PED Differential: BYU vs Utah 2009

[pelagius]

Let me try this again. There are a limited number of "official" statistical measures you will find tracked by the NCAA (rushing/passing yards, turnovers, penalties, etc. etc. etc.). The point I'm trying to make is that pass efficiency (one of the official measures) has the greatest correlation with MOV of all the official measures you want to stack up against it. Of all the official statistics:

• Pass efficiency has the highest correlation with average points scored.
  
• Pass efficiency differential has the highest correlation with average MOV.
  
• Pass efficiency differential-differential has the highest correlation with predicted MOV.

This does not deny the existence of models that utilize entirely new, unofficial statistics (that go beyond simple differencing), such as a new pass efficiency formula that has different weightings or some model that is a massive composite of all of the official statistics and so forth.

Indy,

Jay wasn't really challenging the previous claim. He is saying there is more information in past MOV than past PED differential in terms of predicting future outcome of games (either W/L or MOV) . In its most sophisticated form he is suggesting that, for example, the Sagarin predictor ratings are better predictors of future outcomes than PED differential. MOV is clearly an official statistic and that is all a computer model like Sagarin uses (besides taken into account home field). Jay is suggesting this will be a better predictor. He may be wrong but it is not like it is an unreasonable question/response. In fact, I am a little surprised you haven't tested this before.

[Indy Coug]

Indy,

Jay wasn't really challenging the previous claim. He is saying there is more information in past MOV than past PED differential in terms of predicting future outcome of games (either W/L or MOV) . In its most sophisticated form he is suggesting that, for example, the Sagarin predictor ratings are better predictors of future outcomes than PED differential. MOV is clearly an official statistic and that is all a computer model like Sagarin uses (besides taken into account home field). Jay is suggesting this will be a better predictor. He may be wrong but it is not like it is an unreasonable question/response. In fact, I am a little surprised you haven't tested this before.

But MOV is the end result of many things. What one "mainstream" (and I know I'm using that term a little loosely here) measure is the single biggest contributor towards MOV?

The concept of PED Differential is already pretty technical for the average person. If I'm trying to communicate how important the pass efficiency statistic is (which frankly has only become recently accepted by mainstream media) compared to some of the pet statistics that get quoted in some kind of correlative context (eg. BYU is 27-2 when running for more than 200 yards), it's best not trying to gum up the works by making it even more technical.

I've taken the PED Differential concept even farther than what I'm describing here and using that model against Vegas spreads, if my expected spread is more than a certain absolute difference of the Vegas spread, then my model will beat the spread over 65% of the time. There's where the predictive value kicks in.

[jay santos]

But MOV is the end result of many things. What one "mainstream" (and I know I'm using that term a little loosely here) measure is the single biggest contributor towards MOV?

The concept of PED Differential is already pretty technical for the average person. If I'm trying to communicate how important the pass efficiency statistic is (which frankly has only become recently accepted by mainstream media) compared to some of the pet statistics that get quoted in some kind of correlative context (eg. BYU is 27-2 when running for more than 200 yards), it's best not trying to gum up the works by making it even more technical.

I've taken the PED Differential concept even farther than what I'm describing here and using that model against Vegas spreads, if my expected spread is more than a certain absolute difference of the Vegas spread, then my model will beat the spread over 65% of the time. There's where the predictive value kicks in.

1. I think the PED concept is maybe good in illustrating the importance of the pass relative to run. i.e. if you can prove PED correlates to MOV and W-L more than say average run per carry differential. But that's where the value ends. I find it extremely odd you would use it to predict a score.

2. Do you realize what you're saying when you imply you can beat Vegas 65% of the time? Even saying you can beat Vegas 55% of the time is like saying you have a license to print money.

[Indy Coug]

2. Do you realize what you're saying when you imply you can beat Vegas 65% of the time? Even saying you can beat Vegas 55% of the time is like saying you have a license to print money.

Yes, I'm familiar with the concept of arbitrage.

[jay santos]

Yes, I'm familiar with the concept of arbitrage.

Don't mess with arbitrage. If you can win 65% of the time, bet on every game every week.

[jay santos]

Indy, see this thread

http://cougaruteforum.com/showthread.php?t=3559

How is what you're doing with PED any different than what Locke did with his offensive defensive differential stat?

[Indy Coug]

Indy, see this thread

http://cougaruteforum.com/showthread.php?t=3559

How is what you're doing with PED any different than what Locke did with his offensive defensive differential stat?

You mean what I'm presenting in this thread as PED versus what Locke is doing? They're fairly analogous I guess, but I don't know much about basketball efficiency ratings and haven't done any research to see how good they are.

[jay santos]

You mean what I'm presenting in this thread as PED versus what Locke is doing? They're fairly analogous I guess, but I don't know much about basketball efficiency ratings and haven't done any research to see how good they are.

What I'm saying is that I think what both your models are doing is to pick a winner on a variable highly correlated to MOV with no control for SOS.

I'd bet your model would predict goofy results when you take a team with a really tough SOS and match up with a weak SOS. Take Boise vs Florida for example.

[Indy Coug]

What I'm saying is that I think what both your models are doing is to pick a winner on a variable highly correlated to MOV with no control for SOS.

I'd bet your model would predict goofy results when you take a team with a really tough SOS and match up with a weak SOS. Take Boise vs Florida for example.

1. I've shown how good even an unadjusted PEDD is at predicting MOV.

2. The "ultimate" model is an adjusted statistic, and that is what I used to compare against the Vegas spread, but I'm not sharing that model here.


Maybe you will find this home/away conference/non-conference PEDD post interesting:

http://www.cougarboard.com/noframes/...tml?id=2302955

[Indy Coug]

Offense Pass Efficiency

BYU - 110.20
Utah - 152.4

Defense Pass Efficiency

BYU - 121.31
Utah - 137.38


Pass Efficiency Differential

BYU: -11.11
Utah: 15.02


PED Differential: 26.13


Factor that in with home field advantage and Utah should be around a 12 point favorite.

[Indy Coug]

Here are the caveats:

1. Utah's pass efficiency defense the last 4 games: 186.6. Memphis is currently last nationally in season PED with 170.7. http://twitter.com/#!/CougarStats/st...18275275051008

2. For the first time in BYU history, BYU scored at least 144 points over three straight games AND allowed no more than 24 (55-7, 49-10, 40-7) http://twitter.com/#!/CougarStats/st...74973258076162

3. Heaps' efficiency the last 4 games is 159.3 versus 86.9 prior to that.

4. Since Bronco took over as DC, BYU has allowed 71.5 yards rushing/game and an average of 2.68 per carry.


In other words, the 12 point spread indicated by the PEDD statistic may not be as valid due to significant recent shifts in performance by both teams.

[Moliere]

Offense Pass Efficiency

BYU - 110.20
Utah - 152.4

Defense Pass Efficiency

BYU - 121.31
Utah - 137.38


Pass Efficiency Differential

BYU: -11.11
Utah: 15.02


PED Differential: 26.13


Factor that in with home field advantage and Utah should be around a 12 point favorite.

Once Indy speaks, the thinking is done. I'm preparing myself for a two touchdown beat down. Whit is going to confuse the heck out of Heaps. Hopefully Heaps and Ashworth did their home teaching this month.

[Indy Coug]

Once Indy speaks, the thinking is done. I'm preparing myself for a two touchdown beat down. Whit is going to confuse the heck out of Heaps. Hopefully Heaps and Ashworth did their home teaching this month.

Go back and read my caveats.

[BoylenOver]

Here are the caveats:

1. Utah's pass efficiency defense the last 4 games: 186.6. Memphis is currently last nationally in season PED with 170.7. http://twitter.com/#!/CougarStats/st...18275275051008

2. For the first time in BYU history, BYU scored at least 144 points over three straight games AND allowed no more than 24 (55-7, 49-10, 40-7) http://twitter.com/#!/CougarStats/st...74973258076162

3. Heaps' efficiency the last 4 games is 159.3 versus 86.9 prior to that.

4. Since Bronco took over as DC, BYU has allowed 71.5 yards rushing/game and an average of 2.68 per carry.


In other words, the 12 point spread indicated by the PEDD statistic may not be as valid due to significant recent shifts in performance by both teams.

Can we chalk up the recent shifts to the performance of Utah/BYU, or can we chalk it up to the level of competition? IMO it's a combination of both, but I zealously choose to believe it's more the competition.

[Moliere]

Go back and read my caveats.

I did. I was making a joke followed by my professional skeptical analysis. Lighten up Francis!

Whit's defense always seems to screw with BYU QBs. I'm not liking the idea of our freshman QB going up against Whit's defense. I hope we can pound the ball down Utah's throats ala SDSU. Given their average LB corp I think this is the best way to beat them. I'll be as happy as a Ute fan in the topless section if Heaps passes for 400 yards en route to a blowout win but that is not likely.