Deconstructing Wind Speed Narratives
Forecasts of high winds normally draw these narratives: shorter passing and more running. A 2012 study by Brian Burke shows an average drop in AY/A (quarterback efficiency metric that tracks yards per attempt while accounting for touchdowns and interceptions thrown) as wind speed climbs. To break down the link between wind speed and QB output, the study was expanded. All QBs with at least two seasons from 2001 to 2016 were added. Game stats scraped from Pro-Football Reference were used to include pass attempts, fantasy points, and run:pass ratio.
The narratives seem to hold merit based on the plots. Attempts, AY/A, and fantasy points scored trend down while run:pass ratio trend up. At extreme conditions, AY/A is down 9.56%, attempts are down 12.23%, and run:pass ratio is up 13.31%. From this view, owners would be advised to stay away from teams playing in these conditions.
However, the trend lines imply low correlation (average R-squared value is .06). Additionally, the decline is shallow (slopes of -0.09, -0.04, -0.13, and 0.008, respectively). Plus, previous articles reference games showing this behavior can’t be universally applied.
Colder temperatures also produce similar narratives with fewer pass attempts and more rushing. The same data set was compared to temperature with the results shown below.
Using a macro lens, casual owners would draw similar conclusions from the data. Cold weather definitely slows the passing game. The plot clearly shows a declining average number of pass attempts as it gets colder. But again, this data carries a low correlation (0.001) and only a soft incline (0.019).
The trend lines, albeit simple linear fits, point to the weather having an effect. However, the low significance values are a sign that more detail must be applied to the data. Any useful trends that exist beneath the surface require a closer look at each QB. The next section gives a preview of what’s to come.
Beneath the Surface
Detailed analysis was limited to QBs with starts in 2017. Wind speed brackets contextualize the conditions from baseline (0-4 mph) to extreme (15-19 mph). Stats within each bracket were averaged as a reference point. Furthermore, a range of outcomes was made from the associated error.
Each bar represents the range of outcomes (average ± error) for each condition. Larger bars are likely due to small sample size, making conclusions shaky, but the averages are noteworthy.
- Population average at baseline conditions is 34.34 attempts per game.
- Drew Brees has the highest average at 38.94. Alex Smith has the lowest average at 28.53.
- Population average at extreme conditions is 33.04 attempts per game.
- Blake Bortles has the highest average at 46. Tyrod Taylor has the lowest average at 25.5.
- Population average delta shows a 1.65% decrease in pass attempts.
- Taylor has the greatest delta at 25%. Matthew Stafford has the lowest delta at 0.08%.
Russell Wilson’s and Taylor’s below-average pass attempts at baseline conditions were noteworthy, but their rushing floors balance their fantasy output. At extreme conditions, all QBs most affected saw a minimum decrease of five pass attempts. Bortles’ and Kirk Cousins’ increases in attempts are buoyed by small samples with high volume. Tom Brady’s stats are most noteworthy. Over a 25-game sample, his pass attempts actually increase at extreme conditions by 3.67%.
- 10 out of 20 QBs drop in average pass attempts at extreme conditions.
- Data from QBs with samples size of more than 10 games deemed ‘safe’ to extract usage data for future studies.
- Hypothesized ‘safe’ data would lead conclusions about changes in personnel usage.
- Future study required for specific cases.
The decrease in pass attempts leads to the second narrative of increased rushing. Stats were used to calculate the run:pass ratio for each QB. Low values denote more passing while higher values indicate the opposite.
- Population average at baseline conditions is 0.79.
- Bortles has the lowest average at 0.38. Cam Newton has the highest average at 0.90.
- Population average at extreme conditions is 0.91.
- Wilson has the highest average at 1.26. Bortles has the lowest average at 0.17.
- Population average delta is a 13.31% increase in R:P ratio.
- Taylor has the largest delta at 55.04%. Andy Dalton has the smallest delta at 0.9%.
Data on Bortles points to game flow having a significant effect. Conservative teams or those with elite running backs have R:P values closer to 1.
- 14 of 20 QBs saw shifts towards increased running at extreme conditions.
- Of QBs most affected, non-mobile QBs see largest drop off in fantasy points per game.
- Future study required on correlation to RB scoring.
If rushing is up and attempts are down, are short passes consequently a larger part of the plan? Personnel usage is most likely more significant, but the values shed some light.
- Population average at baseline conditions is 7.53.
- Rodgers has the highest average at 8.85. Fitzpatrick has the lowest average at 5.16.
- Population average at extreme conditions is 6.90 (nice).
- Stafford has the highest average at 7.93. Bortles has the lowest average at 4.73.
- Population average delta is a 11.31% decrease in AY/A.
- Palmer has the largest delta at 36.95%. Brees has the smallest delta at 2.04%.
The data suggests a shift in offensive approach for the 5 QBs well below average at extreme conditions. If true, usage data would correlate with these findings.
- 15 of 20 QBs exhibit decrease in AY/A at extreme conditions.
- Correlation study required to understand cause and determine actionable knowledge.
The detailed analysis gives us a better picture. For each QB, it’s clear how weather affects their play on an individual level. It’s also easier to contrast QBs. Not just for the sake of measuring who’s better, but it opens the door to more studies. Personnel usage, positional usage, or throwing mechanics are all now a part of the discussion.
So, Let’s Review
“High winds disrupt the deep passing game.”
Yes, but it affects each QB differently. It’s important to review their track record and matchup. You’re not sitting Aaron Rodgers because of high winds.
“High winds mean less passing and more rush attempts.”
Yes, but again, this is QB specific. More importantly, it doesn’t necessarily equate to lesser fantasy output. “Konami” QBs can aid their day with their legs. Volume is king, but game script can be just as big of a limiting factor.
Each metric, when compared to wind speed, can lead to new conclusions for fantasy owners. Future studies will focus less on answering, “What’s happening?”, and understanding, “Why’s this happening?”. The first will consider efficiency and affected QBs across all three metrics. These QBs include: Brees, Jay Cutler, Carson Palmer, Philip Rivers, Rodgers, and Wilson. Rodgers and Wilson share similar AY/A averages, but Rodgers has more attempts. In extreme conditions, Wilson’s reliance on efficiency takes a hit due to his drop in AY/A and attempts. Future articles on the study will include this and much more.