Wednesday, October 21, 2015

Turnover Index Week 7 - Expect the Jets to Regress

We're off to a good start for the 2015 season. Our first bet against the spread was successful, with the Browns covering (just barely) against the Broncos as a 4 point underdog. Here are the results (we'll add a row to this table each week):

weekbetswonstarting
bankroll
amount betprofitsending
bankroll
6 1 1 $1,000 $49 (5.0%) $45 $1,045

There have been no shortage of surprising starts this NFL season (it's been positively weird). Among those surprises are the New York Jets, who have won four of their first five, despite an offseason marred by coaching changes, bad debt, and sucker punches.

Over those first five games, the Jets have averaged three turnovers per game on defense, which currently leads the league. But turnover success on defense is fleeting, and history suggests the Jets are unlikely to maintain this pace throughout the season. Going back to the 1989 season, here is how teams have fared after racking up at least 15 defensive turnovers in their first five games (48 teams in total):

Teams with at least 15 Defensive Turnovers in their 1st 5 Games
turnovers/gmwin percentagainst the spread
Games 1-53.272%68%
Games 6-162.052%45%


We see that both turnover rates and win percentage regress heavily in the remaining 11 games - though 2 turnovers per game is still above the league average of 1.5, suggesting that defensive turnovers are not entirely random. Performance against the spread, however, does more than regress. It swings, like a dampened pendulum, from highly profitable in the first five games (68%) to moderately unprofitable in the remaining eleven (45%).

This pendulum effect nicely illustrates the principle behind the turnover index. The market is being fooled by randomness, conflating turnover luck with repeatable team skill. And where there is market inefficiency, there are profits to be had. With that in mind, here are the week seven picks against the spread (three games in total):

turnovers
gameawayhomediffper gamepickprobbet
Ravens @ Cardinals 4 13 9 1.5 Ravens 53.6% 2.6%
Cowboys @ Giants 3 12 9 1.4 Cowboys 53.4% 2.1%
Jets @ Patriots 15 7 -8 -1.6 Patriots 53.8% 3.1%

We are wagering 7.8% of our $1,045 bankroll this week on the Ravens, Cowboys, and Patriots.

5 comments:

  1. Inspired by this, I ran a regression the other day that suggested that 'fumbles forced' is a particularly predictive stat against the spread. Though - contrary to the turnover index - it expects teams that force more fumbles to do well against the spread.

    ReplyDelete
  2. This comment has been removed by the author.

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
  3. So how did the model perform? Not clear to me what point spreads you were using.

    ReplyDelete
    Replies
    1. 1-2. Post forthcoming. Also, there are no bets meeting the criteria for week 8.

      Delete