ESPN: Predicting Interceptions
I wrote a piece for ESPN+ today about using passes defensed to predict interceptions. The article features the projections for top contenders but I wanted to run the whole table here. The basic idea is that passes defensed are a better future predictor of interceptions than past interception totals. There's randomness to interceptions, similar to fumble recoveries. These numbers predict future interceptions using both passes defensed and schedule strength.
Here are the correlation of certain stats from Weeks 1-9 and the interception projections to interceptions in Weeks 10-17:
- Offense interceptions: .17
- Offense PD: .26
- Offense projection: .37
- Defense interceptions: .08
- Defense PD: .21
- Defense projection: .38
And here's a table with projected interceptions both offense and defense for all 32 teams in Weeks 11-17:
| Team | Games Left |
OFF INT 1-10 |
OFF PD 1-10 |
Proj OFF INT 11-17 |
DEF INT 1-10 |
DEF PD 1-10 |
Proj DEF INT 11-17 |
|---|---|---|---|---|---|---|---|
| MIN | 6 | 3 | 18 | 3.3 | 9 | 55 | 5.7 |
| SEA | 6 | 2 | 28 | 3.5 | 8 | 45 | 4.2 |
| GB | 6 | 2 | 31 | 3.9 | 9 | 41 | 5.7 |
| KC | 6 | 1 | 32 | 4.0 | 6 | 35 | 4.3 |
| TEN | 6 | 6 | 34 | 4.3 | 9 | 50 | 4.5 |
| WAS | 7 | 9 | 36 | 4.7 | 8 | 26 | 4.5 |
| OAK | 7 | 4 | 29 | 4.7 | 7 | 36 | 6.7 |
| NYJ | 7 | 12 | 36 | 4.9 | 5 | 36 | 6.2 |
| PIT | 7 | 7 | 32 | 4.9 | 14 | 49 | 7.8 |
| ARI | 6 | 5 | 39 | 5.0 | 4 | 32 | 3.8 |
| NE | 7 | 6 | 26 | 5.1 | 19 | 52 | 5.7 |
| DEN | 7 | 5 | 29 | 5.1 | 5 | 38 | 5.1 |
| HOU | 7 | 6 | 28 | 5.2 | 5 | 42 | 6.7 |
| NO | 7 | 5 | 28 | 5.2 | 4 | 35 | 6.0 |
| CAR | 7 | 6 | 38 | 5.3 | 12 | 44 | 6.1 |
| BUF | 7 | 8 | 29 | 5.3 | 6 | 47 | 5.4 |
| LAC | 6 | 10 | 49 | 5.7 | 7 | 25 | 1.6 |
| SF | 7 | 8 | 33 | 5.9 | 11 | 45 | 5.1 |
| CLE | 7 | 12 | 54 | 5.9 | 4 | 26 | 5.7 |
| ATL | 7 | 10 | 42 | 5.9 | 2 | 20 | 4.5 |
| NYG | 6 | 10 | 57 | 6.0 | 8 | 35 | 4.6 |
| DAL | 7 | 9 | 37 | 6.1 | 4 | 36 | 5.2 |
| CHI | 7 | 5 | 41 | 6.2 | 6 | 39 | 5.1 |
| PHI | 7 | 5 | 40 | 6.2 | 8 | 40 | 5.9 |
| DET | 7 | 6 | 39 | 6.2 | 3 | 46 | 6.9 |
| BAL | 7 | 6 | 35 | 6.3 | 8 | 42 | 6.4 |
| MIA | 7 | 13 | 53 | 6.4 | 6 | 35 | 6.4 |
| JAX | 7 | 4 | 44 | 6.4 | 7 | 33 | 4.7 |
| IND | 7 | 7 | 40 | 6.6 | 6 | 23 | 4.2 |
| TB | 7 | 14 | 52 | 6.7 | 6 | 44 | 6.1 |
| LAR | 7 | 10 | 50 | 6.9 | 6 | 37 | 4.3 |
| CIN | 7 | 9 | 58 | 7.0 | 3 | 28 | 6.0 |
Comments
1 comment, Last at 13 Nov 2019, 11:37am
#1 by Pat // Nov 13, 2019 - 11:37am
Holy crap, New England's interception rate is projected to *plummet*.
If you look at "current interception rate/game" versus "predicted rate over remaining games", the biggest difference (in interceptions/game):
NE: -1.30
SF: -0.49
CAR: -0.46
PIT: -0.44
LAC: -0.44
The Chargers are interesting there because they already have the 3rd fewest picks in the league.
and on the flipside (projected to increase)
DET: 0.65
CIN: 0.52
ATL: 0.42
NO: 0.41
HOU: 0.40