Looking Deep Into The Relationship Of Runs and Fantasy Points
Vegas lines are at the heart of my own daily fantasy models. Sportsbooks do a really good job of creating accurate lines, so good strategy mostly involves following their lead. The difficult job we have to do is figure out how to interpret those lines. How do the run projections of various teams convert to actual fantasy point production?
Researching the relationship between runs and fantasy points can be done fairly easily in R. There is a package in R called “retrosheet,” which was created by a company that supplies game logs and play by play data all the way back to the turn of the century. You can read more about Retrosheet at their website. With Retrosheet data easily accessible in R, we can analyze a lot of data to see how runs relate to actual fantasy point output.
For this analysis, because it involves fantasy points per run, I removed all games where either the home or visiting team scored 0 runs. The data I’m looking at is from 2010-2014. I’m going to use FanDuel fantasy points for this study, but did not include stolen bases in the fantasy point calculation.
The first question that we can answer is how many fantasy points we can expect per run scored. We can look separately for home teams and visiting teams and see if there is a difference. VFPR is visiting fantasy points per run while HFPR is home fantasy points per run.
The home team has slightly more fantasy points per run here, but the number is so small it could easily just be variance. Overall, the fantasy points we can expect per run is about 4.63.
Now how well do runs predict fantasy points? We can find that out by looking at the correlation between total fantasy point output and total runs in a game.
> cor(rtd$TFP, rtd$truns)
Pretty freaking high, a nearly perfect correlation. For all intents and purposes, we can say that runs equals fantasy points. So the higher amount of runs Vegas projects a team to score, the more fantasy points we should expect for that team.
Does fantasy points per run change depending on the run output? A thought I had when I first started playing daily fantasy was that fantasy points per run should increase as run output increases. In other words, is the fantasy points per run in a high run game over 4.63, and in a low run game less than 4.63?
The data below is fantasy points per run output depending on the score. The first entry is fantasy points per run when a team scores 1 run, the second entry is when a team scores 2 runs, the third entry is when a team scores 3 runs, and so on, all the way up to 10 runs.
 5.478067, 4.914304, 4.638244, 4.545431, 4.384771, 4.305757, 4.224895, 4.156250, 4.163960, 4.106199
It turns out my initial thought was quite wrong. As run total increases, fantasy points per run decreases. It’s not by much though, especially when we consider the normal range of run projections by Vegas spreads is 3 to 5 runs. Overall, we should only expect a fraction more fantasy points per run in a low over/under game than a high one, and of course total fantasy point output will still be much higher in a high over/under game. But what this finding does show us is that high over/under games are not magical beacons of fantasy points, nor are low over/under games fantasy death traps.
Let me know in the comments if there’s any other research you would like to see on the runs and fantasy points relationship.
View all posts by Daniel Steinberg