Even though quarterback is by far the highest-scoring position, quarterbacks do not usually get drafted early. One reason is that, relative to other positions, quarterbacks are considered highly replaceable. The position is deep — this thinking goes — and even the bad ones are pretty good. Which is where the Late-Round Quarterback strategy was born via JJ Zachariason and his work on the strategy.
We are all familiar with this commonly held belief about quarterback replaceability. But is it true? To answer this question, let’s begin by taking a look at how the top 12 quarterbacks from 2018 stacked up against one another in one of my PPR leagues last year. In this league, quarterbacks receive one point per completion, negative one point per sack, five yards per point for rushing and receiving yards, and negative two points for fumbles, interceptions, and pick-sixes. All other quarterback scoring is Yahoo default. Under this scoring scheme, here is how the top twelve quarterbacks shook out:
The difference in average fantasy points per game (AFP) between QB12 (Trubisky) and QB1 (Mahomes) is 8.8 points. For comparison, among the top 12 RBs and WRs, this difference was 16.1 and 6 points, respectively. From this angle, running backs are the least replaceable, wide receivers are the most replaceable, and quarterbacks are in the middle as slightly more replaceable than wide receivers.
But the situation is more complicated. Since we usually have to play one QB, two RBs, and three WRs, to measure replaceability across positions we might instead want to consider the range of all the players in each position that will have to get played every week. This means looking at the ranges for 12 QBs, 24 RBs, and 36 WRs. This cashes out to 8.8 for QBs, 21.5 for RBs, and 16.1 for WRs. From this angle, QBs (8.8) are the most replaceable and RBs (21.5) are the least replaceable. I think in general this chimes with most of our intuitions. And we see it reflected in the way the three positions typically get valued.
But the situation is even more complicated. There is yet another way of thinking about replaceability. It’s not quite right to say that we get one QB, two RBs, and three WRs. We could also view it as getting the following: QB1, RB1, RB2, WR1, WR2, and WR3. Here RB1 and RB2 represent running backs 1-12 and 13-24 (in a 12-team league), respectively, and so on with the wide receivers. This provides us with six groupings for which we can compute replaceability. Here they are in table form:
Tables are nice, but I find it easier to compare the replaceability of these positional groupings via a simple line plot. Like this:
The first takeaway is that getting a top RB1 is essential. In these data (2018, PPR) there are over 16 points a week up for grabs. I doubt anyone will find this very surprising because RBs dominate the first round of most drafts.
In contrast, the second takeaway is pretty surprising. From this graph, it is clear that quarterbacks are not as replaceable as many people think. The graph is telling us that getting a top QB1 is the second most important move you can make. Here there are almost 9 points up for grabs.
Lastly, the third takeaway is also interesting. None of the other positional groupings matter much from the perspective of replaceability. Indeed, with less than three points per week on the line, WR3s are so replaceable that choosing one is basically a low stakes dart throw.
So the strategic basis for drafting a quarterback early (but not before securing a top RB1) might be stronger than a lot of people realize. Maybe. Let’s go even deeper.
The line plot above shows us the AFP ranges for each positional grouping. Note, though, that the criterion for grouping the players into sets of 12 is a somewhat arbitrary artifact of the fact that we’re talking about a 12-person league. Will our perspective on positional replaceability change if consider AFP by position without artificially grouping the positions? To consider this, let’s look at all of the 2018 PPR “starters”, which I’ll define as the top 12 QBs, 24 RBs, 36 WRs, and 12 TEs. Like this:
The key here is to look at the top eight running backs (the topmost cluster of yellow dots) and next cluster of running backs (beginning with James White just before the 25 AFP line). This explains why the replaceability of RB1s seemed so low. The real story is that running backs fall into two clusters, an elite group of 8 and the rest of the pack. Limiting our RB1 set to these eight brings the replaceability to 7.4, which is more in line with the other groupings. And guess what that means?
It means that quarterback is the least replaceable position on your roster. Another way of saying this is that, in PPR last year, having a top QB1 was more valuable than having a top RB1 or WR1. One fact which tempers this conclusion is that the real RB1 cluster only had 8 members, making it a scarcer group to access.
I’ll conclude with a few additional takeaways.
First, it’s hard to justify taking a non-RB in the first round of a snake PPR draft. Every non-RB that goes off the board in the first ten or so picks is a gift.
Second, understanding the outsized value of quarterbacks doesn’t mean you can draft early willy nilly. The more picks you can make before the top quarterbacks start moving the better — and this will vary with each league. If your league is sophisticated, and everyone is well versed in late-QB strategic thinking, then you can wait. Personally, I would never take the first QB because, historically, top performers rarely repeat. But I do want one of the first five or so. If I miss getting a top-five quarterback from being on the wrong end of the draft order during a run, then I pivot to a late-QB strategy.
Third, the Konami Code which was made famous by Lord Reebs (the idea that mobile quarterbacks are like a cheat code because of the extra points they score from running yards and touchdowns) is much weaker in PPR. For quarterbacks, there is a clear trade-off between running more and passing more. And among the elites, it seems that the points for completions outweigh the points from running. As we can see in the graphs, only two scramblers make the QB1 cut: Newton at #5 and Watson at #7.
Making matters worse, the lower-scoring running approach also entails greater injury risk. Both Newton and Watson got injured last season, and their rankings would have been even lower had I not computed their value as an average per game played. Finally, scrambling entails more sacks and turnovers. In leagues that penalize these — like the one from which these numbers are taken — the difference will be greater. In short: all things equal, fade running quarterbacks in PPR.
Fourth, the Big 3 tight ends clocked in right about where the high-end WR2s meet the low-end WR1s. This presents a tricky situation. Consider the case of Kelce. In his top 150, Evan Silva puts Kelce at #18 overall, sandwiched between JuJu Smith-Schuster (Silva’s WR6) and Antonio Brown (his WR7). Last year Kelce would have scored as the WR10. What are the tradeoffs?
Obviously, a #6 is better than an effective #10, but they are so close that it probably doesn’t matter. However, in this case, an effective #10 probably represents the ceiling of the TE position, whereas most top ten WRs have a plausible path to a top WR finish. So from this angle, the WR option has more upside.
On the flip side, let’s assume that taking a WR here entails accepting a low-end TE later, and, conversely, that taking the top TE here entails taking a low-end WR3 later. To see how this played out last year we can look to the lower left-hand corner of the last graph. There you will see the lowest-scoring starters, 7 TEs represented by blue dots, attached to the next lowest scoring cluster of about 12 WRs, represented by green dots. The TEs’ AFPs only range 2.1 points; the WRs’ AFPs only range 2.5 points; the whole group’s AFPs only range 5.4 points in total.* Thus, the stakes are pretty low.
Now suppose your WR selection actually becomes the top WR this year. What does that get you? 3.5 points. The stakes are low once again. By my lights, then, this is a wash and it doesn’t really matter. I’m letting the higher injury risks of TEs be the dealbreaker, and I’m passing on the Big 3 in this situation.
Finally, let’s just reiterate the big takeaway here. Quarterbacks are probably not as replaceable as you think. So don’t miss the opportunity to draft one early if you find yourself in the right position to do so.
*For those paying attention, these numbers don’t add up because for clarity I excluded Gronk in the previous calculations because he’s not really part of either group.