Why 3rd Down Conversion Rate is Worthless

Not all 3rd downs are created equal. Clearly a 3rd and 2, is much easier to convert vs. a 3rd and 10, right? Than, why do we give so much credence to a RAW “3rd Down Conversion Rate”? I noted this issue last year when we saw how low Mitch Trubisky’s 3rd conversion rate was, when those same noting that didn’t realize the Bears had one of the highest “yards to go”/3rd down.

Instead, wouldn’t a better practice be to bucket, and weight each third down (like baseball’s slugging percentage / how we’ve applied the same “multiplier” in “Pass Rushing Slugging Percentage“. As in, weight the 3rd downs based on the yards to go. Say, as follows:

  • 3rd and Short (1-2 yards to go)
  • 3rd and Medium (3-5 yards to go)
  • 3rd and Long (6-10 yards to go)
  • 3rd and TOO Long (usually a toss away play in neutral game scripts) (>10 yards to go)

As it turns out, running a simple regression, you can “explain” more than 1/3 (.35) of a team’s “traditional” 3rd down conversion rate based on how far the third down needed to go.

Applying this logic, the 2020 NFL season’s Adjusted 3rd down conversions shape out like this:

Tot. 3DsAdj Tot% of 3D by YTG1-2 YTG3-5 YTG6-10 YTG>10Tot. 3D Conv.3D RkWt.ed 3D %Wt.ed RkDiff
SEA168418SEA19.64%27.98%36.31%16.07%38%2792%1116
NE152369NE26.32%21.05%36.18%16.45%41%17111%413
CAR181462CAR20.44%21.55%40.33%17.68%39%2484%1311
MIN164409MIN23.78%20.73%37.80%17.68%41%16100%79
MIA209491MIA25.36%28.71%31.58%14.35%39%2679%197
TEN166396TEN25.30%25.90%33.73%15.06%45%7115%34
HOU186447HOU24.73%25.81%33.87%15.59%43%1296%93
NYJ206516NYJ14.08%32.04%43.20%10.68%34%3166%283
LAR193458LAR22.80%30.05%34.20%12.95%41%1490%122
NYG203538NYG14.78%27.09%36.45%21.67%36%2968%272
BAL168415BAL24.40%17.86%44.05%13.69%49%2117%11
LV182421LV26.92%24.18%39.56%9.34%46%6109%51
SF202520SF18.81%24.75%36.63%19.80%39%2375%221
CLE192466CLE24.48%25.00%33.85%16.67%45%896%80
IND200508IND18.50%26.00%38.50%17.00%41%1880%180
CHI217550CHI18.89%24.88%40.09%16.13%33%3261%320
GB179439GB20.11%26.82%40.78%12.29%51%1116%2-1
NO184445NO23.91%25.54%35.33%15.22%47%5105%6-1
ARI216550ARI20.83%24.54%33.80%20.83%40%2272%23-1
CIN230587CIN22.17%21.74%34.78%21.30%36%3062%31-1
DET197512DET17.26%23.86%40.61%18.27%41%1580%17-2
JAC211529JAC18.96%27.49%37.44%16.11%41%1977%21-2
PHI235599PHI17.02%28.51%37.02%17.45%37%2862%30-2
PIT216526PIT24.07%22.69%38.89%14.35%43%1381%16-3
DEN220599DEN14.55%22.27%39.55%23.64%39%2565%29-4
TB213527TB20.19%29.11%33.80%16.90%44%984%14-5
DAL229584DAL20.96%23.14%35.81%20.09%40%2069%25-5
WAS237585WAS22.36%21.94%42.19%13.50%40%2168%26-5
BUF208506BUF22.12%27.40%35.58%14.90%47%492%10-6
ATL224561ATL17.86%29.91%36.16%16.07%44%1178%20-9
KC225572KC19.56%28.00%31.11%21.33%48%384%15-12
LAC249617LAC21.69%26.51%34.14%17.67%44%1072%24-14

Biggest takeaways when you weight each third down conversion:

  • Seattle, New England and Carolina were considerably “better” at converting 3rd downs
  • LA Chargers and (believe it or not) the Chiefs were far worse at converting 3rd downs

The main point here: 3rd down conversions, especially as a raw statistic, really has no value when looking at it in a vacuum. Without knowing how far you need to convert, what’s it matter if you convert at x percent or y percent? Unless you adjust for difficulty, you don’t glean good information.

Leave a Reply