Turnovers in NFL Handicapping

Turnovers in NFL Handicapping At the end of Old School's power ratings thread for Week 6, he notes the team that's plus in TO's is 50-4 ATS this year. Another poster asks for ideas in predicting TO's. I don't think Turnovers can be predicted. While not entirely random, there's enough randomness in them to prevent reliance on any attempts to predict them. I do think turnovers can be integrated into handicapping in the following way, and would appreciate feedback: If a team has performed well (in either point differential or W/L record) despite a poor T.O. ratio, wouldn't that suggest the team is actually better than its W/L or point differential would suggest? Conversely, if a team has performed poorly despite having a plus T.O. ratio, wouldn't the opposite be true? Examples: Through Week 5, G.B.'s point differential is only +11, yet its T.O. differential is 2nd in the league at +7. The other teams in the top 5 of T.O. differential show point differentials ranging from +56 to +80. Through Week 5, Dallas and Pitt show point differentials of +24 and +15, yet both show a T.O. differential of -4. Other teams at or around -4 are showing point differentials of from -55 to -81. To me (and in the simplest terms), this suggests G.B. is intrinsically not as good a team as its record or perception may be. And that Dallas and Pitt are both better. Obviously, opponents' strength would play into this somewhat. I don't think T.O.'s are predictable, but they can be used in handicapping just the same. Thoughts?

Great analysis IMO. May take Pittsburg vs Cleveland coming off a lucky win at Buffalo. I have GB only an 11 point favorite, and the line is 13 1/2.

In college handicapping you must consider the style of play. For example Air Force is +16 in turnovers after posting +13 last year and +10 in 2007. Obviously they don't pass often so interceptions are rare. That said, you have to consider the Falcons overrated as they are just 2-2-1 ATS despite the +16 margin.

There is going to be some predictability I think. I'd guess fumbles are going to be mostly random (qb fumbles less random). Interceptions are going to be highly coorelated to passes attempted, and effectiveness of the QB and defense. An effective QB with little rush pressure from the defense, is going to result in far less turnovers than a shitty QB under heavy pressure. A problem that could come up is that teams that are losing are going to be in more bad passing situations.. so the losing is actually going to cause some of the turnovers and skew the ATS record of turnover margin. This is like the rushing yards and attempts coorelation to winning. Yes, having a better running game is benficial to winning, but leading in the game also skews the rushing numbers in that teams direction. My guesses to factors that would predict turnovers would be.. Passes attempted, sacks forced/allowed. Completion % and completion % allowed.

In the NFL, I have found it profitable to actually play on the team with the worse turnover margin. The sample size is anywhere from 650 plays to over 1000 plays going back to 1983 depending on the different parameters used. The idea is the turnovers will eventually even out. It's not quite that simple as there are other parameters to this but looking at the worse turnover team is a good place to start with finding teams that will cover the spread.

[QUOTE=Sixth Sense;5753]In the NFL, I have found it profitable to actually play on the team with the worse turnover margin. The sample size is anywhere from 650 plays to over 1000 plays going back to 1983 depending on the different parameters used. The idea is the turnovers will eventually even out. It's not quite that simple as there are other parameters to this but looking at the worse turnover team is a good place to start with finding teams that will cover the spread.[/QUOTE] Is it possible this is an angle that has worked in the past that may not work in the present and future? People are understanding more and more about past turnovers and how predictive they are of future games. If the market priced them incorrectly in the past (thus giving teams with high TO margins a bigger power rating in future games), but don't make that mistake as much any more, then that angle of betting on the team with more TOs would no longer be any good.

Don't have the numbers in front of me but it was profitable last year and I believe has been profitable just about every year going back to 1983. Would have to check as there may have been a year or two where it was a game or two under .500. The situation I use doesn't apply until week five and ends after week 12. What I have found is late in the year, the situation doesn't perform as well. I imagine that is because the bad teams (which are usually poor turnover margin teams) have quit for the season. Keep in mind it is just one piece of handicapping. There can be other situations going against the turnover plays, either letdown, bounce back, fundamental rushing plays, etc. Also, the line needs to be fair as well. So, a lot more goes into this than just blindly playing the situation. If it keeps me off a team, that can be just as good as playing on a team.

thank you for the reply, makes sense

[QUOTE=Sixth Sense;5753]In the NFL, I have found it profitable to actually play on the team with the worse turnover margin. The sample size is anywhere from 650 plays to over 1000 plays going back to 1983 depending on the different parameters used. The idea is the turnovers will eventually even out. It's not quite that simple as there are other parameters to this but looking at the worse turnover team is a good place to start with finding teams that will cover the spread.[/QUOTE] Are you talking about using the one team with the worst T.O. margin in the entire league each week? Or in each game, using the team with the worst margin as between those two opponents? (Understand it's just one factor and not automatic) Also, any data on those rare situations where a team with the worst margin is playing against a team with the best? Seems like that type of 'perfect storm' might show even more dramatic results.

Awesome thread Gaslamp!