Valuation of players and draft picks

Temporary forum for the discussion of the Tuesday Morning project, to enhance the Madden 08 off-the-field experience.
Drizzt_13
All-Pro
Posts: 202
Joined: Wed Apr 18, 2012 7:32 pm

Re: Valuation of players and draft picks

Postby Drizzt_13 » Mon May 21, 2012 9:39 pm

torontogrudlies wrote:
Drizzt_13 wrote:Is the game log only just one game, because I would prefer the entire seasons worth of stats.


There is a team file which will have an entire season of stats for particular players on a team. Please post more detail about what stats we're looking for.



QB: Yards per game
HB: Yards per game receiving and yards/game rushing
FB: Pancake blocks, Yards per game receiving
WR: Receptions, Yards Per Game, Drops
TE: Receptions, Yards Per Game, Drops, Pancake Blocks, Sacks Allowed
OL: Pancake Blocks Sacks Allowed
DL: Sacks, Tackles, Tackles for loss
LB: Sacks, Tackles, Tackles for loss, Pass Deflected, Interceptions, Catches Allowed
DB: Tackles, Tackles for loss, Pass Deflected, Interceptions, Catches Allowed

As I mentioned above I think If I plugged in a seasons worth of data with player ovr as the input and y/game as the output into a spreadsheet and did a curve fit that would be the best way to get an equation for relating stats to a perceived overall rating. I was also thinking that for different facets of the game I would have different equations. For OL I would run different curve fittings for both Pancakes/game and sacks allowed /game, and then use a weighted average of the two as the perceived overall.

The Equations for each positions would look something like this. I will use the notation CF(x) to indicate that x will be scaled to overall using curve fitting. For example, CF((Y/ga)*(rec/rec+Drops)) means that I will take whatever numbers are output by the formula (Y/ga)*(rec/rec+Drops) and scale them to produce an ovr between 70-99.

QB: CF(Y/game)
HB: Coach Run%*CF(Rushing yards per game)+Coach pass %*CF(Receiving Yards/game)
FB: Coach Run%*CF(pancake Blocks)+Coach pass %*CF(Receiving Yards/game)
WR CF((Y/ga)*(rec/rec+Drops))
TE: Passing Team Value:Coach Agressiveness%*CF((Y/ga)*(rec/rec+Drops))+Coach Conservative%*CF(Sacks allowed/game)
Running Team Value: CF(pancake block/game)
Final Value: Runing Team Value*Coach Run%+Passing Team Value+Coach Pass%
OL: Coach Run%*CF(Pancake/Game)+Coach Pass Percent*CF(Sack/Game)
DL: Coach Run %*CF(Tackles+1.5*tackles for loss)+Coach Pass %*CF(Sack/Game)
LB: Run Stop: CF(Tackles+1.5*tackles for loss)
Pass Stop: A*CF(Sack/Game)+B*CF((Pass Defl.+3*Interceptions)/Catch allowed)
A+B=1, a and B will change if it is a 4-3 or 3-4 scheme as 3-4 OLB's rush the passer more and 4-3 OLB's are supposed to cover more
Final Value: Run%*Run Stop+Pass%*Pass Stop
DB: Run Stop: CF(Tackles+1.5*tackles for loss)
Pass Stop: A*CF(Sack/Game)+B*CF((Pass Defl.+3*Interceptions)/Catch allowed)
The CF(x) will probably have to be done separately for SS, FS and CB.

HRun Stop: CF(Tackles+1.5*tackles for loss)
Pass Stop: A*CF(Sack/Game)+B*CF((Pass Defl.+3*Interceptions)/Catch allowed)

For certain parts of this we could always replace/game with per snap but I'm not sure that is worth all the craziness that comes with tracking snaps

Hope this is somewhat coherent.

Re: Valuation of players and draft picks

Sponsor

Sponsor
 

User avatar
torontogrudlies
MVP
Posts: 806
Joined: Fri Jul 15, 2011 4:56 pm

Re: Valuation of players and draft picks

Postby torontogrudlies » Mon May 21, 2012 10:10 pm

?

Code: Select all

INDIVIDUAL STATS



PASSING                        CMP  ATT  YDS  PCT   YPA SACK   TD  INT LONG RATING
Lonnie Lofton                  180  355 3329   50   9.3   31   32   13   74   98.1
Maurice Thomas                  38   89  884   42   9.9    7    6    5   87   78.1
David Alexander                  7   15  132   46   8.7    2    2    2   33   77.6
Steve Mason                      8   20  162   40   8.0    4    2    2   43   62.8

RUSHING                        ATT  YDS   AVG LONG   TD  FUM
Sean Wright                    344 1001   2.9   23    4    2
Lomas Brooks                   128  524   4.0   59    3    2
Benji Walker                    12   79   6.5   28    0    0
Calvin Johnson                  18   65   3.6   22    0    0
Todd Jones                      21   35   1.6    6    2    0
Terrell Jones                   22   14   0.6   13    1    1
Thomas Jones                     1    1   1.0    1    0    0
Steve Mason                      1    1   1.0    1    0    0
David Alexander                  1   -1  -1.0    0    0    0
Lonnie Lofton                   28   -5  -0.1    8    2    2
Maurice Thomas                  11  -10  -0.9    8    0    1

RECEIVING                      REC  YDS   AVG LONG   TD DROP  YAC
Greg Payton                     46 1099  23.8   73    5    4  122
Jason Houston                   46 1057  22.9   66    9    2  122
Leslie Gaines                   36  734  20.3   67    7    1  103
Darrius Donaldson               11  331  30.0   87    4    0   79
Ted Smith                       20  314  15.6   31    1    3   54
Sean Wright                     22  252  11.4   53    4    1   81
Jeremiah Brown                  15  243  16.1   33    4    2   10
Eddie Woolford                  15  183  12.1   37    3    2   34
Lomas Brooks                     4  130  32.5   74    3    1   47
Rodney Pritchett                 4   69  17.2   24    1    0    1
Todd Jones                       8   46   5.7   15    1    0   13
Julius Lane                      3   25   8.3   17    0    0    0
Steve Holmes                     1   14  14.0   14    0    0    4
Benji Walker                     2   10   5.0    5    0    0    0
Ernie Owens                      0    0   0.0    0    0    1    0

KICKING                        FGM  FGA  PCT FGSBLOCKED  XPA  XPM  PCT XPSBLOCKED KICKOFFS TOUCHBACKS
Tony Edmunds                    40   45   88          0   53   53  100          0        0          0
Robert Lopez                     0    0    0          0    0    0    0          0      108          4

PUNTING                        ATT  YDS   AVG LONG BLOCKS IN20 TOUCHBACKS
Robert Lopez                    71 2925  41.1   55      0   31          8

KICK RETURNS                   ATT  YDS   AVG   TD LONG
Darrius Donaldson               38 1172  30.8    1  101
Greg Payton                     21  445  21.1    0   49
Steve Holmes                     2   39  19.5    0   22
Eddie Woolford                   1    4   4.0    0    4
Jason Houston                    2    6   3.0    0    6
Carlos Frazer                    1    0   0.0    0    0
Alonzo Diaz                      0    0   0.0    0    0

PUNT RETURNS                   ATT  YDS   AVG LONG   TD
Darrius Donaldson               16  147   9.1   19    0
Alonzo Diaz                     12   81   6.7   14    0
Greg Payton                     17   71   4.1   12    0
Steve Holmes                     1    4   4.0    4    0
Carlos Frazer                    0    0   0.0    0    0
Jason Houston                    0    0   0.0    0    0
Eddie Woolford                   0    0   0.0    0    0

DEFENSE                        TOT LOSS SACK   FF FREC  YDS   TD  INT  RET   AVG DEFLECTIONS SAFETIES CTH ALLOW BIG HITS
Steve Clayton                   60    1    0    1    0    0    0    5   26   5.1           8        0         3        0
Barry Hall                      59    3    0    1    0    0    0    0    0   0.0           8        0         9        0
Alonzo Diaz                     54    2    1    1    0    0    0    3   52  17.3          11        0        29        0
Jerry Perz                      53   13    1    1    0    0    0    6   74  12.3          19        0        26        0
London Henry                    51   11    1    1    1    8    0    2   27  13.5          13        0        10        0
Levon Jones                     48   13    3    0    0    0    1    4   52  13.0           6        0         7        0
Jake Crosby                     41    3    0    2    1    8    0    5   51  10.1          19        0        25        0
Matthew Jenkins                 40    8    1    2    1    2    0    1    5   5.0           6        0         4        0
John Young                      31    7    0    0    0    0    0    3   19   6.3           4        0         5        0
Justin Matthews                 30   12    9    2    0    0    0    0    0   0.0           0        0         0        0
Emmitt Hornsby                  30   10    8    1    1    2    0    0    0   0.0           0        0         0        0
Thomas Jones                    24    9    2    0    1   17    0    0    0   0.0           0        0         0        0
John Parker                     23    2    1    0    0    0    0    0    0   0.0           3        0         0        0
Chris London                    22    2    0    0    0    0    0    4   48  12.0           5        0         6        0
Lorenzo Knight                  20    0    0    0    0    0    0    1   29  29.0          10        0         0        0
Lamont Rogers                   20    1    0    0    0    0    0    1    0   0.0           8        0         3        0
Jamel Curry                     19    0    0    0    0    0    0    1    8   8.0           1        0         0        0
Jimmy Hales                     18    5    8    0    0    0    0    0    0   0.0           5        0         0        0
LeMarcus Wright                 16    4    2    0    0    0    0    0    0   0.0           0        0         2        0
Jason Houston                   10    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Leslie Gaines                    7    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Darrius Donaldson                4    0    0    0    0    0    0    0    0   0.0           0        0         0        0
R. J. Sylvester                  4    1    0    1    0    0    0    0    0   0.0           0        0         0        0
Steve Holmes                     4    0    0    0    0    0    0    0    0   0.0           4        0         3        0
Greg Payton                      3    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Sean Wright                      2    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Harry Spikes                     2    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Austin Nelson                    1    0    1    0    0    0    0    0    0   0.0           0        0         0        0
Benji Walker                     1    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Brad Gray                        1    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Steve Mason                      1    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Eddie Woolford                   1    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Adron Short                      1    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Stephen Hanson                   1    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Jeremiah Brown                   1    0    0    0    0    0    0    0    0   0.0           0        0         0        0
Carlos Frazer                    1    0    0    0    0    0    0    0    0   0.0           0        0         0        0

BLOCKING                      PANCAKES SACKS ALLOWED
Harry Spikes                        59             2
Carlos Frazer                       52             4
Antrel Hendricks                    52             3
Stephen Hanson                      47            11
Adron Short                         46             9
Eddie Woolford                      32             0
Calvin Johnson                      25             0
Todd Jones                          13             0
Julius Lane                         10             0
Ted Smith                           10             0
Leslie Gaines                        8             0
Jeremiah Brown                       5             0
John Parker                          4             0
Greg Payton                          4             0
Jason Houston                        3             0
Sean Wright                          2             1
Lonnie Lofton                        2             0
Benji Walker                         1             0
Terrell Jones                        1             0
Lomas Brooks                         1             2
Darrius Donaldson                    1             0
Omar Clayton                         1             0
David Alexander                      1             0
Donny Blackmon                       1             0

Drizzt_13
All-Pro
Posts: 202
Joined: Wed Apr 18, 2012 7:32 pm

Re: Valuation of players and draft picks

Postby Drizzt_13 » Mon May 21, 2012 10:28 pm

Yeah this is what I need, How do I get it?

User avatar
torontogrudlies
MVP
Posts: 806
Joined: Fri Jul 15, 2011 4:56 pm

Re: Valuation of players and draft picks

Postby torontogrudlies » Mon May 21, 2012 10:32 pm

In the Stats subfolder, under Teams. You also have to make sure you have this functionality turned on....and it can be in txt or html format I believe.

Drizzt_13
All-Pro
Posts: 202
Joined: Wed Apr 18, 2012 7:32 pm

Re: Valuation of players and draft picks

Postby Drizzt_13 » Mon May 21, 2012 10:41 pm

How do I turn on that functionality? I also don't seem to have a teams folder in tthe stats subfolder

User avatar
torontogrudlies
MVP
Posts: 806
Joined: Fri Jul 15, 2011 4:56 pm

Re: Valuation of players and draft picks

Postby torontogrudlies » Mon May 21, 2012 10:51 pm

Menu--Settings--Game Options--Stats

It will have options for text/html etc. as well as asking what you want your stats directory to be.

User avatar
torontogrudlies
MVP
Posts: 806
Joined: Fri Jul 15, 2011 4:56 pm

Re: Valuation of players and draft picks

Postby torontogrudlies » Mon May 21, 2012 10:53 pm

Is this something which you're going to need to copy/paste hundreds of times? If it's something that tedious, let me know and I can set aside time to design something to spit out a CSV file or something.

Drizzt_13
All-Pro
Posts: 202
Joined: Wed Apr 18, 2012 7:32 pm

Re: Valuation of players and draft picks

Postby Drizzt_13 » Mon May 21, 2012 11:12 pm

Just doing it once or twice shouldn't be a big deal, if we feel the need to test it extensively then that could be useful.

User avatar
torontogrudlies
MVP
Posts: 806
Joined: Fri Jul 15, 2011 4:56 pm

Re: Valuation of players and draft picks

Postby torontogrudlies » Wed May 23, 2012 12:28 am

Drizzt_13 wrote:
As I mentioned above I think If I plugged in a seasons worth of data with player ovr as the input and y/game as the output into a spreadsheet and did a curve fit that would be the best way to get an equation for relating stats to a perceived overall rating. I was also thinking that for different facets of the game I would have different equations. For OL I would run different curve fittings for both Pancakes/game and sacks allowed /game, and then use a weighted average of the two as the perceived overall.




What are your thoughts regarding the evaluation of players who haven't played as much? Say, a backup center?

Drizzt_13
All-Pro
Posts: 202
Joined: Wed Apr 18, 2012 7:32 pm

Re: Valuation of players and draft picks

Postby Drizzt_13 » Thu Mar 28, 2013 10:01 am

OK so I've been working on correlating stats with player overall for injury stats but I think it is going to be more relevant to the ways in which teams evaluate players. Stats don't correlate very strongly with the number of snaps a player plays but they do correlate pretty well with their overall rating.What I propose we do is take a stat like tackles/run snap or sacks/ pass snap, use the best fit line to generate an OVR from that stat, and then average it with the OVR generated from other stats. This average would be weighted based on what a team is looking for in that player, so if they want a pass rushing DE the OVR generated from sacks would be higher than that generated from sacks.

I'm going to give the equation I generated for the position and the average and median margin of error for that position.

3-4 DE
OVR Based on Tackles/Run Snap
OVR=(Tackles/Run Snap+.1904559)/.0032928715
Average Error: 2.40888
Median Error: 1.7815
OVR Based on Sacks/Pass Snap
OVR=(Sack/Pass Snap+.031182284)/.0004715774
Average Error: 5.432
Median Error: 4.889

As you can see it's much easier to predict run stopping ability than pass rushing ability, the question is should we automatically weigh run stopping a little higher to make the equation more accurate or should we base it only on team run pass preferences.

4-3 RE
OVR Based on Tackles/Run Snap
OVR=(Tackle/Snap+0.095965194)/0.0020803411
Average Error=8.718
Median Error=7.08929
OVR Based on Sacks/Pass Snap
OVR=(Sack/Snap+0.0511181967)/0.0008058839
Average Error=5.262
Median Error=4.97

It makes since that as 4-3 DE's are primarily pass rushers that sacks would be a more accurate measurement. Interestingly the biggest deviation from the average comes with specialized players especially on tackles. Freeny recorded very few tackles despite being rated very highly and so was rated 20 points lower than he should have been as a run defender, similarly there were a number of players who were primarily run stoppers and came out rated much higher based on their run defense then they were overall.

This brings up issues when it's going to come to negotiating salaries. Run defenders tend to get paid less then pass rushers because the NFL is a passing league. We need a way so that even if a team is only looking for their DE's to be run suffers they don't pay them like pass rushers.

4-3 LE
OVR Based on Tackles/Run Snap
OVR=(X+0.1278032308)/0.0025490323
Average Error=5.6244
Median Error=3.555

Sacks Doesn't correlate with OVR at all for LE so I compared it to SPD and ACC and got some equations that fit a little better.

SPD Based on Sacks/Pass Snap
SPD=(X+0.0445515706)/0.0008446923
Average Error=6.234
Median Error=5.289
ACC Based on Sacks/Pass Snap
ACC=(X+0.0774318643)/0.0011647376
Average Error=4.787
Median Error= 3.31
The easiest thing would be to just plug the new Stats based speed and acceleration into the OVR equation for LE's and generate a new OVR based on that. I did some quick equations to see how much that effected the OVR and raising the speed and acceleration 6 points each jumped the OVR by about 4 points and lowering them both did the same. Raising just one increased it by about 2. That seems acceptable to me that on average the stats generate a rating that is within 4 points of the actual rating.



3-4 Nose Tackles
OVR Based on Sacks/Pass Snap
OVR=(X+0.0281238983)/0.0011816029
Average Error=10.375
Median Error= 9.747

Most nose tackles recorded 0 sacks and one recorded 3 so I didn't bother trying to figure out a pass rushing component. It makes sense to me that Nose Tackle would be a position that is really hard to evaluate based on the tape.


4-3 DT's
This is another one where stats do a really poor job of matching up with OVR's so I compared to specific rating instead
SPD Based on Sack/Pass Snap
SPD=(X+0.0132438702)/0.000303879
Average Error= 7.483 (If you remove the highest outlier this drops .5 points
Median Error=5.65
ACC Based on Sack/Pass Snap
ACC=(L3+0.0114995015)/0.0002225043
Average Error=8.397
Median Error=7.7966
TCKL Based on Tackle/Run Snap
TCKL=ln((M2/0.0012376247))/0.05838888121
Average Error=6.193
Median Error=5.166

If you raise or reduce everything by the average amount in the OVR formula you get a 6-7 point shift in the OVR.

4-3 MLB's
OVR based on Tackle/Snaps
OVR=(X+0.0849724496)/0.0023704521
Average Error=5.577
Median Error=5.153
AWR based on Ctch Allow/Pass Defended
AWR=(X-9.1235453086)/-0.086300349
Average=10.397 (Two of these have an error of 30-40 points when they are removed the error shrinks to 7.41
Median Error=6.666
ACC based on Ctch Allow/Pass Defended
ACC=(X-24.2697767061)/-0.2535815347
Average Error=3.914
Median Error=2.6295
SPD based on Ctch Allow/Pass Defended
SPD=(X-17.5793095)/-0.19078707
Average Error=4.78
Median Error=3.63

Interestingly tackle rating doesn't correlate well with any stat. Since we are basing a lot of things off the same stat Catches Allowed/Passes Deflected even though it would make sense to adjust every possible ability score we can we're really just doubling the influence since we're basing it off the same statistics. That is not necessarily a bad thing it's just something to consider.

Feel free to check these stats against some of your own franchises generated with different sliders. I'm fairly certain that ultimately we're going to have to make a program that automatically calculates the equation of how different stats correlate to different ratings if we want to handle using sliders other than the ones I used for this test specifically. The main purpose of this is just to figure out what correlates with what and whether it's best fit as a line or a curve. Also to get an idea for what we can expect the range of deviation from these equations to be so that we can change the way things are weighted. Since 3-4 DE's fit the stats generated for them better than NT's we can either just say predicting nose tackles is simply harder for teams or we can automatically weight the actual OVR higher for NT's so that the amount of deviation is decreased.

Anyway this is all food for thought and there's more coming soon, any fresh perspectives on this would be nice.

Edit: Just had a Eureka moment. One thing that has been frustrating me for a while now is that teams will always error in rating players in the same direction. If a player who is projected to have 10 sacks has 15 every team in the league will rate him higher than he should be (though in differing amounts based on how good they are at scouting). This is problematic because in the NFL sometimes one team will underrate a player while another will overrate them. I'm thinking instead of basing everything off of season stats teams should select specific games and only look at the stats from those games, how these games are selected could differ from team to team. Some teams might weigh playoff games and end of the season game higher, some teams would focus on games in the division or against them. Also generating individual ratings from those games would give us flexibility in scouting. Say a player has a really up and down year, half his games he rated 5 points below where he should be and the other half he rated 5 above where he should be. If we simply do a total he comes out average, but some teams might be focused more on a players upside than on their downside and so would weigh the above average games higher than the below average games. A team looking for a reliable player might do the opposite. Doing it this way permits us a lot of ways to have teams scout differently.


Return to “Tuesday Morning Manager”

Who is online

Users browsing this forum: No registered users and 2 guests