Photo: Christopher Hanewinckel, USA Today Sports

If you don't spend much time on fantasy football Twitter during the offseason, you probably missed a debate that was driving the community apart for a few days. It all started when Dave Kluge asked a relatively innocuous question:

Who would you rather roster in fantasy football?

Running back A: 15 points per game and plays all 17 weeks

or

Running back B: 25 points per game but misses the second half of the season

— Dave Kluge (@DaveKluge) February 22, 2022

It doesn't seem like much, but it drew 36,000 votes, 289 replies, and 115 quote-tweets, including from some of the most popular fantasy analysts on the platform. Much like "the dress" in years past, people saw things one way and were incredulous that anyone else could see it differently.

And much like "the dress" had a correct answer (eventually the woman who posted the picture confirmed what color it really was), this question also has a correct answer. But getting to that answer requires going through several layers of analysis, each of which is a bit counterintuitive. So let's walk through it for those who missed out.

Before we begin, many observed that these hypotheticals closely match the actual seasons from Derrick Henry (8 games at 23.9ppg in PPR scoring) and Ezekiel Elliott (17 games at 14.6ppg in PPR scoring, though one of those games came after championships were decided, so let's use 16 games to keep things simple). Based on this observation, I might refer to running back A as "Elliott" and running back B as "Henry", but this is just a shorthand to help us put a face to the production. Really when I say "Elliott" I mean "players who have a production profile broadly similar to Ezekiel Elliott's in 2021", and the same when I refer to "Henry".

Layer 1: Total Points

This is the surface-level analysis that most new fantasy football players perform. Sixteen games at 15 points per game (or ppg) equals 240 points. Eight games at 25 points per game equals 200 points. The team with the 15ppg running back would have outscored the team with the 25ppg running back. And indeed, with two-thirds of the vote going for the first option, many assumed the voters were performing this type of surface-level analysis and leaving it at that. This is partly why everything took off: this layer of analysis is wildly incomplete. Why? Because of...

Layer 2: Replacement Points

The team with "Derrick Henry" isn't exactly taking a zero in all of the games Henry misses. Instead, he's starting a backup running back. Worst case scenario, he's starting a player off of waivers. Whoever he's starting, that player is getting him points. How many points? Well, the replacement only needs to average 5 points per game for the "Henry" team to match the "Elliott" team at 240 points over the full season, and running backs on waivers typically average substantially more than that. Factoring in replacement points, the right answer would seem to be running back B.

You saw a lot of experienced fantasy analysts making exactly this point and criticizing everyone who voted for running back A in the poll because experienced fantasy analysts have had it drilled in by now that what matters isn't how many points a player scores, but how many points a player scores above a theoretical replacement. That concept is called VBD, and it's no stretch to say that this entire website was founded on the concept more than 20 years ago.

Because the team with "Henry" and a replacement will score more points than the team with "Elliott", that team will likely win more games. So this set up as a classic example of the masses believing one thing but the experts knowing better. Except as I said, there are layers here, and there was another twist to come. Because...

Layer 3: Timing Matters

The goal of fantasy football is not winning the most games. The goal is winning a championship. To that end, performance in the fantasy playoffs is more important than the regular season.

Everyone knows this, but few fantasy analysts pay much attention to it because it's not really an actionable insight. Yes, if you roster a playoff superstar your chances of taking home a title are much higher. But who on earth could have known in advance that this year's playoff superstars would have been Amon-Ra St. Brown and Rashaad Penny? All we can do is grab the best players we can and hope that their best games come at the right time of the year.

But in this particular hypothetical, that's not the case. We know for a fact that the "Derrick Henry" back is going to miss the playoffs because it's stipulated right there in the question. So to really grapple with who is more likely to help you win a title, we have to look at how much "Elliott" boosts your odds in the playoffs.

My general rule of thumb is that each playoff week with byes is worth about 2.5 times as much as a regular-season week, while each playoff week with no byes is worth about 5 times as much as a regular-season week. How I get to this value is a column unto itself, but those are the weights I use, and I tend to feel that if anything they underestimate the relative difference. Based on that, "Elliott" will give more "value" as long as replacement level is below ~10.5 points per game, while "Henry" will give more value if replacement is above that mark.

Usually, if you're adding a running back off the street you'd expect fewer than 10.5 points per game, so that would suggest running back A is the correct answer. Surprise, the masses were right and the analysts were wrong. Right? Well...

Layer 4: Lots of Other Stuff Matters, Too

Here's the actual, honest-to-goodness, 100% scientifically proven correct answer to the question:

IT DEPENDS

Really, that's the truth. Which side is more valuable depends on a ton of different variables. How deep are the rosters in your league? (This has a big impact on how many points you can expect from a running back off of waivers.) What percentage of teams make the playoffs? (Running back B only helps you make the playoffs, so the more teams make it, the less necessary his help becomes.) Are you generally better than average (and therefore more likely to make the playoffs without running back B's help?) or worse than average (where that big early-season boost is more likely to be the difference between making the postseason and missing it entirely?)

How deep are the starting lineups (the shallower the lineups, the more a single difference-maker like Henry would matter)? Does your league play for money, and if so, what's the payout structure (leagues that give a cash bonus for making the playoffs put an added financial incentive to regular season performance)?

I think in general, for most managers in most leagues, the second running back would result in slightly more wins over the course of the season, while the first would result in slightly better title odds. But the differences are likely small and the key features of your league will have a huge impact on which way things lean.

And that's my favorite part about this question. Everyone is so confident about the right answer, but really, the right answer is just "it depends".

For fun, I built out a spreadsheet where you can enter values for several of these variables and it'll estimate which running back gave you better odds at winning a title. Here are my best guesses for the most representative values, the ones that are closest to average across all leagues.

Here are screenshots of the corrected assumptions and corrected results.

Key takeaway here is just that performance in the playoff weeks is really, really, really disproportionately valuable in determining championship odds.

(Though this generally is not an actionable insight.) pic.twitter.com/dqZO2wbJ8E

— Adam Harstad (@AdamHarstad) February 23, 2022

Using my assumptions above, the "Henry" team likely won more games (8.2 vs. 7.6), had a better chance of making the playoffs (71% vs. 60%), a better chance of earning a bye (27% vs. 19%), and a better chance of still being alive in the semifinals. But ultimately the "Elliott" team probably had a better shot at taking home the title (9.9% vs. 9.0%).

But that's just one set of assumptions, most of them pretty rough and untested. If the Derrick Henry team actually has a 46% chance of winning each playoff game instead of a 44% chance, suddenly his title odds are slightly better than the Ezekiel Elliott team's. If you want to play with the spreadsheet just to see how small changes make a huge difference in the end result, you can find it here; just save a copy, edit the yellow cells as you see fit. The second page will dynamically update to calculate how many wins each team will get on average, how likely they are to make the playoffs, how likely they are to earn a bye, and how likely they are to win a title.

It makes intuitive sense that regular-season performance is more important in a league where 33% of teams make the playoffs than one where 66% of teams do. It's interesting to see that intuition in action as we alter the odds of a 7-7 record earning you a playoff berth, for instance.

Anyway, the debate was fun and everyone involved largely had a good time. It's a shame that it's now settled. In the interest of starting a new one: if everyone believed one answer or the other was correct, and instead it turns out that the correct answer is "it's really close and largely depends on league particulars", does that mean that everyone was right... or no one was?