Stat Geeks

I know there a few here that like to comb through statistical analysis and that there are opinions on which ones are most important when it comes to wins and losses.



This is based on the NFL but should be applicable to most any level. HOw about TOXIC DIFFERENTIAL.

It essentially combines turnover differential with explosive play differential (defined as plays gaining more than 20 yards).



It has long been a proven adage that if you have a plus-2 turnover advantage on your opponent you will win the game 80 percent of the time. If you have a minus-2 explosive-play advantage it equally leads to winning about 80 percent of the time. If you have both you win close to 95 percent of the time. No other statistical combination that you can control has quite the same rate of predictive success.

[/b]



http://msn.foxsports.com/nfl/story/toxic-differential-little-known-stat-key-to-predicting-success-wednesdays-with-billick-112812" onclick="window.open(this.href);return false;

MTM wrote:
I know there a few here that like to comb through statistical analysis and that there are opinions on which ones are most important when it comes to wins and losses.



This is based on the NFL but should be applicable to most any level. HOw about TOXIC DIFFERENTIAL.

It essentially combines turnover differential with explosive play differential (defined as plays gaining more than 20 yards).



It has long been a proven adage that if you have a plus-2 turnover advantage on your opponent you will win the game 80 percent of the time. If you have a minus-2 explosive-play advantage it equally leads to winning about 80 percent of the time. If you have both you win close to 95 percent of the time. No other statistical combination that you can control has quite the same rate of predictive success.

[/b]

What if I can almost get to 95% with one stat and not a combination.

Well, by itself PER Differential is already at 86.1%



Here is BYU since 2006 with 135 PER:

51 games with a 135 PER (Record 48-3) (.941)

This explains a lot of Riley's limitations. He rarely has big play ability and we now how turnover prone he is. His toughness and scrambling make up for some really deficits. Take away his scrambling with injuries...............ah forget it, broken record.

SpiffCoug wrote:

I know there a few here that like to comb through statistical analysis and that there are opinions on which ones are most important when it comes to wins and losses.



This is based on the NFL but should be applicable to most any level. HOw about TOXIC DIFFERENTIAL.

It essentially combines turnover differential with explosive play differential (defined as plays gaining more than 20 yards).



It has long been a proven adage that if you have a plus-2 turnover advantage on your opponent you will win the game 80 percent of the time. If you have a minus-2 explosive-play advantage it equally leads to winning about 80 percent of the time. If you have both you win close to 95 percent of the time. No other statistical combination that you can control has quite the same rate of predictive success.

[/b]

What if I can almost get to 95% with one stat and not a combination.

Well, by itself PER Differential is already at 86.1%



Here is BYU since 2006 with 135 PER:

51 games with a 135 PER (Record 48-3) (.941)

You are the stat geek I thought of when I read the article!



I was mostly surprised that big play differential was as equally as relevant as turnover differential in predicting outcome.



I know you are a big PER fan. I’d be curious to see if your stat applies to all or most teams.



I did a quick check on Utah’s 2008 team and Brian Johnson’s PER was less than 135 in 4 of the 13 games they won. (I chose that year because it was the last time Utah has had 1 QB play the whole season and I don’t have time to try to put together a full analysis of multiple QBs.)

Right so in 2008, Utah was 1.000 when reaching the PER standard. Maybe this weekend I'll do the record of a few teams this year when reaching the standard.

Golson has been <135 PER 5 times this season. I haven't combined Rees' stats for the games he played/started so I don't know the Team PER for each game.



I'm curious to see your full analysis because I'm too lazy to do it!

Cougs_Rule wrote:
This explains a lot of Riley's limitations. He rarely has big play ability and we now how turnover prone he is. His toughness and scrambling make up for some really deficits. Take away his scrambling with injuries...............ah forget it, broken record.

Tough to have big plays when you don't go downfield much. Doman hasn't done it much the last two years. Heaps wasn't very effective until Anae took the reigns off and let him go downfield too.

I had today off. Looked at record for the teams ranked 1st, 11th, 21st, 31st, 41st, etc and last in Total offense.



I looked at their W/L record. Their W/L record with a 135 PER. Their record when having the better PER. And the W/L record of the team with the better PER in those games were.



All tolled, I looked at 13 teams and 155 games. The teams and their total offense ranking are:

Louisiana Tech, 1st, 9-3 W/L; 6-2 135 PER W/L; 5-1 Better PER W/L

West Virginia, 11th, 6-5 W/L; 6-2 135 PER W/L; 5-0 Better PER W/L

UCLA, 21st, 9-3 W/L; 8-1 135 PER W/L; 8-0 Better PER W/L

San Jose St, 31st, 10-2 W/L; 9-1 135 PER W/L; 10-2 Better PER W/L

Air Force, 41st, 6-6 W/L; 4-1 135 PER W/L; 5-1 Better PER W/L

Oregon St, 51st, 8-3 W/L; 4-0 135 PER W/L; 8-0 Better PER W/L

UCF, 61st, 9-3 W/L; 6-0 135 PER W/L; 9-0 Better PER W/L

North Texas, 71st, 4-8 W/L; 2-1 135 PER W/L; 3-1 Better PER W/L

Miami (OH), 81st, 4-8 W/L; 3-2 135 PER W/L; 4-3 Better PER W/L

New Mexico, 91st, 4-9 W/L; 2-2 135 PER W/L; 2-0 Better PER W/L

Florida, 101st, 11-1 W/L; 6-0 135 PER W/L; 10-1 Better PER W/L

Idaho, 111th, 1-11 W/L; 0-3 135 PER W/L; 1-2 Better PER W/L (each game they reached 135 PER they also gave up 135 PER)

Maryland, 120th, 4-8 W/L; 2-4 135 PER W/L; 4-4 Better PER W/L



Totals: W/L 85-70 (.548); 135 PER W/L: 58-19 (.753); Better PER W/L: 75-15 (.833). The teams were a combined 12-53 (.185) when they did not have the Better PER for a combined Better PER W/L of 128-27 (.826).



Now for the BYU context. The team with the better PER record is 89-13 (.873) under Bronco. A total W/L of 73-29 (.716). And 135 PER record of 53-4 (.930). BYU's Bette PER W/L is 66-6 (.917).



Some other random other PER notes since I've been in my database today. The median PER for BYU since 1984 is 141.41. The median is the exact middle of all the numbers.



BYU's average points scored since 1998 with a 135 PER is 39.8. With a record of 60-5 in those games.



BYU only had a 135 PER 25 times in those 65 games against teams that won six or more games, going 20-5 in those games. So BYU went 40-0 against teams that won five or fewer games and reaching the 135 PER standard.



Of the 57 games under Bronco of a 135 PER, just 22 have come against teams that won 6 or more games, going 18-4 in those games.



Of the 45 games where BYU has fallen below a 135 PER, only 10 of them have come against teams that won five or fewer games, going 7-3 in those games.



BYU has had 15 games under 100 PER under Bronco, going 4-11 in those games. Nine of those games have come in the last three years. The average record of the opponent in those games was 8.9-3.6.



BYU is 10-0 under Bronco with a PER above 200, with just one in the last three years. The average record of the opponent in those games was 4.7-7.5