Third Down Conversions: up is downThird down is the down in NFL football since if you pick up the needed yardage you keep going, if not the drive is typically done. Teams even have "third down specialists" on both sides of the ball who have roles suited primarily to the key down plays.On the surface then it would figure that teams with good third down conversion (or good defensive stops on third down) would be teams to play on the line. This is dubious in reality. With small sample sizes in an NFL season the third down plays are a little bit like turnovers in that luck weighs in heavily and yet can be a big factor in outcomes. I started my research with the theory that maybe like turnovers you would want to play the team with the poor third down results so far. FootballOutsiders.com has also had some articles on this subject although for them it's more about predicting the next season results based on last season 3rd down stats. Pick = whichever team has the higher number on [Def. 3rd Down % - Off. 3rd Down %]
This kind of confirms the basic theory that playing on the teams doing badly on 3rd down is better than playing the teams doing well. 52% overall isn't bad for a seven year set of games. Throw out the first column where the differences are pretty minimal and you've got a 54% success rate. The next idea was to combine this indicator with something else. For starters I tried the solid DC drive chart predicted score. Pick = worse 3rd down team (by at least 6%) that's also picked by the DC projection, with the DC overlay being at least 3.5 points (more than a field goal).
Yes! Now we have a 62% historical success rate. Not a ton of plays at an average of 17/season. When they happen they are dynamite. If you have a dog qualifier getting more than a field goal point spread points you are talking 24-6. Some people don't like when there are multiple filters. If we don't look for a DC big overlay of 3.5+ but just go with any DC agrees with the pick as the bad 3rd down team, it's 139-100 (58%). Next up I looked at Bad 3rd Down Teams also picked on the DC/TO which boosts the bad turnover teams up.
That's a 58% two indicator system. I still have the 6%+ minimum 3rd down difference in effect. I also see that when the DC/TO overlay gets up to 4+ it's 61% and higher territory. At this level almost all the plays are big dogs. The theory again is that bad turnovers and bad third down will under-rate a team and they will be ready to pop unexpectedly the next game out! If I don't cherry pick quite so much on the filter and use a 2+ overlay instead of the 4+ it is 180-110 over the seven years. Looking at it year by year shows:
Only one iffy season on the dogs at 21-18 in the favorite heavy 2005 season. The funny thing is if you look for all three indicators in agreement (3rd down + DC + DC/TO) then it's not all that exciting. When two of the three agree and the third disagrees then you have something. You can find the info you need to make these bets in the Investor Subscription for NFL 2008.
|