Turnovers in football
sosoangry
Senior Member
(I posted below in picks section but felt it was also relevant to this forum)
Very tough start for me this football season, but went 5-1 yesterday in NFL with Giants being the only loss. Not to sound too much like a famous doctor in the sports service business but it is important to review your picks and assess those that were bad, i.e. deserved to lose, and those that were good, i.e. a good pick that lost. In my opinion, the Giants were a good pick that lost. Why? Due to the most important determining stat in football: turnovers.
See the chart below based on NFL games since 2003 (or over 1000 games):
Net turnover margin, SU win pct., Margin of victory
+1 69% 6 points
+2 83% 10 points
+3 91% 13 points
+4 97% 17 points
As you can see above, a team that has just a +1 turnover advantage in a game ends up winning the game nearly 70% of the time, by an average margin of 6 points. And the win percentage and margin of victory get progressively higher with each incremental net turnover advantage.
Back to the NY Giants game -- what was the Arizona Cardinals net turnover over margin? A whopping +4, meaning history said the Cards had a 97% chance of winning that game, by an average margin of 17 points.
Now if you believe as I do that turnovers are random and cannot be handicapped, then when you look at the stats in the NYG/Ariz game, the NYG led in most stats (first downs, total yardage, etc.). Those are 'handicap-able" stats, turnovers are not. Over time, turnovers tend to even out for teams -- again, because they're more or less random, unpredictable events. That said, given the Giants led in many stats -- in particular total yardage (341-266) and yards per play (5.0-4.4) -- I believe the Giants were the right side considering a meager -2 line.
Another way to assess picks in relation to net turnover margins is to compare the final scores versus the historical margin of victories. Here are some examples from yesterday:
Teams with +1 net run over margin: Chargers, Redskins, Cowboys. All three teams won SU by more than the historical average win margin of 6 points. Therefore in isolation one can surmise all three teams deserved to win SU as they did not overly depend on turnovers to win the game.
Teams with +2 net run over margin: Bills and Browns. The Bills won by 19 points over the Dolphins, well ahead of the 10 point historical margin, meaning the deserved to win and didn't overly count on turnovers. However, the Browns won by just two points -- or 8 points below the 10 point average margin of victory historically -- meaning the +2 net turnover margin for the Browns appeared to help them attain the SU win.
As for +4 net turnover margin games yesterday, benefiting were Arizona, Chicago, Houston and the Patriots. The historical average margin of victory for +4 net turnovers is 17 points. NE won by more that 17, so they did not overly depend on turnovers to win, but Houston won by 16 which is just under the 17, and Arizona won by 11 and Bears won by 8, both well under the 17 point average margin of victory and both teams seemingly depending more highly on turnovers to win their games.
Very tough start for me this football season, but went 5-1 yesterday in NFL with Giants being the only loss. Not to sound too much like a famous doctor in the sports service business but it is important to review your picks and assess those that were bad, i.e. deserved to lose, and those that were good, i.e. a good pick that lost. In my opinion, the Giants were a good pick that lost. Why? Due to the most important determining stat in football: turnovers.
See the chart below based on NFL games since 2003 (or over 1000 games):
Net turnover margin, SU win pct., Margin of victory
+1 69% 6 points
+2 83% 10 points
+3 91% 13 points
+4 97% 17 points
As you can see above, a team that has just a +1 turnover advantage in a game ends up winning the game nearly 70% of the time, by an average margin of 6 points. And the win percentage and margin of victory get progressively higher with each incremental net turnover advantage.
Back to the NY Giants game -- what was the Arizona Cardinals net turnover over margin? A whopping +4, meaning history said the Cards had a 97% chance of winning that game, by an average margin of 17 points.
Now if you believe as I do that turnovers are random and cannot be handicapped, then when you look at the stats in the NYG/Ariz game, the NYG led in most stats (first downs, total yardage, etc.). Those are 'handicap-able" stats, turnovers are not. Over time, turnovers tend to even out for teams -- again, because they're more or less random, unpredictable events. That said, given the Giants led in many stats -- in particular total yardage (341-266) and yards per play (5.0-4.4) -- I believe the Giants were the right side considering a meager -2 line.
Another way to assess picks in relation to net turnover margins is to compare the final scores versus the historical margin of victories. Here are some examples from yesterday:
Teams with +1 net run over margin: Chargers, Redskins, Cowboys. All three teams won SU by more than the historical average win margin of 6 points. Therefore in isolation one can surmise all three teams deserved to win SU as they did not overly depend on turnovers to win the game.
Teams with +2 net run over margin: Bills and Browns. The Bills won by 19 points over the Dolphins, well ahead of the 10 point historical margin, meaning the deserved to win and didn't overly count on turnovers. However, the Browns won by just two points -- or 8 points below the 10 point average margin of victory historically -- meaning the +2 net turnover margin for the Browns appeared to help them attain the SU win.
As for +4 net turnover margin games yesterday, benefiting were Arizona, Chicago, Houston and the Patriots. The historical average margin of victory for +4 net turnovers is 17 points. NE won by more that 17, so they did not overly depend on turnovers to win, but Houston won by 16 which is just under the 17, and Arizona won by 11 and Bears won by 8, both well under the 17 point average margin of victory and both teams seemingly depending more highly on turnovers to win their games.
Comments
Over the last 4 years, Eli Manning has thrown 83 picks. I didn't check, but I would be extremely surprised if anyone else in the NFL was even close to that number over that time frame. Do you think he is just randomly throwing interceptions over and over again, or that there might be something to it?
Yes, from 2010-2013, four seasons, Eli has 83 interceptions, but in that time period the average INT % for all NFL QBs was 3.6% and in that same time period Eli had an INT % of 3.7%, or a tad higher than the average. In 2010 and last year, he was higher than average at 4.6% and 4.9% respectively, but in 2011 and 2012 he was lower than average at 2.7% and 2.8% respectively. So if anything Eli appears to be erratic as opposed to a non-random INT thrower, and he can somewhat prove my point that bad luck one year (2010) was followed by better luck the next (2011) = random. Random occurrences over time will tend to even out. In this case, Eli regressed to the average INT % (or close to it) in the four year period.
That said I prefer not to isolate on one player as my observation about turnovers being random is based on many games (1000+) over many years. You can always attempt to find outliers or single "data points" to prove a point, but I prefer to look at a much larger set of occurrences.
In my opinion, no matter how good the basic stats look, if you have a ballhawking, turnover-creating defense playing vs a turnover prone QB like Manning is, you throw those stats out the window. My pick was mostly based on the probability of the NYG committing the most turnovers in this game.
If you look ahead to their next matchup vs the Texans, you'll see that on the turnover front, a -6 team is playing a +5 team. The question then becomes: Can the Giants keep the turnovers at a minimum, so they'll have a positive turnover differential? If that happens, they've been gifted +5.5 points by the oddsmakers with the current line. That's a +EV bet. However, my research shows that teams that start bad, continue to be bad for a while.
But I've never heard of anyone successfully handicapping based on predicted turnovers. If you can, more power to you, but I sure cannot.
Looks like all the responses were over here anyway so hope you understand.
I do not believe this, so I guess I'm done reading
I cannot either, but my model does warn for over-/undervaluation of a team depending on their turnover differential.
but as Durito alluded to, not all turnovers are created equal. Some are the result of bad decisions by the QB and others are the result of bad luck. Look at the Pittsburgh - Baltimore game of last thursday. The Steelers ended with a -3 turnover differential. However, none of them were caused by mistakes. One was a tipped pass and the other 2 were forced fumbles. So, Pittsburgh scores only 6 points, but they could've scored 18 and kept Baltimore to 14.
Looking at the upcoming Pittsburgh vs Carolina game, my numbers show a mediocre offense, playing on the road, at a team that has excelled at defense and creating turnovers. Which team has the highest chance of getting the positive turnover differential? Carolina or Pittsburgh? It's Carolina off course, even though the current numbers for both teams should regress to normal. Pitt on pace for -32 and Carolina on pace for +48, both are not going to happen. But as Pittsburgh has created zero turnovers and Carolina has committed zero turnovers, the probability that Carolina wins and covers is the greatest here, just because of the probability that they'll receive a couple of free extra possessions from Pittsburgh And I have not looked at a single yardage or completion stat. But I'm pretty sure my handicapping makes more sense than all those other numbers combined.
Agree on fumbles.
I don't think anyone can say Eli Manning attempting 39 passes against Arizona's defense and throwing 2 picks is horribly unlucky, or even that uncommon for that many pass attempts.
This really can't be emphasized enough. In addition to the randomness of fumbles & (somewhat less of) interceptions, where on the field they happen & when in the game are huge. A hail Mary intercepted with 2 seconds to go in the 1st half is nowhere near as damaging as a fumble in the red zone in the 4th quarter but there's no difference between those two on the stat sheet. To me that's what makes turnovers unhandicapable.
As opposed to discussing turnovers and there effect on outcomes, I think per chance this discussion has become more about dislike for Eli Manning. Wasn't my point to start with....
Yup, not surprising given the numbers I show for 1000+ games back to 2003:
Net turnover margin, SU win pct., Margin of victory
+1 69% 6 points
+2 83% 10 points
+3 91% 13 points
+4 97% 17 points
23-5 = 82% win percent. I show above just a +1 net turnover margin = 69% win rate. Assuming the 28 games of the 23-5 mentioned average out to +2 net turnover margin, the 82% falls in line with the 83% win rate for +2 turnover margin going back to 2003.
I guess I would also say that if turnovers were predictable, or easy to predict, we'd see more handicappers winning at 70+% rate since the ATS results are not far off from the SU results going back to 2003:
+1 turnover margin = 69% SU and 70% ATS
+2 " " = 83% SU and 82% ATS
+3 " " = 91% SU and 89% ATS
+ 4 " " = 97% SU and 95% ATS
Correctly predict turnover margin = world is your oyster.
Wanted to jump in on this one earlier but I had to leave the house for a bit.
Eli's number are bad. But back in 2010 when he threw for 25 INT's, I believe the guys at PFF did a breakdown of every single INT and turned out that about 10 were plain Eli's mistakes, while the rest were the fault of the receiver running the wrong route, tipping the ball in the air and/or a hail mary at the end of a half/or game.
This is exactly what I saw in their first game. A couple of tips that were on target but the receivers bobbled the ball, and also Eli not being in sync with his receivers. This will also lead to wrong routes and more INT's. So it's not all his fault. As a bettor, just wait till you see the Giants offense click and watch the cash roll in.
Predicting the turnover margin is very difficult. But you're on the right track for the fact that at least your including it in your analysis.
That's why it's funny to see how I used to cap games a couple of years back, while using yards, 3rd down% and points etc etc. That method is akin to coin flipping, no matter how you turn it.