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.