This is a time of year where there’s constant chatter about stacking. It’s peak best ball season. The Scott Fish Bowl has recently been drafted. Home redraft leagues are about to kickoff drafting. The word correlation is being thrown around left and right. In a dynasty best-ball auction startup this offseason I was desperate to get some stacks on my team (All I ended up with was Tua/Hill/Gesicki and Geno/Lockett, thanks for asking). I had listened to the podcasts. I had read the articles about correlation which stacking is built on. And it all made so much sense? For those wondering what the idea being correlation is?
Quick summary – when quarterbacks score well, so do their pass catchers. A very obvious, simple idea really. However enticing the idea of stacking is, I’m still yet to see clear evidence in a well put together study of its clear advantages over not stacking. Even in best ball and season long redraft leagues (please send links if you believe there is clear evidence out there!). Even so, it didn’t appear there had been the same research done in dynasty leagues, and if it made a difference to all play win percentage.
I wanted to keep the actual analysis simple. My two questions:
- Did teams who had a stack end up with a higher all play win percentage than those that didn’t stack?
- Did the presence of multiple stacks result in a correlated increase in all play win percentage?
With help from code that @EdnaBEASTmode used in their recent article I was able to classify rosters that had single, double, or triple or more stacks on their roster, and how many of each they had.
Single stack = QB and 1 x WR or TE from the same team
Double stack = QB and 2 x WR or TE from the same team
Triple stack = QB and 3 x WR or TE from the same team
To ensure we weren’t including any stacks with backup QBs I limited the QBs to those that started 7 or more games. This number just came from the number of starts Jameis had before his injury.
The Results
2,557 dynasty teams across 2020 and 2021 from Sleeper were pulled, as were their league standings. The leagues were non-bestball leagues.
Answering Question 1
Below are three charts simply comparing all play win percentage for teams who had one or more single stack, double stack, or triple or more stack on their roster. As you can see, no difference. I’ve also included the results from statistical analysis of the data behind those charts. A p-value of less than 0.05 indicates a statistically significant difference in values. None of these are below that number, or even close.
Answering Question 2
Below are three scatter plots with trendlines. The X axis is the number of single, double, or triple stacks on a roster. The Y axis is all play win percentage. The trend lines? Flat as a tack. No clear correlation.
Conclusion
There’s multiple ways to skin a cat, and different approaches to the analysis of a problem. In this case, stacking, may result in slightly different results or conclusions. However, given with this analysis there doesn’t even appear to be a trend towards stacking being beneficial in relation to all play win percentage in dynasty leagues. I’m fairly confident that my main strategy going forward is still going to be pick the players who I think will score the most. Except RBs, eff that position.
Link to code: GitHub