Player Impact Rating (PIR)

Most of the threads on this board are dealing with betting systems but I have been thinking over the past week how it could be interesting to find,and express numerically, the impact a player while he is in the game. This could be a useful stat, especially for betting, as it could show how much the loss of a player from the lineup would affect the team. This stat could show how useful and influential a certain player is while he is playing. This is my first "draft" of the formula and I would like some suggestions and feedback from others on this board on how to edit and adjust the formula to make it more accurate. I am posting here because I think it could be an interesting idea to delve deeper into the statistics of players.
Method and Formula
The way I went at finding the PIR was to look at the player stats, as well as, and compared to, the stats of the team as a whole while the certain player is in the game. The complete formula is (20)*(G/GF)+(20)*(A/GF)+(0.4)*GD-(0.2)*R-(OG/GA). G=Goals scored (excluding penalty kick goals), GF=Goals For, A=Assists, GD=Goal Differencial, R=Red Cards, OG=Own goals, GA=Goals Against.
I tested this formula on players from Major League Soccer using stats from the 2002 season. The results seemed to be around what I expected, but the formula could be fixed to be more accurate. I could post my results if anyone wants to see them. I would really appreciate any feedback and suggestions.

Sounds like a hellofalota work but I for one will be very interested in seing what such a beast would look like.

I think I'll need a few examples, though. That formula looks like ancient Egyptian to me...

[Joe]

I think Paul Steele's book has stats on this, i.e. % of wins with/without player.

Hi soccerfan

I'd be interested in seeing/discussing this.

You can email me at

tmmneural AT yahoo DOT co DOT uk

M.

Hi SF,
I tried something similar a season back.

Check out the RacingPost's Smartbet pages,they've got an interactive database that allows you to examine results for teams with & without players or combinations of players.It's presentation of the results does leave a lot to be desired though.

I didn't think that taking a straight win/loss percentage told you much as it doesn't say anything about the strength of the opposition.So I looked at how much better or worse the team did compared to the mid point of the supremacy spread for that particular game.

I tended to find that the most influential players play in the most obvious positions.Centre forward/central defence/goalkeeper.

I ended up with figures along the lines of Man U playing around 0.2 of a goal per game better than the spread firms had estimated when RVN played compared to when he didn't.You can then convert this extra goal supremacy to an extra likelyhood of winning the game.

Equally interesting is finding out which big named(over hyped) players cause a team's form to dip when they're on the park. A couple of French World cup winners spring to mind.

W.

If anyone is interested, I could email you the Excel I used spreadsheet so you could have a better idea of the formula.