FOF7 and Draftee Combine Scores
 

So, how does that combine score really relate to the quality of the player drafted? I did some black box analysis of this in 2007 and I will slowly (note: slowly) be doing the same analysis for FOF7.

Methodology:
I started a league with the combine accuracy set to 100 and advanced the league through the first season to the Pre-Free Agency stage.

I will use the FOF 2004 draft file creator to create 1,000 player drafts of players at each position (QB, RB, FB, ... , S) and each quality level (625, 600, 575, ... , 375).

I will import each draft into FOF7 and see what combine scores are generated from the quality of player.

I will export the data using the built-in Data Export Utility.

The output will be brought into Excel and parsed.

I will present the raw data at the top and my observations at the bottom.

I will try to make a new post for each position as I get the data exported and analyzed.

Hopefully this data will be useful because it's a metric ton of work.

Data to follow.

Edit:

Suggestions for use:
First and foremost, let me say that this analysis is not guaranteed to produce results. It is not intended to provide absolutes. Jim has created an incredible game that has outliers who "break the mold". This was intended as black box analysis of the most accurate look at draftees that we have available.

That being said, here are some ways this data could be used:
1. Finding delineation points in draftee scores. Data tends to indicate that there is little difference in QBs with a 4.60 40 yard dash time vs. 4.63 but there is a bigger difference between 4.63 and 4.66.
2. "Banding" of players. It would be very easy to write an Excel macro or function to estimate or "band" a combine rating based on position player. For example, a QB with a dash of 4.63 looks like it is between a 625 and 550 while a Solecismc score of 42 for the same player would indicate somewhere between 625 and 475. These bands could be summed, averaged, what have you to compare between two similar-looking and/or -graded players of the same position.
3. Mathematical weights and values based on individual combine scores. I am going to try this for a few seasons to see how this works out. If it does, I might post my method (if I get back to my winning ways in the CFL). :)

My feedback would be to put the accuracy to 50
I think most leagues are going there and would probably have the most impact for the community.

I started this with 625s but got derailed do to work. Will be glad to help you crunch the numbers.

QBs
 

Skipping Workouts
Out of 11,000 QBs 461 did not receive combine scores. These 461 QBs were scattered somewhat evenly throughout the ratings.


40 Yard Dash

Solecismic

Strength

Agility

Long Jump

Position Drill

Notes

Players' combines must be taken as a whole. Individual combine scores are not guarantees of the exact quality of the player. For example, a QB draftee may have a 40 time of 4.43, indicating he could be one of the best in the game. However, if his Solecismic is 9, Strength is 9, Agility is 8.3, Long Jump distance is 90, and his position skill is 50, this is likely just a really fast QB who won't be a very good one.

Indicators of good players:

• Speed kills. A dash time of 4.57 or better could possibly be an indication of a Superstar QB.
• Strength kills. A strength of 14 or higher could possibly be an indication of a Superstar QB with 16 or higher suggesting but not guaranteeing a future hall-of-famer.
• Long Jump kills? Long jump of 114 or better (9'6") could possibly be a Superstar QB.
• Position drill is the strongest indicator of quality. An 87 or higher could possibly be a Superstar QB.

Indicators of bad players

• A dash of 4.97 or higher tends to indicate a backup or roster-filler.
• A strength of 9 or lower tends to indicate a backup or roster-filler.
• Long jump of 101 or lower (8'5") tends to indicate a backup or roster-filler.

Cautions

• Solecismic differences are very fine and can be used to differentiate at the top and bottom of the scales but be careful to try to pick out the difference between a 550 and 525, for example.
• There are so many outliers with Agility it is hard to have a "hard and fast rule" about what to draft. I would probably avoid any Agility higher than 7.80 though.

Crud. I've done a truckload of work at 0 already with QBs done and RBs almost done. I may have to complete those at 0 and possibly come back later at 50.

When you say "each quality level" does that mean that you set all ratings to that level? (So a 550 means that every skill was 550, for example?)

That is correct.

:popcorn:

+1

** in the correlated skillsets.

Which, by the way, old 2k7 combine correlations may or may not apply any longer. It might not be nearly as simple as "bench is 50/50 deep/long" or "agility is pass blocking."

Are you suggesting there's evidence that the old correlations no longer exist?

Or just that we shouldn't necessarily take it for granted that they do?

If the former...please share at least a skeleton of what you're seeing.

Because of this...
Be very careful about how you phrase these:

• FOF drafts aren't generated with all flat ratings. There are plenty of non-superstar QBs with dash scores better than 4.57, for example. And even on position drill, I'm seeing a QB with a Position drill of 90, but a 7 strength, 8'5" jump, 4.81 dash. (Overall grade is 3.8, too.) Pretty sure that guy isn't a superstar. Some people will jump on your phrasing and say "Oh...87 drill = sure superstar!!!"

I will rephrase these. I meant "indicative" as "points to but doesn't guarantee". I'll change these now.

Also, I am editing the first post to add "Suggestions for use."

+ 1 here.

First cut I look at on these is around the 500 or 425 level. Does this guy meet each combine at a 95% confidence he's over a 500 level player. A player with > 500 on all his sills. is going to be a damn good player. A player with over 425 in all skills will be a decent performer. Anything else is gravy.

I really look to see a cut of players that I have a high confidence will perform. From there, I try to cull out the superstars.

This is correct. However, before game obfuscation wouldn't you expect a, say, 600-level player to have higher average combine scores than, say, a 425-level player? We can't look at data before obfuscation, therefore we need to have SOME type of fixed point of reference off which to build. If we just set the overall value, then you may be looking at a 600-level player with a horrible single score that skews the data.

Let's talk concrete examples. Suppose you have two QBs that the game, before obfuscation, rates as 475-level QBs. So, suppose the QB individual skill ratings look like:


That could result in combines that look like:

The presence of the great speed and great strength for QB2 are offset by the weak combine scores elsewhere. As I edited the posts above to say, the 4.50 and 16 do not guarantee a superstar QB but those combine numbers are that of an outstanding QB whether the QB is actually outstanding.

None of this works, however, unless we have some picture of what a 625-rated 40-time is, and strength, and solecismic score, etc. That's why the individual ratings were set in the input file. Failure to do so will put our data all over the map and render it useless.

By looking at a draftee's each individual combine score and seeing what range it falls in, we might be able to determine, approximately, where the QB could wind up overall, no matter what our scout says. That 19/25 may wind up being a 50/50 decent backup and "creep up."

It makes sense in my head and I'm probably doing a bad job of explaining it.

RBs
 

Running Backs
Skipping Workouts
Out of 11,000 RBs 439 did not receive combine scores. These 439 RBs were scattered somewhat evenly throughout the ratings.


40 Yard Dash

Solecismic

Strength

Agility

Long Jump

Position Drill

Notes

Players' combines must be taken as a whole. Individual combine scores are not guarantees of the exact quality of the player. For example, a QB draftee may have a 40 time of 4.43, indicating he could be one of the best in the game. However, if his Solecismic is 9, Strength is 9, Agility is 8.3, Long Jump distance is 90, and his position skill is 50, this is likely just a really fast QB who won't be a very good one.

Important high rating delineation points:

• Dash < 4.47.
• Strength > 19.
• Agility < 7.09.
• Long Jump > 126.

Important low rating delineation points:

• Dash > 4.65.
• Strength < 15.
• Agility > 7.35.
• Long jump < 111.

Cautions

• Solecismic differences are very fine and can be used to differentiate at the top and bottom of the scales but be careful to try to pick out the difference between a 550 and 525, for example.

You're right. Here's where I'd rate that QB:

Dash = 4.81 = ~500 rating
Sol = ?? = unknown, skipping
Agi = ?? = unknown, skipping
Str = 7 = ~400 rating
Jump = 101 = ~425 rating
Pos = 90 = ~600 rating

Ignoring the unknown values, (500+400+600+425)/4 = 481.25. You probably have a QB (depending on what the Solecismic and Agility combines are) that has the combines of between a 475 and 500 player.

These scores can be used for approximations. We'll never be guaranteed to know but it can help differentiate between potential 45/45 players and 32/32 players that never amount to much.

This is great stuff thanks so much.

The latter, Quik. I'd share if it were the former but I haven't even played the game. It looks like a lot about how combines are generated has changed -- at least existing knowledge about correlations should probably be re-verified?

This study is kinda a Malcpow-for-FOF7, and that one relied on correlation knowledge in order to draw conclusions. Every sign about FOF7 points to far more obscuring and far less 'guarantees', so even a subtle change in correlations can affect the value of a study greatly.

Fullbacks
Skipping Workouts
Out of 11,000 FBs 404 did not receive combine scores (Error 404 - Workout not found. Ahem...moving on...). These 404 were scattered somewhat evenly throughout the ratings.


40 Yard Dash

Solecismic

Strength

Agility

Long Jump

Position Drill

Notes

Players' combines must be taken as a whole. Individual combine scores are not guarantees of the exact quality of the player. For example, a QB draftee may have a 40 time of 4.43, indicating he could be one of the best in the game. However, if his Solecismic is 9, Strength is 9, Agility is 8.3, Long Jump distance is 90, and his position skill is 50, this is likely just a really fast QB who won't be a very good one.

Important high rating delineation points:

• Dash < 4.78
• Strength > 28
• Agility < 7.14

Important low rating delineation points:

• Strength < 20
• Agility > 7.60
• Long jump < 102

Cautions

• Once again Solecismic differences are very fine and can be used to differentiate at the top and bottom of the scales but be careful to try to pick out the difference between a 550 and 525, for example.

Tight Ends
Skipping Workouts
Out of 11,000 TEs 430 did not receive combine scores. These 430 were scattered somewhat evenly throughout the ratings.


40 Yard Dash

Solecismic

Strength

Agility

Long Jump

Position Drill

Notes

Players' combines must be taken as a whole. Individual combine scores are not guarantees of the exact quality of the player. For example, a QB draftee may have a 40 time of 4.43, indicating he could be one of the best in the game. However, if his Solecismic is 9, Strength is 9, Agility is 8.3, Long Jump distance is 90, and his position skill is 50, this is likely just a really fast QB who won't be a very good one.

Important high rating delineation points:

• Agility < 7.33

I know that's not a lot but the TEs are pretty blurry as far as ratings go. The only obvious lines are the ridiculously good values.

Important low rating delineation points:

• Agility > 7.75
• Long jump < 102

Cautions

• Most rating differences (Solecismic, Strength, etc.) are very fine and can be used to differentiate at the top and bottom of the scales but be careful to try to pick out the difference between a 550 and 525, for example.

Wide Receivers
Skipping Workouts
Out of 11,000 WRs 413 did not receive combine scores. These were scattered somewhat evenly throughout the ratings.


40 Yard Dash

Solecismic

Strength

Agility

Long Jump

Position Drill

Notes

Players' combines must be taken as a whole. Individual combine scores are not guarantees of the exact quality of the player. For example, a QB draftee may have a 40 time of 4.43, indicating he could be one of the best in the game. However, if his Solecismic is 9, Strength is 9, Agility is 8.3, Long Jump distance is 90, and his position skill is 50, this is likely just a really fast QB who won't be a very good one.

Important high rating delineation points:

• Dash < 4.41
• Strength > 14
• Agility < 6.97
• Position > 63

Important low rating delineation points:

• Dash > 4.61
• Strength < 9
• Agility > 7.34

Cautions

• Ignore Long Jump when it comes to evaluating WRs. It would appear it is still used for another rating like punt/kick returns that have a pretty even spread among the ratings.
• Solecismic rating differences are very fine and can be used to differentiate at the top and bottom of the scales but be careful to try to pick out the difference between a 550 and 525, for example.

Could you outline how you go from using the data export utility to getting the draft combines into a spreadsheet?

Right now I hit 'export data' and get 27 league named files, but have no idea how to find the draft combine data in them or how to get that information into a spreadsheet.

When you click Data Export, it will create a bunch of CSV files in a directory. The file that you're looking for is called "rookies.csv". This file will only export if Free Agency has begun (I know this is right) but the draft has not completed (I believe this end point is right).

Sorry for the recent lack of data, all. I have been working 55-60 hour weeks and I just haven't had a lot of time for analysis. I hope to get some work done on this sometime this week.

Thank you, found the file ('rookies'). I really appreciate your help!

So what exactly should I assume is going on with guys whose combine scores don't match their scouting bars at all?

Let's say I'm looking at a QB with scores of 50 Solecismic, 4.50 dash, 16 bench, 7.00 agility, 9'8" jump, and 99 position drill, all apparently elite scores, but he's just middling in bars. Should the scouting bars always just get ignored in a situation like this?

In the old version, pretty much yes. This version, most likely, but it remains to be seen if anything has changed. Unlikely.

I have such a hard time committing to draft a guy in the early rounds (1, 2, 3) based solely on combine scores, especially if his bars are not even "middling" but hovering toward zero. Is he really going to creep above 50?

In later rounds, I'll try and grab good combine scores and did notice one guy I gave up on early approach 50 with another team but it wasn't a big loss.

A MLB who can't stop the run? A DE with no pass rush skills? A CB who can't cover M2M? I don't care how good his combine scores are if I see players like that.

I agree

My approach is:

If combines match bars, very good.

If combines are higher than bars, good.

If bars are higher than combines, not so good.

I have to disagree with the answer you were given. While we can't speak to exactly how everything is working in this version, in the older version you needed to look at specific bars as they were static bars. For instance, with a QB the Sense Rush bar was the static bar. So, if the above listed QB had a high SR bar it's a sign that his other bars are most likely masked.

It's not as simple as that, and neither is drafting, so it's tough to give you cut and dry examples. MalcPow's old drafting thread will give you an idea of what to look for, however how much it correlates to the new game is something I'm still working out. This thread is giving us the data to be able to compare the old to the new.

In relation to the example you gave, if ALL the bars were crap then the guy will probably be crap, regardless of his combine scores. However, if there's some indication that he may be better than the sum of his bars say, then he may end up being the elite player his combines showed.

Did you see better correlation between bars and final developed players in FOF7??

I saw a WR with elite combines but low (around 20s) AD, RR and END starting bars, around 35 3rd down catching and he´s 29% developed also... And am afraid of Drafting him... What would you recommend?? Looking for a better option? Or ignore these bars and go for him?? :popcorn: :popcorn: Our Draft scout was set to 75

In FOF2K7 I didn´t have any doubt but here...

How elite?

What was his BPR like?

I'm waiting for TC yet but it looks like between 85-100 bar!

Any chance you're gonna finish this, stretch? I enjoy using it as a point of reference.