Sharpe Ratio - Part II
If you haven't read the previous article about what the Sharpe ratio is and how to compute it I would suggest reading that first.
Applications for the Sharpe Ratio
Earlier we looked at one very general circumstance for calculating the Sharpe ratio for a play over the course of the season. This can be helpful but also might be too generic when building your gameplan. I'll look at three circumstances. One, determining your team's strengths by calculating the Sharpe ratio every time you ran a given formation/set; two, comparing the success (Sharpe ratios) when breaking down your opponents game film when the opposing teams ran inside, left, right, etc (depending how precisely you'd like to break it down. If you wanted to break it down by hole or gap); and, third, determining a given play's success on 3rd down and a set distance--how often it results in a 1st down.
By Formation or Personnel Grouping
This could be accomplished in the same way as the earlier calculation. You would look at your compiled statistics (remember, you can use anything that you feel is relevant. Game statistics are best but there are fewer of them and it may not apply to a given opponent. You can use practice stats, recorded info off gamefilm, past seasons, etc) and simply look at yardage gained or lost (you can use a similar convention as earlier for turnovers, such as -15 or -20).
If you want to make sure you're only calculating relevant stats, you may want to eliminate plays that did not happen in the middle of the field, such as in the red zone or when backed up, or also eliminate short yardage plays or only focus on 1st and 2nd down.
As before pick a benchmark. We used two yards earlier for the QB sneak. Similar logic can apply to passes, you could always pick the sneak or even throw a bubble screen/quick raise pass rather than a 3 or 5 step pass. For our purposes we use 2 yards again to remain conservative.
Shown below are a bit of the results. The Sharpe ratio here is X. We could then repeat this analysis for our other main personnel groups or formation packages. Below is a fictitious example of this.
Thus, we would say that we are more successful when in our Ace package than when in our Spread package, and we are not very successful when in our Pro set package. The reasons for this can vary, such as what players or in the game or maybe we are simply making poor play calls when in that formation.
Regardless, this can be enlightening because it could show that we are underutilizing a particular strategy (personnel grouping) and we could be more successful by using it more.
(Note - Skip if this is confusing:
There is an interesting application of the Sharpe ratio for what are termed "uncorrelated strategies." Basically, it says that if you have two uncorrelated strategies, and you want to know how much to use one versus the other, you can use the ratio of their Sharpe ratios. For example if the Sharpe Ratio of strategy A was 2 and the ratio of B was 1, you would want to use A twice as much as B. In the context of personnel if it was personnel group A and B with Sharpe ratios of 2 and 1, you would want to use personnel group A 66.67% of the time, all else equal, if this relationship holds true. More food for thought than anything else.)
This analysis applies to all sorts of self-evaluation. Furthermore, we can do the exact same thing for defense. 4-4-3 lineup vs 4-3-4, 4 down linemen versus 3, etc. Any comparison you can think of. In the case of defense you would want to find the lowest Sharpe ratio; in other words, the worst ratio of offensive success to risk.
Game Film - Opponents' Success When Running to Different Sides
This is pretty straightforward if you've followed through the other discussions. I mostly wanted to apply it in a very general sense, not a specific play, and use it for evaluating the defense. One would compile the statistics on the rushing yards against by your opponents' every time they ran left, right, or inside, for simplicity's sake.
As before the computation can be done in excel with columns, with screen captures shown below. If we noticed that the Sharpe for running to the right was higher, what would this tell us?
As always, the Sharpe ratio is an indicator telling us where to look, but not necessarily giving us the solutions--this is what being a coach is all about. Here, if the Sharpe was higher for the right side we might study the gamefilm sooner and see that they are weak at defensive end. Or maybe even that they have a solid player, but he is way too aggressive and can be taken advantage of. These tendencies can be deduced from watching film, but analysis of the data told us precisely where to look and, in the end, should help us save time. Then we decide if the appropriate response are crack blocks, traps, options, etc.
Applications for the Sharpe Ratio
Earlier we looked at one very general circumstance for calculating the Sharpe ratio for a play over the course of the season. This can be helpful but also might be too generic when building your gameplan. I'll look at three circumstances. One, determining your team's strengths by calculating the Sharpe ratio every time you ran a given formation/set; two, comparing the success (Sharpe ratios) when breaking down your opponents game film when the opposing teams ran inside, left, right, etc (depending how precisely you'd like to break it down. If you wanted to break it down by hole or gap); and, third, determining a given play's success on 3rd down and a set distance--how often it results in a 1st down.
By Formation or Personnel Grouping
This could be accomplished in the same way as the earlier calculation. You would look at your compiled statistics (remember, you can use anything that you feel is relevant. Game statistics are best but there are fewer of them and it may not apply to a given opponent. You can use practice stats, recorded info off gamefilm, past seasons, etc) and simply look at yardage gained or lost (you can use a similar convention as earlier for turnovers, such as -15 or -20).
If you want to make sure you're only calculating relevant stats, you may want to eliminate plays that did not happen in the middle of the field, such as in the red zone or when backed up, or also eliminate short yardage plays or only focus on 1st and 2nd down.
As before pick a benchmark. We used two yards earlier for the QB sneak. Similar logic can apply to passes, you could always pick the sneak or even throw a bubble screen/quick raise pass rather than a 3 or 5 step pass. For our purposes we use 2 yards again to remain conservative.
Shown below are a bit of the results. The Sharpe ratio here is X. We could then repeat this analysis for our other main personnel groups or formation packages. Below is a fictitious example of this.
Thus, we would say that we are more successful when in our Ace package than when in our Spread package, and we are not very successful when in our Pro set package. The reasons for this can vary, such as what players or in the game or maybe we are simply making poor play calls when in that formation.
Regardless, this can be enlightening because it could show that we are underutilizing a particular strategy (personnel grouping) and we could be more successful by using it more.
(Note - Skip if this is confusing:
There is an interesting application of the Sharpe ratio for what are termed "uncorrelated strategies." Basically, it says that if you have two uncorrelated strategies, and you want to know how much to use one versus the other, you can use the ratio of their Sharpe ratios. For example if the Sharpe Ratio of strategy A was 2 and the ratio of B was 1, you would want to use A twice as much as B. In the context of personnel if it was personnel group A and B with Sharpe ratios of 2 and 1, you would want to use personnel group A 66.67% of the time, all else equal, if this relationship holds true. More food for thought than anything else.)
This analysis applies to all sorts of self-evaluation. Furthermore, we can do the exact same thing for defense. 4-4-3 lineup vs 4-3-4, 4 down linemen versus 3, etc. Any comparison you can think of. In the case of defense you would want to find the lowest Sharpe ratio; in other words, the worst ratio of offensive success to risk.
Game Film - Opponents' Success When Running to Different Sides
This is pretty straightforward if you've followed through the other discussions. I mostly wanted to apply it in a very general sense, not a specific play, and use it for evaluating the defense. One would compile the statistics on the rushing yards against by your opponents' every time they ran left, right, or inside, for simplicity's sake.
As before the computation can be done in excel with columns, with screen captures shown below. If we noticed that the Sharpe for running to the right was higher, what would this tell us?
As always, the Sharpe ratio is an indicator telling us where to look, but not necessarily giving us the solutions--this is what being a coach is all about. Here, if the Sharpe was higher for the right side we might study the gamefilm sooner and see that they are weak at defensive end. Or maybe even that they have a solid player, but he is way too aggressive and can be taken advantage of. These tendencies can be deduced from watching film, but analysis of the data told us precisely where to look and, in the end, should help us save time. Then we decide if the appropriate response are crack blocks, traps, options, etc.
posted by Chris at
5/18/2005 09:11:00 PM
0 Comments:
Post a Comment
<< Home