The website is unlocked and in off-season mode. The 2023 season is displayed, and you can explore all the site features freely!

Pairing/Stacking Analysis: Correlations Between the Different Fantasy Positions

In extension of my "QB-Kicker Pairing" analysis, here is a look at all the fantasy positions.

Why is this interesting?

Especially for critical matchups, people often ask about the effects of using players from the same team.  It is pretty well known that a "QB-WR" stack can have an amplifying effect, as they key example. What about stacking the other positions? Is there a strategy around combining DST-Kicker?

My approach to the analysis

I'm laying out a similar thought process as I did in-depth for QB+K, but treating that as background, I will simplify the explanation.

(1) Look at the first correlation table, showing how the actual fantasy point results correlate with each other.  All positions are "total team score", so "RB" represents all the running back scoring on a given team for that week.

Re-cap: Why is this not good enough?  If you used this first table alone, you could be misled in your line-up considerations.  This is because you should first and foremost choose players with higher point projections.  The table only shows how results correlate, not how many points are scored. (You don’t want to choose things that correlate if it’s driven by low scoring.)

(2) So let's look at the middle table: the correlations between projections.  What you can see here is that all positions are expected to move in concert.  I.e. if a team is expected to score well, then all positions are expected to do well.  That is why they are all shaded blue.  So seen in advance, interestingly, no positions are expected to "steal points" from each other.  That is just something that happens in course of the games.  Which leads us to...._

(3) The right-hand table represents how much the fantasy points change relative to what was expected.  To my surprise, it actually looks a lot like the first chart.  But at least we know we're doing things rigorously enough to make conclusions.

Of note: 

  • QB-WR correlates most, as expected.  

  • QB-TE is second.  

  • QB-K is a small but positive correlation-- in agreement with my earlier analysis.  

  • Next runner-up is K-DST, which apparently move together (lower floor, higher ceiling).  

  • Negative correlation between RB-WR (stealing TDs from each other)

  • To my surprise, WR-DST with a negative correlation.  I presume related to increased passing when a defense does poorly?  (Garbage time?)  Otherwise it has to be related to pace of game and defensive tiring.

Making a more informative Conclusion: look at Points instead of correlations

Correlations don't tell the whole story, and they don't really help your decision-making.  What is more useful is to know "how many points change can we expect, from pairing?"

To see this, click on the second lone chart.  Notice it is no longer symmetrical, as with the correlation charts.  You are now looking at the number of points change that can be expected when another position (from the same team) does better-or-worse than expected.

Of note:

  • WR points have the highest dependency, with most sensitivity to QB performance.

  • QB performance also is affected by WR performance, but not quite to the same degree.

  • QB performance is affected by TE performance, to the same extent as by WR performance.

  • Aside from these 3 examples, all point-change dependencies are less than 0.4 points.   Implying that in general you should not put too much more thought in trying to make clever pairs!

  • There is a modest level of point-movement in picking the D/ST to go with your Kicker.

I would really love to conclude for you that there is a really interesting strategy to be had from pairing  (that's why I did this!).  But in general there is not so much apparent gains from other pairing.  QB-WR and QB-TE stacks have the most effect.  DST-K and WR-DST might be worth a second look, but otherwise: go for the most points!