The Fantasy Architect: How to ‘Fix’ QB Scoring
Quarterback scoring is broken!
It’s a refrain you’ll hear from many a fantasy manager. The gripe is, in theory, simple. Fantasy quarterbacks get 0.04 points per passing yard, or 1 point for every 25 yards. However, they get 0.1 points per rushing yard, in addition to a bonus per rushing touchdown (+1 or +2 depending on how many points a passing TD is worth in your league). This discrepancy is intended to keep things balanced between QBs and RBs/WRs. Since QBs generate so much more pass yardage than RBs generate rushing yardage, the ‘points multiplier’ is smaller to keep the overall fantasy score relatively similar.
Recently, though, mobile QBs took off (pun intended). Rushing has become a much more crucial part of the position in the past five years, and it has unlocked a sort of fantasy ‘cheat code’ for QBs that have the ability to scramble. This has led to some monster seasons: Josh Allen carried the ball for 763 yards and 6 scores this year on his way to the overall QB1 finish, and Lamar Jackson broke all sorts of records with 1206 and 7 during the 2019 season.
However, this ‘scoring bump’ has some other unintended consequences. Jason puts it like this: it doesn’t matter in the actual NFL if a quarterback runs or throws for 10 yards. In fantasy, though, the 10 yards of rushing is rewarded more, specifically +0.6 points more. This leads to quarterbacks that have a lot of fantasy success without actually replicating that success in real life football (and vice versa). Jalen Hurts is a good example: his 782 yards on the ground and 10 rushing TDs ensured that he was a top-10 QB in fantasy. However, most football fans outside of Philadelphia could agree that his in-game performance did not warrant a top-10 designation.
Now, you might not actually consider this a problem. I personally don’t: I think it’s ok for relative NFL skill to not map perfectly to relative fantasy football skill. As long as the rules are clear – rushing is worth more, which gives mobile QBs a big boost – fantasy managers can adapt and adjust accordingly. Still, I recognize the value of fantasy scoring reflecting actual football performance, and understand that our game has to adapt as the NFL evolves. In this article, then, we’re going to explore what commissioners can do to address this scoring ‘issue’ and what it means for leagues. All data is from nflfastR.
Expected Points Added
A good place to start is obtaining a metric for actual NFL skill, and a natural choice is the Expected Points Added variable from nflfastR, or EPA for short. This is a number that is calculated for every play, and means exactly what it sounds like: based on the outcome of the play, how much did the ‘projected’ points that the offense will score on this drive change? If a QB makes a good play – throws or runs the ball down the field – his team has a higher likelihood of scoring on that drive, and thus he has positive expected points added. The reverse is true for a ‘bad’ play: incompletions, loss of yardage or turnovers.
Looking at the EPA produced per game, then, can give us a reasonable measure of QB skill. Remember, QBs can generate EPA on passes or runs, both of which have an identical impact (a 15-yard pass and 15-yard run will have the same EPA). Let’s see how per-game EPA and per-game fantasy points compare:
There’s certainly a positive correlation here, which makes sense. QBs that are productive in real life (high EPA) are generally better in fantasy. However, there are some notable outliers. Kyler Murray sticks out as a player far above the trend line: his EPA is slightly better than average, but his (per game) fantasy performance is elite. Jalen Hurts and Josh Allen fit in this camp as well, while a player like Mac Jones is the opposite: average EPA but terrible per game fantasy performance. The missing piece, of course, is rushing. Kyler, Josh and Jalen all had very productive rushing seasons, which makes their fantasy finish much better than their ‘real life’ performance, and vice versa for Mac.
We can summarize these results by looking at the QBs who have the largest difference between their fantasy points ranking and their EPA ranking. On the left, we see players that are much better in fantasy: Kyler was 4th in fantasy scoring but 13th in EPA. On the right, we have the opposite: players who were great on the field but much less impactful for fantasy. Teddy Bridgewater leads the way: 9th in per-game EPA but 21st in fantasy PPG.
What can we do? Probably the most natural solution is to reflect ‘real life’, where we make rushing yards and touchdowns the same amount of fantasy points as their passing counterparts (.025 points per yard and 4 points per TD). This will mean a lower overall fantasy score for all players but, naturally, a much lower score for some players. Here are the players that would be most affected by this change, based on how much their fantasy ranking moves. For example, Matt Stafford moves up 5 spots in fantasy PPG, while Tyrod Taylor moves down 6 spots.
Our main query, though, is if this scoring change more accurately reflects real life. Here’s the same chart as above after the scoring change:
We still see plenty of big discrepancies, but the broader picture seems to have improved slightly. Kyler’s number was -9 before, now it’s -5. Teddy Bridgewater and Mac Jones both improved by one slot. Overall, before the scoring change, the median distance between the fantasy points and EPA rank was 4.0, and now it’s 3.5 (a 12.5% improvement). The improvement is even better among the top 15 or so QBs (3.5 median to 2.0 median).
The mapping is clearly still not perfect. That’s because factors other than rushing scores create differences between EPA and fantasy points. For example, 10 yards on a first down might have a different EPA than 10 yards on a third down, because in the latter case there was a lower probability that the drive would continue. Field position matters too: 10 yards from your own 20 has a lower impact on TD probability than does 10 yards from your opponent’s 20. In this case, fantasy scoring doesn’t care, but EPA does!
This is how we get cases like Jared Goff, who had a much better per-game fantasy performance than actual EPA performance. It’s not because of rushing: Goff had just 87 yards on the ground (and no scores) last year. Instead, it’s likely that Goff is racking up fantasy points and yardage in situations where the EPA does not increase by a commensurate amount. This is an entirely different source of discrepancy that is much more difficult to address. It would be something along the lines of making ‘garbage time’ yardage from your own territory be worth less fantasy points, but that opens a whole other can of worms. Perhaps in another article…
Equating rushing and passing yards does decrease the difference between EPA (‘real life’) ranking and fantasy ranking. Charting the two shows generally tighter clustering around the trend line, although outliers persist. The most notable shakeups are higher-ranked players. Skilled rushers Kyler Murray, Jalen Hurts, Josh Allen and Lamar Jackson all see a dip back closer to the trend line, whereas less mobile players like Teddy Bridgewater, Jameis Winston and Jimmy Garoppolo see a bump in fantasy scoring that moves them closer to reflecting actual on-field performance.
It’s up to you and your league to decide what scoring system to use, and hopefully this article gave you a better sense of what happens when you try to make QB scoring more realistic.
Questions? Let me hear it on Twitter.