How to target the end of your drafts

Every draft guide tells you the bench is where leagues are won. We went and checked, and for about a third of it the opposite is true: those players were worth less than whatever was sitting on the waiver wire, every week, for free. Now, to be clear, we are NOT saying that the bench is not important, just not in the way you might think.

Let's talk about it. We looked at the draft board for each of the last six seasons, using only what was known that August. Then we asked, for every player: across the whole season, did that roster spot give you more than streaming the position would have? A roster spot is not worth the player's points on their own. Meaning, if that player misses four games you would stream somebody for those four weeks, so the spot is worth that players production plus the replacement's for the weeks they are out. Compare that against the alternative of simply streaming the whole season and you get what that spot was worth over doing nothing.

Through the first four rounds it is worth a great deal (duh). The top 24 players returned about 93 points above what streaming would have given you, the next 24 about 65, the next 24 about 44. Turns out the top of your draft is pretty important. Who knew!

Where it evens out

Then it keeps falling, and it does not stop at simply breaking even. Around the 120th to 145th player on the board (the exact spot depends on your leagues scoring format) the average roster spot stops being worth anything at all. Past that it goes negative: the last two dozen players rostered in a 12-team, 15-round league came in about 14 points below what streaming would have handed you.

Run that across the whole league and 38% of rostered players were worse than nothing. In a 12-team, 15-round league that is roughly 68 of the 180 players taken. Not 68 bad picks — 68 spots where a manager who dropped the player and streamed the position instead would have finished ahead. So, that means roughly 6 players on that 15-round league are essentially useless.

An important note here is that number is the same in PPR and half-PPR, which is why we trust it. The crossing point moves a little between formats and the share does not.

Of course, some of that is unavoidable

You have to draft fifteen players. Byes are real, injuries are real, and you cannot run a season on nine men and the wire. Sometimes a random waiver wire player you would never consider starting blows up, and some fraction of the bench is insurance you are correctly paying for even when it never pays out.

But 38% is a lot more than bye-week cover or simple randomness, and the shape of it points somewhere specific. The value does not fall off a cliff at the end of the draft — it declines smoothly from round one and crosses zero with three or four rounds still to go. Which means the rounds where most managers are being clever are the rounds that, on average, produce nothing.

What we would actually do with this

You want to aim for upside with your late round picks. Take a chance on ambiguous backfields or receiver rooms, and make quick decisions after one or two weeks. We care less about actual production, and more-so who was actually on the field. After Week 1 last year it was pretty clear that Javonte Williams was going to lead.

So, who does that lead us to this year? Six going after round 9, three at each position.

Matthew Golden (GB, round 11). Green Bay lost two receivers from last season in Romeo Doubs and Dontayvion Wicks, Tucker Kraft is coming off a torn ACL, and the man nominally ahead of him (Christian Watson) has missed a third of his career. Green Bay is an offense we like, and he is going 128th.

Cyrus Allen (KC, round 15). Normally, late-drafted rookie WRs wouldn't show up on this list, but the Chiefs are an interesting receiving room. Rashee Rice has had about as bad an off-season as it gets, and the only other person possibly ahead of him is a turning-37 year old Travis Kelce.

Tre Harris (LAC, round 20). No Keenan Allen, and if the Chargers still only put Ladd McConkey on the field in three receiver sets, we could see Tre Harris take a bunch of snaps. From there, all we have to do is fade Quentin Johnston? It's not that crazy.

Jonathon Brooks (CAR, round 10). I mean, duh. This one is low-hanging fruit after the injury to Chuba Hubbard. That said, in some home leagues Brooks is still going later in drafts, and you should snatch him up.

Jordan Mason (MIN, round 10). Perpetually underrated Jordan Mason. We are hoping better than last years RB35, but the Vikings have made some offensive changes that supposedly are going to benefit Mason, and at the end of the day he's a possible starting RB in round 10.

Chris Rodriguez Jr. (JAX, round 12). If Rodriguez gets the goal-line work for this Jaguars offense he will be a value, and that still leaves room for opportunity if he gets more than that.

The most common way to waste a roster spot is holding a player you drafted simply because you drafted them. The measurement says the marginal man on your bench is, on average, worse than a stranger on the wire — so the bar for keeping him should be that he is doing something or on the field, not that you loved him pre-draft.

Lastly, let's talk about handcuffs. A popular draft strategy is to take your handcuffs, but we advise against that. Most handcuffs are not drafted, and only valuable with an injury. Draft someone with upside in August, and then wait until October to pick up your handcuffs off waivers. If you have to pay a little extra FAAB, it's worth it.

You can see where this cuts on your own roster. The rankings show what the board thinks of everyone you drafted, and the waiver page prices what is actually available against your own starting lineup rather than in the abstract — which is the same arithmetic this piece is built on, applied one player at a time.

Measured over six seasons and 3,403 player-seasons, half-PPR and PPR, scoring every player the board projected rather than only the ones who stayed healthy. Players whose seasons were destroyed by injury are the outcome a bench spot is a bet against, so dropping them would score the bet only where it won.