What Should Our Defensive PER Golden Standard Be?

Although PER is imperfect because it does not measure everything about a QB performance, I think it is a very useful tool. It does not measure QB rushing, or subtlties like how effectively the pass or deception on hand offs, or QB blocking helps the running back performance. But is still very useful. Spiff has provided a wealth of data establishing an offensive PER standard of 135 as an average BYU QB performance and a target that almost ensures a BYU win and creates a winning percentage that keeps BYU ranked or at least receiving plenty of votes. Dual threat QBs as good as Hill may change that target lower. You may not need a 135 PER to ensure a high percentage of wins if your QB chalks up 125 rushing yards and makes the backs thusly more effective as well since defenses have to worry about both.



We have no standard for a defensive PER. What target should we expect the BYU defense to hold opponents to on average to ensure that opponents rarely beat us?



Please post your rationale and hopefully, Spiff or others have some data. Re-voting is enabled in case you change your mind after data comes in.



[ Post made via Mobile Device ] http://cougarcorner.com/images/mobile.png" alt="" loading="lazy" style="max-width:100%">

We held 98 UCONN, 116 Texas, and 127 Houston so far in 2014.



UCONN never really threatened us. I do not think many teams would threaten us with a PER under 100. Texas made it interesting for the first half - then BAM. I would guess that we would win a good percentage of games by holding teams to 115. Maybe not the 90% we would like to win, though. Houston gave us a scare. If Houston could have run the ball they would have beaten us with that 127 PER that day.



I chose under 110 until I see more data.



[ Post made via Mobile Device ] http://cougarcorner.com/images/mobile.png" alt="" loading="lazy" style="max-width:100%">

Without doing any research I think 115 is a reasonable baseline guess. I'll do some research this weekend.

100 would be great, and right now, on the season going into UVA, we're at 113.86. As of the Hine kickoff return for a TD in the fourth, we've given up 106.5. Overall at that point, we're at 111.99.

SpiffCoug wrote:
Without doing any research I think 115 is a reasonable baseline guess. I'll do some research this weekend.

Yikes. How did we do today?

Virginia 109.94

SpiffCoug wrote:
Without doing any research I think 115 is a reasonable baseline guess. I'll do some research this weekend.

This year, we seem to win regardless what PER we achieve or we allow, but Hill is anomalous because he runs so well. Even then, I am not sure we could have allowed Virginia 125 and survived even though we achieved over 150. One more turnover by us, or one less by them, would likely have reversed the final outcome.



any luck on historical research yet?



[ Post made via Mobile Device ] http://cougarcorner.com/images/mobile.png" alt="" loading="lazy" style="max-width:100%">

I like Bronco's standard--hold them to fewer than 23 points.



How that is accomplished between DPER and Rush Defense and Turnovers, I could care less.

snoscythe wrote:
I like Bronco's standard--hold them to fewer than 23 points.



How that is accomplished between DPER and Rush Defense and Turnovers, I could care less.

This year, we are 4-0 even though we have failed that target half the time. Fortunate.



Other years, we could have hit that target 100% but lost too many games. It would be interesting to see if that target has gotten the desired win % over time.



I think data about all of the components is interesting as it adds objectivity to constructive criticism of what needs improvement. After Further Review has also been adding clarity to what and why.



By any of these measures, hopefully it is clear that the much maligned offense, is back this year and is carrying the much heralded defense this year.



AFR for the Houston game was interesting. Hail Mary was largely on the new DC. Should have had Dime instead of Nickel and should gave had defenders prepared to get in front not behind and should have had someone tall like Mitch Matthews behind to knock it straight down hard.



Broken assignment on zone by Harvey Jackson. Badly broken. Left receiver wide open. Our lock down corners turning their heads at the wrong time and drawing legitimate penalties - not bad calls. UVA game. That according to news quote by DC. Defensive line is allowing QB too much time for any DBs to cover receivers that long thus opening up the underneath as receivers adjust routes. PoVey PI due to missed assignment by another player VA VA.



[ Post made via Mobile Device ] http://cougarcorner.com/images/mobile.png" alt="" loading="lazy" style="max-width:100%">

I did some research and crunched some numbers. If you go back to 1994, the average PER for a BYU QB is 135.58. This is part of the reason why 135 is the PER standard.



There are total of 257 games since 1994, so the sample size is fairly substantial. BYU has met the standard 133 times winning, 121 of them (.910)



Since 1994, the average opponent's PER is 118.99. If we apply the same standard, and make the average the standard of what any BYU defense should do, here is what we get.



BYU has held opponents to a 119 PER 140 times. BYU has won 114 of those games (.814). To me, it at first appears that that's not good enough. As .814 is a considerably lower winning percentage than the BYU 135 PER standard yielding a .910 winning percentage.



What if we lower the Defensive PER standard to 115? Now we get a total of 128 instances. BYU has won 103 of those games. Ironically and interestingly enough it yields an even LOWER winning percentage of .804. So 115, isn't good enough either.



What about a 110 Defensive PER? Well, now we have 113 such games to look at and BYU is 94-19 (.832) in those games. Better, but I think I want something above .850.



If we drop the Defensive PER to 105 our results now look like 83 wins in 99 games (.838). If we go even lower to 100, we get a record of 68-15 (.819). And now we're well under 100 such games and our sample size is reduced even further.



I'm not sure standard can be one that is only met less than 1/3 of the time. Whereas the Offense PER standard was met in more than 1/2 the games.



What this tells me is that OFFENSIVE PER is more important that DEFENSIVE PER. If you want to see a similar winning percentage, the Defensive PER has to drop all the way to 85 where BYU is 37-5 (.880). But even then we're still three percentage points lower than the Offensive PER and it would become a standard that has been so rarely achieved - just 16.3% of games.



So I think my final decision will bring us full circle back to the 119 PER and I'll settle for the .814 winning percentage. And it's also a standard that has been achieved about half the time.



So there you have it Cougar Corner-ites! Here are the PER "Royal Blue" Standards:

BYU Offense: 135

BYU Defense: 119

It is interesting how that worm turns. Especially in that segment where holding to a lower PER actually got a worse winning percentage. And no matter how good the defense performs, unless you get a lot of defensive scores, you need the offense to put up some points to win.



This raises the question of how offensive and defensive PER relate to each other. Maybe buried in all of this muddle is an even more golden predictor of win %.



Can you have your model chug what Offensive VA Defensive Per Differential assures an even higher win % than the Offensive PER model? Just curious if there is some differential like 15 or 20 points that just about assures a win.



[ Post made via Mobile Device ] http://cougarcorner.com/images/mobile.png" alt="" loading="lazy" style="max-width:100%">

BYU's record when their PER is X points better than the opponent:



100+ Points: 33-0 (1.000)

90-100 Points: 11-0 (1.000)

80-90 Points: 22-0 (1.000)

70-80 Points: 20-0 (1.000)

60-70 Points: 18-1 (.947)

50-60 Points: 30-4 (.882)

40-50 Points: 30-3 (.909)

30-40 Points: 25-3 (.893)

20-30 Points: 23-5 (.821)

10-20 Points: 19-6-1 (.615)

0-10 Points: 16-5 (.762)



The 50+ PER point losses were UCLA 2007 (1st game), Arizona 2006, Stanford 2003, Utah St 1993 (when Walsh's PER was 207.95!!) and Ohio St 1993.



BYU Opponent's record when their PER is X points better than BYU:



100+ Points: 11-1 (.917)

90-100 Points: 2-0 (1.000)

80-90 Points: 6-0 (1.000)

70-80 Points: 4-1 (.800)

60-70 Points: 8-3 (.727)

50-60 Points: 8-1-1 (.800)

40-50 Points: 11-1 (.917)

30-40 Points: 17-2 (.895)

20-30 Points: 5-1 (.833)

10-20 Points: 12-7 (.632)

0-10 Points: 10-13 (.435)



In 1984, Wyoming's PER was 271.51 and BYU's was a "paltry" 169.03. They were 11 of 14 for 243 yards and 3 TDs with 0 INTs. Kevin Lowe, a WYO RB, was 1-1 for 64 1 TD for nice 967.60 PER. Simply an amazing game.



You can watch the game here:

http://www.byutvsports.com/filmroom/watch/?id=1f39b86f-5973-4d13-8cbf-e5823b75af55" onclick="window.open(this.href);return false;



Based on the above, it looks like once a team is 40+ PER points above their opponent, they're going to win 90+% of the time.

What about when out O has 135 and the D is below 119? What is the win record when we get both?

scott715 wrote:
What about when out O has 135 and the D is below 119? What is the win record when we get both?

135-5 (.964).



The losses were Arizona 2006, Washington 1996, Ohio St 1993, San Diego St 1992, UCLA 1985.