|
Last season saw the "Turnover Difference" theory return to prominence with an outstanding year: 79-50 for a 61.2% win rate! However, the turnover plays had been inconsistent the previous four years despite the solid overall record from 1983 on. This got us to thinking whether there was a way to improve consistency/results...
Last year's research centered on breaking out interceptions and fumbles into different groups. While there were some interesting points to be gleaned from this research, and both factors were strong in their own right in 2001 (fumbles: 57-35 for 62%, interceptions: 65-44 for 60%), it didn't mark a dramatic difference from the basic theory.
One of our columnists last season focused on red zone stats, and by season's end the conclusion was the RZ stats were not successful when applied conventionally, but the turnovers might hold some potential. Thus it is we set out to test the turnover theory using only red zone stats.
The starting point remains the same -- the concept is that teams that have turnovers are often as unlucky as they are "guilty" and so by finding match-ups where one team has a far superior record with turnovers, by playing the team with the poor turnover history you can often be finding great value. Turnovers have such a large impact on a game's outcome, and indeed how the game plays out (a rushing-oriented team is suddenly forced to pass since they are behind...) that they can skew the statistics and particularly the public estimation of a team's true ability.
We'll take a look at this approach using a database that includes all the games played from week 5 through week 17 in 2001:
Red Zone
T/O Diff |
1% |
2%-3% |
4%-5% |
6%-7% |
8% + |
ALL |
| HomeFavs |
0 - 1 |
5 - 3 |
4 - 2 |
5 - 0 |
14 - 15 |
28 - 21 |
| HomeDogs |
2 - 1 |
2 - 1 |
3 - 3 |
2 - 0 |
16 - 8 |
25 - 13 |
| AwayFavs |
0 - 0 |
2 - 0 |
1 - 2 |
5 - 4 |
6 - 5 |
14 - 11 |
| AwayDogs |
1 - 1 |
7 - 4 |
3 - 4 |
2 - 1 |
31 - 14 |
44 - 24 |
|
| Favorites |
0 - 1 |
7 - 3 |
5 - 4 |
10 - 4 |
20 - 20 |
42 - 32 |
| Underdogs |
3 - 2 |
9 - 5 |
6 - 7 |
4 - 1 |
47 - 22 |
69 - 37 |
| Home Teams |
2 - 2 |
7 - 4 |
7 - 5 |
7 - 0 |
30 - 23 |
53 - 34 |
| Away Teams |
1 - 1 |
9 - 4 |
4 - 6 |
7 - 5 |
37 - 19 |
58 - 35 |
|
| ALL PICKS |
3 - 3 |
16 - 8 |
11 - 11 |
14 - 5 |
67 - 42 |
111 - 69 |
| WIN % |
50 % |
67 % |
50 % |
74 % |
61 % |
62 % |
| Spread Range |
10+ pts |
5 to 9.5 |
0 to 4.5 |
| Favorites |
1 - 1 |
15 - 12 |
26 - 19 |
| Underdogs |
8 - 5 |
29 - 17 |
32 - 15 |
ANALYSIS: Well, that's a pretty impressive overall record, given you'd be playing every game with even a 1% difference between the two teams! I think all of us would take a 62% season. However, last season was exceptional in regards to underdogs covering, and in a connected circumstance, the success of the contrarian style plays. Even with that caveat, one of the nice things the table above displays is that the favorite plays were a healthy 42-32 (57%) in a year when favorites were getting creamed! Of course the underdogs with their 69-37 (65%) record are more enticing. If you were going to backfit the results you probably make a rule to require a 6%+ difference between the two teams to have a play.
Before we get too excited and start planning our wagers for next season, let's see if we can make a case for why the red zone turnovers could be so predictive. Here are the reasons that come to mind:
- Turnovers in general have predictive value
- Red Zone turnovers are more critical since a team with the ball in the red zone has a high expected points number for the drive (already in field goal range)
- Red zone mistakes can be especially deceptive to the final score -- a team can be moving the ball very well on an opponent but a couple of R/Z errors and it leads to zero points. Likewise a team's defense can be very shaky but a get a couple of turnovers deep in their own half and on the surface it might look like the defense did a good job.
- Red Zone turnovers may have more of a mental effect than a typical turnover in that it can deflate a team's morale (on the negative side) or give an overly large boost (on the positive side)
Fundamentally we believe it comes down to point #2 -- a team that turns over the ball at midfield is sacrificing some potential points, but not nearly as many as a team turning the ball over at its opponents' five yard line!
The next concern is that we've only looked at one season so far. Let's go back then to 1999:
Red Zone
T/O Diff |
1% |
2%-3% |
4%-5% |
6%-7% |
8% + |
ALL |
| HomeFavs |
3 - 1 |
6 - 3 |
6 - 4 |
5 - 3 |
8 - 10 |
28 - 21 |
| HomeDogs |
2 - 2 |
3 - 0 |
3 - 3 |
3 - 2 |
10 - 16 |
21 - 23 |
| AwayFavs |
0 - 2 |
1 - 3 |
1 - 3 |
1 - 3 |
2 - 1 |
5 - 12 |
| AwayDogs |
4 - 2 |
3 - 8 |
7 - 4 |
4 - 3 |
17 - 15 |
35 - 32 |
|
| Favorites |
3 - 3 |
7 - 6 |
7 - 7 |
6 - 6 |
10 - 11 |
33 - 33 |
| Underdogs |
6 - 4 |
6 - 8 |
10 - 7 |
7 - 5 |
27 - 31 |
56 - 55 |
| Home Teams |
5 - 3 |
9 - 3 |
9 - 7 |
8 - 5 |
18 - 26 |
49 - 44 |
| Away Teams |
4 - 4 |
4 - 11 |
8 - 7 |
5 - 6 |
19 - 16 |
40 - 44 |
|
| ALL PICKS |
9 - 7 |
13 - 14 |
17 - 14 |
13 - 11 |
37 - 42 |
89 - 88 |
| WIN % |
56 % |
48 % |
55 % |
54 % |
47 % |
50 % |
| Spread Range |
10+ pts |
5 to 9.5 |
0 to 4.5 |
| Favorites |
0 - 4 |
11 - 12 |
22 - 17 |
| Underdogs |
5 - 12 |
26 - 18 |
25 - 25 |
ANALYSIS: Wow! Those are very different results, even though the general turnover theory was very solid in 1999. About the only area that had a semblance of the same results were the "medium dogs" getting 5 to 9.5 points. Overall though, 62% quickly becomes 50% which is why you never want to get too jazzed about some research until you re-test it against another sample.
We'll dig out a few more years at some point since the 2001 results were so strong. Ultimately we ere having a hard time buying into pure red zone play for the same reason we were doubtful about our columnist's efforts last year: the red zone stats are a small subset of a team's overall data, they tend to overlook the importance of big play ability (eg scoring from out side the red zone, ala exhibit A St. Louis!), and they can reflect too much garbage time with teams playing prevent defenses.
What we do believe is that the next revolutionary step in turnover difference theory is to quantify each turnover by situation to create a numeric t/o value. After all, a fumble at your own 2-yd line can be as damaging as a fumble in the red zone. Given we already track the games at the play-by-play level, this is an easy enough project to undertake...
|
