, just so it doesn't seem like we suddenly hate the game.

[Ben E Lou]

Fair enough. I appreciate from Jim's side that he wants (and I want as a player) there to be some grey area where it's not like

Jim brought up the issue to the community at release time that he wondered if scout error was too great. I wonder if what's troubling me a bit about this is as simple as that. I don't know the margin, but perhaps what we're seeing is that, say the error is +/- 30, when it should be no more than +/- 15. If an 80/80 player is showing up as 50/50, then that's a problem, imo. I threw that number out there because I've personally seen unmasking of right around 30 points several times, so I wouldn't be shocked if there's 30-point masking that never gets removed.

[Ben E Lou]

Heh. Turns out I was wrong about Oliveras maxing out at 48/48. He crept up to 50/50 in year 15 of his career. I'm guessing that's a little more unmasking...

[ddrrbb]

I think some of the problem is that players don't always play near their ratings. Even if a scout has poor evaluation in different areas, he should not be blind. Either a QB can make a long/short throw or not. And if he can't, he should not be able to in a game. I can scout a practice to see if a QB can make certain throws (and the combine is televised now).

If a player has a low rating in a skill and assuming accurate scouting of that player, that player should be bad at that skill. Oliveras should not complete a short pass, EVER. He has had a 0 rating for 15 years, but I'll bet he completes the same % of short passes as someone with a 50 rating. I think there should be less extremes in some of the ratings (and the 0 rating made obsolete...don't get me started). A poor stat should be rated 15-20, not 0-3.

BTW, I guess big DTs can have a 0 pass rush rating as some go whole careers w/o getting near a QB.

[Ben E Lou]

Oliveras should not complete a short pass, EVER.

A zero doesn't mean he's your cousin who played backup QB in high school. It means he's the worst possible a pro-level QB could be at the skill. He should be the worst QB in the league at short passing to get that rating, sure. But a zero doesn't mean that he can't throw a three-yard pass.

[Synovia]

Sky, from what I've seen playing around, overall ratings are simple composites. That makes each one about 8 points overall. Scramble frequency doesnt seem to effect the overall.

We've got him losing 5 from the kick holding (which is irrellevant), 8 from Short passing (which I think is masked), and thats 13 points. That puts him at 63/63. I could see a GOOD to Very GOOD (63/63) quarterback with great recievers putting up those sorts of numbers. If the Read Defense is masked too, then it puts him up at 70.

[Ben E Lou]

Sky, from what I've seen playing around, overall ratings are simple composites. That makes each one about 8 points overall.

No. They're weighted.

[Synovia]

I dunno, we were taking player files, and flopping around the kick holding, and it seemed to make an 8 point or so difference.

If theyre weighted, whats short passing worth?

[Ben E Lou]

I don't know the weightings (never tried to work them out), but just experimenting with a few ratings makes it clear that they're not equal.

[wade moore]

Jim brought up the issue to the community at release time that he wondered if scout error was too great. I wonder if what's troubling me a bit about this is as simple as that. I don't know the margin, but perhaps what we're seeing is that, say the error is +/- 30, when it should be no more than +/- 15. If an 80/80 player is showing up as 50/50, then that's a problem, imo. I threw that number out there because I've personally seen unmasking of right around 30 points several times, so I wouldn't be shocked if there's 30-point masking that never gets removed.

I think my problem isn't the amount of scout error, but that it's really a complete masking. To me if you use the term "scout error" that means that one scout may see things differently than another. This is not a case where your dumb scout sees him as a 50/50, but my superstar best ever scout sees him as an 80/80. The seperation from realism for me is that there's this guy with a 100+ rating and EVERYONE in the league still think he should be holding a clipboard.

I would personally (and I know some would disagree) like to see a system where good scouts get it right more often than not and bad scouts are much more sketchy - rather than this complete masking for all scouts in the league for an extended period of time. I think the max we see right now is around 10 points between scouts. I would rather see it that this overall masking is gone, but there is more error between scouts.

Let's face it, at this stage it seems like scouts mean relatively little in the big picture.

[Ben E Lou]

I think my problem isn't the amount of scout error, but that it's really a complete masking. To me if you use the term "scout error" that means that one scout may see things differently than another. This is not a case where your dumb scout sees him as a 50/50, but my superstar best ever scout sees him as an 80/80. The seperation from realism for me is that there's this guy with a 100+ rating and EVERYONE in the league still think he should be holding a clipboard.Well, in general, I agree with you there. There's not really meaningful scout-to-scout variance, apart from the interview assessment. I was referring to scout error as defined in FOF. I guess at this point, I'm talking within the system that we have. I'd love to see a completely revamped staff system as much as the next guy.

I would personally (and I know some would disagree) like to see a system where good scouts get it right more often than not and bad scouts are much more sketchy - rather than this complete masking for all scouts in the league for an extended period of time. I think the max we see right now is around 10 points between scouts. I would rather see it that this overall masking is gone, but there is more error between scouts.I don't mind overall masking for *some* players. After all, NO ONE saw Tom Brady as better than a 6th round pick (although I'd argue that Brady fits the "random breakout" model more than the "creeper" model, anyway.) It *could* be argued (although I'm not sure I 100% agree with this line of thinking) that the current system is better than the one you propose because it puts the onus on the GM to be good at drafting, rather than just hiring an outstanding scout. I think there needs to be a balance there. There needs to be a better payoff for getting a great scout. At the same time, it's not like the scout is the only guy that looks at film, goes to the combine, watches practice, etc. The ability of the human, who in FOF really plays the threefold role of Owner/GM/Head Coach, should have a lot of influence here as well.

Let's face it, at this stage it seems like scouts mean relatively little in the big picture.As you know from the fairly-recent discovery that Cuervo and I stumbled upon, a good scout can be quite useful in interviews, but other than that, I agree with you here. And the more you know about drafting, the less useful they are--although they're still useful even to an "expert" player (and I'm pretty sure that they'd still be at least somewhat useful even to Jim).

[Ben E Lou]

Oh, and to be clear for those that aren't aware of this, "scout error" in FOF2K7 means that *every* scout in the league saw Oliveras as somewhere in the neighborhood of 15/38 after his rookie year, probably ranging no greater than from around 13/36 to 17/40. We checked this in IHOF on a creeper player that everyone knows is better than his advertised 45/50 ish ratings, and a bunch of people checked in, and no one who did saw him as worse than 45/49, or better than 46/51. I also checked this with some obvoius creepers in a test MP league and saw the same thing: little/no variance from one scout to the next. However, it has been confirmed by examination that there can be very significant differences from one scout to the next in the interview process. The better your scout, the more likely that he's right, but even an Excellent young talent, Excellent position guy can be dead wrong at times.

[gstelmack]

It *could* be argued (although I'm not sure I 100% agree with this line of thinking) that the current system is better than the one you propose because it puts the onus on the GM to be good at drafting, rather than just hiring an outstanding scout.

The problem is that it can take a dozen or more drafts of painstaking work to become good at drafting, and by that time I think you've lost a lot of players. Sure, I may be good in drafting in a year or two, but it's because I don't have the time to play single-player at all and am only in MP leagues that I'm absolutely killed here.

[Vinatieri for Prez]

I think there needs to be a balance there. There needs to be a better payoff for getting a great scout. At the same time, it's not like the scout is the only guy that looks at film, goes to the combine, watches practice, etc. The ability of the human, who in FOF really plays the threefold role of Owner/GM/Head Coach, should have a lot of influence here as well.

I think a balance could be struck in this way. A better scout eliminates the masking you see in veterans, but stays pretty much the same way you see it for the draft. Thus, you maintain the unknown quantity in drafting, but give an edge to GMs in evaluating talent by not having to rely so much on stats (which can be difficult, especially for positions like O-line).

As an example, most scouts would see Tom Brady as the 15/41 guy pre-draft. Now, let's assume Brady is a stud and performs like one for a few years but his ratings are still very masked (he shows as a 35/52. After a few years in the league, the excellent scouts should show him as 80/86. The good scouts maybe show him as 70/75 and the poor scouts still show him at 35/52. Then, as he gets into years 6-7 of his career, it becomes obvious to everyone, he is 90/90. Thus, the good scout is a real help in judging guys in the 2-5 year range. Then, if you combine that with an overhaul to the staff/scouting system in the next FOF game, we are talking a whole new element to the game that I would find fascinating.

[Synovia]

Vinateri, I like that idea, but with a couple caveats, which Tom Brady brings up nicely.

When Bledsoe went down, Brady wasnt #2 on the depth chart. He was #3 behind another young guy: Damon Huard.


Huard turned out to be pretty decent, but its obvious that even at that point, the Patriots scouts felt that while Huard was the better player (hihger on depth chart) that Brady had "shown something".

That makes me think that good scouts should have a better idea than bad scouts, even early. Maybe even if its only something like the bad scouts seeing Brady as 15/41->17/45 PostTC, vs an excellent scout seeing 15/41->17/48.

The good scout should ALWAYS (barring complete misreads) have a better feeling about a good player than the bad scout.


The one thing this would REALLY make a difference on, IMO, would be the AI. AI teams with good scouts would move 'real' good players up the depth chart quicker, and be less likely to release good players. It would also make it easier for Rex to properly set players.

[Vinatieri for Prez]

Yeah, I agree with that mostly. But the Brady example is still a little different. It was his own team's scouts that "saw something." So, with my example, you would have a little "on-team" modifier (like we see now, but more pronounced) to help see these gems, while other teams (even with excellent scouts) would have no clue. So, if you have an excellent scout who is on the team, you get a little bonus there, but if the excellent scout is on the other team, he still doesn't see it because he hasn't had the opportunity.

[Synovia]

That I could agree with.

If you wanted to go even further, there could be a smaller bonus for "In division" because teams watch more film of their division, send scouts to the open training camp, etc.

[Vinatieri for Prez]

I should add that if you go in this direction, where scouts do become more valuable, it may be worth splitting the scout position into separate offensive/defensive ones. Just a thought.

[gstelmack]

I actually think Brady is a good example of wanting to evaluate players like:

Patriots scouts see him as: 15 / 85, 30% chance of reaching it
Some other scout might see: 15 / 80, 5% chance
A third might see: 15 / 50, 90% chance

Or somesuch. The Patriots scouts did see some key intangibles that made them like Brady, but not enough to pick him very high with some of the other needs they had. They liked what they saw in TC, and he was working his way up the depth chart. He was also helped by the fact (playing time wise) that Bellichik did not like Bledsoe a whole lot; he thought he was easy to gameplan against.

But until Brady got some starts, they did not know how he'd REALLY turn out. It didn't take long for them to realise he was pretty good. They always thought he had the potential in him, but weren't sure if he'd ever reach it.

[CU Tiger]

I think a "fuzzy targets" system would actually do exactly what you are saying better.

One where bars are fuzzy at the edges.

So may a given scout sees a rangee of current and a separate rangee at future ratings.

With Tom Brady The pats saw his current as say 15-30 and his future as a 60-80 (even they probably didnt see a 90)

Where as 30 other teams probably looked at him and saw a current of 5-30 and a future of 40-65, by "blurring" the values and not giving us hard numbers we are ledt to guess where in the range they fall. of course over time a scout should zero in on a players abilities but this would vary with familiarity. For exampl after year 1 the Pats may have saw 28-30/75-85 guy while others RANGE PROBABLY HANDT CHANGED MUCH.

But the one thing I have a huge problem with, and I think what Skydog is lameenting the most, is no NFL caliber scout should see Brady today and say, "that guy? hell he is a 45 at best" Much like the Zero short passer can still hit a 5 yard swing the worst NFL scout should still no a HOF Qb isnt a back up.

[Ben E Lou]

Heh.

The final word on C.J. Oliveras is this: I left the Atlanta team to start a dynasty thread (http://www.operationsports.com/fofc/showthread.php?t=59743) with another team. I had drafted a young stud QB a year or so before I changed teams, planning to let him get mentored until Oliveras decided to hang 'em up. As soon as the AI took over the team, it cut Oliveras. Miami signed him to a two-year deal, then promptly sat him on the bench for an entire season. However, in his 17th and final season, Miami's QB got hurt near the end of the year, and C.J. Oliveras, age 39, rated 44/44 at the time, got a final chance to prove his worth in the playoffs. How'd he do, you ask?

Division Playoffs: 19 for 25, 263 yards, 2 TDs, 0 INTs, 135.9 Rating
AFC Championship: 21 for 29, 227 yards, 2 TDs, 1 INT, 103.6 Rating
Super Bowl: 17 for 28, 177 yards, 2 TDs, 0 INT, 102.8 Rating

Yes, sports fans, Oliveras closed out his career with a stellar postseason and a ring, became the first QB in league history (in 2030) to be named to the HoF, and leaves this text simmer wondering if Oliveras ever would have started a game had that 11/35 rookie not been drafted by a human player.

http://www.younglifenorthdekalb.com/oliveras/predraft2009.jpg

[wade moore]

I think this is an area where maybe the biggest improvement to the SP experience could be made with potentially the smallest coding time from Jim.. if he could find a way to have AI teams recognize and utilize players like this...

Edit: But the trick is you probably don't want AI teams to be able to just immediately pick them out.

[IanW]

This is the Pats media release about their second 2001 preseason game (Brady as a 2nd year QB, around 4th on the depth chart)

"Charlotte, N.C. – Forget Jimy Williams line-up changes. How about Bill Belichick starting Drew Bledsoe for the first series, inserting Tom Brady for the next four and then putting Bledsoe back in the game in the second quarter to finish the last four series of the first half. It was part of Belichick’s plan to use as many players as possible early in the game against Carolina’s best and overall, the Patriots looked impressive, beating up on the Panthers, 23-8.

Both Brady and Bledsoe looked good with Brady passing for 11 of 18 for 122 yards and a touchdown and Bledsoe completing 9 of 16 passes for 73 yards. Bledsoe had one pick, a ball that bounced off Troy Brown and into the hands of Panther Michael Rucker.

And again, the Patriots defense looked solid, with only a third quarter touchdown breaking its scoreless streak at six quarters. Bobby Hamilton, Richard Seymour and Matt Stevens had active games on that side of the ball.

New England’s first touchdown of the preseason came on a drive that began at the end of the first quarter as Brady impressively moved his team from its own 20. David Patten, who was called on often early in the game (6 catches for 65 yards in the first half), snagged a nice 16-yard slant to put the Patriots up, 10-0 with the extra point.

The big plays in that drive were a Brady to Bert Emanuel hook-up for 14 yards on third and 8 and a Patten reverse for another 14 yards. The drive numbers: 10 plays, 80 yards, 4:46."

Source http://www.patriots.com/games/index.cfm?ac=completereportsdetail&pid=1960&pcid=41&special_section=na

Quick note on game 4


"- Brady was also strong, connecting on 11-of-19 for 166 yards and the touchdown to Faulk. Brady also handled most of the holding duties on field goals and extra points. Damon Huard did not play, but Belichick said that was because the game was well in hand and he didn’t see the point of putting Huard in just to hand the ball off."

So, yeah, they might have seen something ... but if a low-round QB does 11/18 for 122 and then 11/19 for 166, then they are probably worth playing in FOF. I guess the moral of the story is to spread QB time around in preseason and see who can play.

[gstelmack]

In my office league draft, we had one punter go through undrafted. No workout, but someone grabbed him in late FA. His attributes show as 61/61 to me (63/63 to him), but only 61% developed. I'm curious to find out what is lying after TC and the season go by. Based on all these discussions, I expect to find it's the ratings. But I may have to fire my scout, given the whole "yes coach, he's only 61% developed, but he's not going to get any better" conversation I'll be having with him...

[Ajaxab]

But g, won't you have to have that discussion with every scout you hire? ;)