Week to Week Performance
It’s Sunday morning and you’re doing some last minute research for your daily fantasy lineup. Last week you told yourself you’d set your lineups by Friday so you’d have time to run a few errands and get to the gym, but alas, here you are.
You say to yourself: “Yourself, Nick Chubb sure had a big week last week, I wonder if he can do it again. He was only projected to get 12 points but he ended up with 25!”
Then you have a thought: just because he crushed his projection last week, what does that mean for his expectation this week? Will his projection go up? Will his price in DFS go up? How likely is it that he’ll match his total from last week? How are my competitors thinking about this? Did I remember to pay my credit card bill?
Guess what: we did a bunch of research and have the answers to all those questions! We pulled data for the last 3 years of NFL games, along with Rotowire projections, and ownership percentages from major DFS sites.
There are other factors in week-to-week performance changes such as quality of opponents, injuries, playing time, etc, but those effects should be minimal when averaged out over the large data sample.
Exceeding Point Projections
To measure consistency in exceeding projections, we measured [Points Scored Week X] / [Projected Points Week X], and charted that against [Points Scored Week X + 1] / [Projected Points Week X]. We are looking for consistency in exceeding the original expectation.
What we found was virtually no predictability week to week, which means that a player under- or over-performing in one week had no effect on the next week. In more technical terms, the measured correlation of each player metric from one week to the next was 0.01. There’s a stronger relationship between (insert joke here).
What does this tell us? When building your DFS rosters, look beyond players that merely exceeded expectations in the prior week. Past performance is no guarantee of future success. It may be promising that your running back got more carries or wide receiver got more targets, and there may be specific cases when you can rely on that, but as a whole building a team of over-performers from the prior week will not be a winning strategy.
Changes in Projection
Ok so Mr. Football Player beat his projection last week. What does that mean this week in terms of his projected points? Those nerds must have some complicated algorithms to get the most accurate projections, right?
For all the data points in our sample, we looked at the relationship between projected points from week to week, and player performance compared to their prior week projection. Overall, 67% of the time projections remained within 10% of the projection from the prior week.
But in the cases when the projection did change, over 80% of the time the projection moved in the direction you would expect: over-performers saw their projections increase, under-performers saw their projections decrease. But don’t buy the hype! As we saw in the previous section, overall there is no correlation in week to week performance.
Want a hot tip? In the rare instances where a player underperformed in the prior week by more than 20% and then saw his projection increase (we saw it happen about 60x per year), those players beat their new projection by an average of 18%! So be on the lookout for under-performers who see an increase in their projections.
Change in Ownership
Finally, we wanted to look at changes in ownership in daily fantasy based on all the above factors. After all, if you’re trying to get rich, you can’t just do what everyone else is doing. We pulled ownership percentages from major daily fantasy sites to see what’s going on.
If in the prior week a player met or fell short of expectations, he was just about as likely to see his ownership increase as decrease. However, if a player outperformed his projection by more than 20%, he was 75% more likely to see his ownership percentage increase as decrease.
What does this mean? We love shiny objects. We love the hot new thing. We love keeping up with the Julio Jones’s. We see a player crush it in a prior week and we gotta have him. We can’t miss out on another elite performance!
We all know that going after players with a lot of public interest is never the way to win lots of money. In Jock MKT, if you bid up on prior week top performers, you’re probably not going to do very well.
So what did we learn? Or did you just scroll past all the words and pretty charts?
- Player performance relative to their projection has no correlation from one week to the next.
- Projections move as you would expect based on prior week performance, so be wary of any major changes week to week.
- Ignore the crowd! The masses will target the studs from last week. Don’t be like them.
Be yourself. Be unique. Go win lots of money.