Yesterday, I discussed how expected shooting percentage (aka shot quality) varies when teams have the lead. Today I want to look at how even-strength shooting characteristics vary both when teams are leading and trailing. First, raw shot totals since the lockout:
Margin | Leading | Trailing | Trail/Lead |
1 | 51080 | 58110 | +13.8% |
2 | 21950 | 26980 | +22.9% |
3 | 8914 | 11307 | +26.8% |
We can see that the further a team gets behind, the more they out-shoot their opponents. This shouldn't be surprising – there's limited downside to giving up an additional goal when you're already down, so you can go all out on the attack. A team with the lead is content to sit on it.
Actual shooting percentage:
Shot % | Actual | Relative to Tied |
+3 | 7.07 | +28.2% |
+2 | 7.80 | +41.4% |
+1 | 6.71 | +21.8% |
Tied | 5.51 | |
-1 | 6.11 | +10.9% |
-2 | 6.22 | +12.9% |
-3 | 6.33 | +14.9% |
This version of shooting percentage includes goals, shots on goal and missed shots. We can see that shooting percentage increases as teams have bigger and bigger leads, but doesn't change much as they fall further behind – no amount of offensive pressing can really generate that many more scoring chances.
In terms of shot quality, we see roughly the same effect – expected shooting percentage increases as a team gets in the lead, but stays essentially constant as it falls further behind:
Expected Shot % | Expected | Relative to Tied |
+3 | 6.25 | +13.3% |
+2 | 6.17 | +12.0% |
+1 | 6.08 | +10.2% |
Tied | 5.84 | |
-1 | 5.96 | +8.2% |
-2 | 5.97 | +8.3% |
-3 | 5.95 | +7.9% |
After a lot of database machinations, I'm convinced I have these numbers right. If anyone has any conflicting data, please let me know!