2008-2009 Shocker Men's Prouty Ratings

[im4wsu]

Just got back from Alamo Bowl, so am tardy in updating the +/- vs Prouty ratings and evaluation of line-up combinations.



Look at Hatch in Creighton game. In 7 minutes, all statistic line items are zero except for missed shots leading to a very low Prouty. WSU scored 5 more points than it allowed when Hatch was on the floor.

Conversely, Hannah had very good stat line (Prouty .499), but in his 31 minutes, the Shockers gave 16 more points than they scored.

[1979Shocker]

1979 - Murry has no 'a' in his name .... The spelling police

Thanks. It's fixed everywhere.

[1979Shocker]

Wichita State vs Bradley Prouty Ratings

              BU       All 
Player       Game    Games 
Hannah       0.213   0.387
Clemente     0.376   0.361
Murry        0.372   0.342
Hawkins      0.337   0.334
Stutz*       0.377   0.309
Durley       0.381   0.288
Griskenas    0.012   0.284
Chamberlain -0.030   0.266
Kyles        0.056   0.234
Hatch*       0.561   0.230
Michael      0.000   0.227
Ellis        0.054   0.172
Britton      0.000   0.063
Steven       0.000   0.001
TEAM         0.360   0.371

Team Winning Percentage: 0.462

* Improved from previous game.

[ShockTalk]

I need a refresher course: How does Clemente, Murry, Hawkins, and Durley have a higher Prouty Rating for the Bradley game than their overall rating, yet have an arrow pointing downward (which I assume means their overall rating went down).

[1979Shocker]

I need a refresher course: How does Clemente, Murry, Hawkins, and Durley have a higher Prouty Rating for the Bradley game than their overall rating, yet have an arrow pointing downward (which I assume means their overall rating went down).

There are four parts to the Prouty Rating formula, Offensive Efficiency, Total Offense, Possessions Gained, and Win Rating.

For Clemente, his Offensive Efficiency and Total Offense were better in the Bradley game than his overall. His Possessions Gained and Win Rating were worse. Prior to the Bradley game, his Total Rating was 0.367.

For the Bradley games Clemente's ratings were:

Offensive Efficiency 0.600
Total Offense        0.481
Possessions Gained   0.111
Win Rating           0.312
Total Rating         0.376

For all games Clemente's ratings are:

Offensive Efficiency 0.519
Total Offense        0.319
Possessions Gained   0.283
Win Rating           0.323
Total Rating         0.361

[im4wsu]

I need a refresher course: How does Clemente, Murry, Hawkins, and Durley have a higher Prouty Rating for the Bradley game than their overall rating, yet have an arrow pointing downward (which I assume means their overall rating went down).

Normally with "ratings" or "averages", being below that number in the additional information would lower the "rating" or "average" or being higher than that level would increase the "rating" or "average". This is not always the case with Prouty rating because the team winning percentage applies to both past and current data. Even if all statistics for Clemente for just the Bradley were EXACTLY equal to the numbers that comprised the "rating" coming into the Bradley game (therefore one would expect the r"rating" to remain unchanged), the "rating" would go down after including the Bradley game stats because the team winning percentage went down due to the loss.
Thus, the team losing the Bradley somehow changed the rating for all the games played before it. IMHO, that is an inherent flaw in the Prouty rating formula.

[ShockTalk]

Got It! Thanks im4wsu. A player could actually be playing better each game, but if the team is losing the games, his rating could go down and visa versa.

[im4wsu]

Updated to include Bradley game:



Hatch had good shooting percentage, hence a good Prouty of .561, but WSU was outscored by 2 in the 8 minutes of playing time for Hatch, a +/- per 40 of -9.7.

Kyles didn't shoot so had a nonexistent Prouty of .056, but in his 8 minutes, WSU outscored Bradley 12-7, converting to a 25.3 +/- score.

[1979Shocker]

I've been looking over the Win Rating part of the Prouty Rating. I was trying to figure out if I've been calculating it right for all games.

The formula: [ MIN / (Team MIN / 5) * Team Win Percentage ] is for individual games. The [Team MIN /5] will always be 40 when there's no overtime. So the Win Rating would be the MIN / 40 * Team Win Percentage.

For all games, I took the total Team MIN / Games Played / 5 which will still come out to 40 with no overtime games.

The part I'm not sure about is the player MIN. Currently I'm dividing the player MIN played by Games they played in. As an example, a player that played 40 minutes in 1 game out of 10, assuming a 5-5 record, would have a Win Rating of

(40 / 1 game) / 40 * 0.500 = 40 / 40 * 0.500 = 0.500. This would be their Win Rating when they actually played.

If you used the average MIN instead, the calculation would be:

(40 / 10 games) / 40 * 0.500 = 4 /40 * 0.500 = 0.050. This would be their Win Rating whether they played or not.

So I don't know if a player's Win Rating should be for only games they played in or for all games. This won't affect anyone who has played in every game. If I calculate the formula using all games, a player who hasn't played in all games will see their overall Win Rating go down along with their Prouty Rating.

Does anyone have any thoughts on this?

I'm thinking of using the average minutes. The only question would be, did the player not play because they weren't producing? The average MIN would work better for this situation. Or, did the player not play because they were injured. The [ MIN / Games Played ] might work better for this. This wouldn't penalized the player's Win Rating because of the injury.

[im4wsu]

I've been looking over the Win Rating part of the Prouty Rating. I was trying to figure out if I've been calculating it right for all games.

The formula: [ MIN / (Team MIN / 5) * Team Win Percentage ] is for individual games. The [Team MIN /5] will always be 40 when there's no overtime. So the Win Rating would be the MIN / 40 * Team Win Percentage.

For all games, I took the total Team MIN / Games Played / 5 which will still come out to 40 with no overtime games.

The part I'm not sure about is the player MIN. Currently I'm dividing the player MIN played by Games they played in. As an example, a player that played 40 minutes in 1 game out of 10, assuming a 5-5 record, would have a Win Rating of

(40 / 1 game) / 40 * 0.500 = 40 / 40 * 0.500 = 0.500. This would be their Win Rating when they actually played.

If you used the average MIN instead, the calculation would be:

(40 / 10 games) / 40 * 0.500 = 4 /40 * 0.500 = 0.050. This would be their Win Rating whether they played or not.

So I don't know if a player's Win Rating should be for only games they played in or for all games. This won't affect anyone who has played in every game. If I calculate the formula using all games, a player who hasn't played in all games will see their overall Win Rating go down along with their Prouty Rating.

Does anyone have any thoughts on this?

I'm thinking of using the average minutes. The only question would be, did the player not play because they weren't producing? The average MIN would work better for this situation. Or, did the player not play because they were injured. The [ MIN / Games Played ] might work better for this. This wouldn't penalized the player's Win Rating because of the injury.

79--I think you read the formula literally and use total minutes for the individual for the year and divide by [the total minutes for the team for the year divided by 5]. As you said, the denominator will be 200 unless there are overtime games.

This is the best I can find using google:

http://www.eba-stats.com/glossary/questions/prouty_basketball_efficiency_rating.htm

[1979Shocker]

79--I think you read the formula literally and use total minutes for the individual for the year and divide by [the total minutes for the team for the year divided by 5]. As you said, the denominator will be 200 unless there are overtime games.

That's what I was thinking. I'll have to look at some of the previous game ratings to see player ratings changed.

[rayc]

1979Shocker-

From a strictly mathematical perspective, your current method is probably the correct one. When a player doesn’t play in a game his rating for that game is not zero, it is undefined (one of those nasty division by zero things).

I think to get to the kind of seasonal rating you want, you will probably need to define one for yourself as I haven’t seen your question addressed elsewhere. You may want to consider weighting the seasonal figure that you currently calculate by the percentage of total games played in. By defining it yourself, you gain some leeway in defining a factor for games not played due to injury.

Example: a player has a current season ratings of 0.30. By weighting it, the seasonal rating would be say 0.3*(3/5) = 0.18 (if games played in =3 of 5), or 0.3*(5/5) = 0.3. If a player missed the two games due to injury, you can adjust your weighting factor to (5/5) versus the (3/5) if he just didn’t see any time. This way will give you the same results as your using average minutes without accounting for injuries, but allowing you the ability to adjust for injuries if you desire. A rather simplistic addition, but it’s a thought.

1979Shocker Player Season Prouty Rating = (Rating as you currently do it)*(games played in + games missed due to injury)/ Total Games

[1979Shocker]

Wichita State vs Northern Iowa Prouty Ratings

              UNI     All 
Player       Game    Games 
Hannah       0.362   0.380
Clemente     0.167   0.346
Murry        0.281   0.333
Hawkins      0.232   0.323
Stutz        0.190   0.298
Griskenas*   0.334   0.294
Durley       0.202   0.279
Chamberlain* 0.324   0.268
Hatch*       0.349   0.238 
Michael      0.000   0.227 
Kyles       -0.017   0.224 
Ellis        0.178   0.170 
Britton      0.005   0.043 
Steven       0.005   0.001 
TEAM         0.313   0.359

Team Winning Percentage: 0.429

* Improved from previous game.

[im4wsu]

On this update through the UNI game, I have highlighted the performances that are better than the team's average performance.

[ShockerInKC]

I think to get to the kind of seasonal rating you want, you will probably need to define one for yourself as I haven’t seen your question addressed elsewhere. You may want to consider weighting the seasonal figure that you currently calculate by the percentage of total games played in. By defining it yourself, you gain some leeway in defining a factor for games not played due to injury.

I understand that there's a flaw in the formula that penalizes a player when they don't play in a loss, and conversely rewards them for wins.

Another way that might give you a good seasonal rating would be to take each of the individual game Prouty ratings, multiply them by minutes played in that game, then sum them, and divide back through by minutes played in all games. This way there is no reward or penalty for games missed. I haven't been back through the formulas (I'll look later) and see if this method rewards or penalizes too much based on minutes played alone. This is an interesting endeavor.

I also wonder about the effect on sitting down a selfish player (who might have a good Prouty rating) but has an overall negative effect on the team's performance. This could explain why the team combination with Hatch in the game is better than the standard starting lineup with Durley (instead of Stutz).