What Have We Learned Thus Far?
*LT regularly compiled over 370 touches per season as a Charger and yet somehow avoided the dreaded Curse.
The apps on this site should fall generally into two categories. Some apps, like the Custom Similarity Scores tool will be aimed at giving you actionable fantasy intelligence. But some apps, like the Curse of the 370 app, are meant to do something a little less straightforward. They’re meant to teach something which can only be learned. If I throw a graph up on the site and say “See, no curse!” it will have less staying power than if I throw up an app and let you figure it out for yourself. So that’s why I say they’re meant to teach something which can only be learned.
Having said that, some people are lazy and would rather just have things spelled out for them and I can’t say I really blame them. So from time to time I’ll come back and try to summarize things.
So what have we learned thus far?
The last half of the season doesn’t matter the most in terms of explaining next year fantasy results (for wide receivers anyway). If you want the best shot at predicting next year fantasy results for receivers, you’re better off looking at a full season dataset. The last half of the season does explain more than the first half of the season, so among season splits it’s the most explanatory, but still the full season is probably the best dataset to look at. This even holds true if we’re trying to predict the performance of younger players. But looking at younger players also unearthed another trend that should give us pause when trying to pick the next breakout star receiver. Let’s take a look at the r^2 when we try to use prior year fantasy performance to predict next year performance for receivers up to age 25.
The r^2 is .33
Now let’s look at a similar sized group of receivers but at a different point in their career. This is what happens when we filter by age 26-31. You can see that the r^2 improves dramatically.
I can imagine a number of explanations for the fact that prior year results tend to explain more of next year results for the receivers that we would regard as being in their prime. The younger receivers are probably less entrenched members of their own offenses. Younger receivers that perform really well might be playing in offenses where there aren’t very many good options, which is to say that the offense itself might be bad and not very reliable. Older receivers have probably learned a good amount about being a professional and keeping their bodies in good condition. There might also be survivor bias at play here. It could be the case that reliable receivers are the ones that stick around past age 25.
The Curse of the 370 May or May Not Exist
It’s possible to look at running backs who carry the ball 370 times and come away feeling like there must be diminishing returns for those running backs at some point. For instance, look at the following graph which shows Season N carries on the X axis and Season N+1 Fantasy Points on the Y axis. It looks like at 350 carries, running backs start to experience diminishing returns.
But there are a couple of problems here. First, you can see that the number of observations greater than 370 aren’t very numerous. That means we have to add randomness as a potential explanation for the so-called “Curse of 370”. The other problem is that this graph only includes carries. What if we include receptions also? This is what the graph looks like:
You can see that the peak at 350 carries disappears, the peak shifts to about 390 total touches, and the drop-off becomes much less severe. To make things even more problematic for the “Curse” is the fact that if we look at just players aged 25 and older, the peak/drop disappears entirely.
The Curse of the 370 may or may not be related to something that happens when running backs compile a lot of usage in one season. But I would probably be skeptical of the curse given what happens when we look at total touches. This might all be academic going forward though as Arian Foster might be one of the last tests of the Curse. Foster’s 460 combined touches (including the playoffs) seem to be an outlier in a league moving towards more and more running back timeshares.