Handcuffing: How Roster Construction Should Handle Backup Running Backs

Handcuffing is the practice of spending a late pick on your own starting running back's direct backup. The reflex version treats that pick as insurance bought on principle: one handcuff per stud, filed under obvious, never examined. A projection model with distribution bands can do better than principle, because the question underneath has an actual number attached.

That number is how much of the starter's projected value moves to the backup when the starter's season ends. Call it the transfer rate. It runs from near zero to near one, and what moves it is whether the vacated work lands on one player or gets divided.

What The Roster Spot Is Actually Buying

A backup running back's projection is two projections wearing one name. There is the standing-role world, where he takes a change-of-pace snap share and finishes below every startable back on your bench. There is the succession world, where the starter's carries, snaps, and goal-line work land on him and his median jumps into flex territory.

The board shows you the blend of those two worlds, weighted by how likely the model thinks each one is. The blend is the honest summary of the player and it is close to useless for this specific decision, because you never hold a handcuff through the blend. You hold him until the branch resolves, and the resolution decides whether he starts every week or gets cut.

That asymmetry is the entire case for the roster spot. A bench slot on a handcuff is an option: the downside is one waiver flier you didn't roster, and the upside is a weekly starter acquired at a last-rounds price. Options price off the good branch and the odds of reaching it, which means the object that matters is the upper region of the backup's distribution. The bands behind every median are where that region gets described.

Transfer Rate And Where It Breaks

Transfer rate is the succession-branch median divided by the starter's median. A bell-cow backfield with no third back and no passing-down specialist transfers high, because the vacated touches have one place to go. A three-man committee transfers low, because the vacancy gets split before it ever reaches your guy.

The rate on its own will mislead you, and the way it misleads is worth watching. A low-volume starter can produce a high transfer rate — his backup inherits nearly all of a small pie — while the inherited workload still lands under the free streamer you would have added anyway.

The number that survives contact is the succession-branch median measured against replacement level, the same yardstick every other pick on your board is priced against. That pricing argument is made in full on the value-based drafting page; assume it here.

An Invented Stretch Of Board

Take an invented stretch of running back board — every number below is made up to show the shape, none of it from a live run. Three starters, each with one direct backup, all medians in points per game.

Starter A projects 15.0 with nothing on the depth chart behind his handcuff, and the model's succession branch puts that backup at 10.5. Starter B projects 13.0 at the front of a three-man committee, and his backup's succession branch reads 5.7. Starter C projects 10.5 as a low-volume grinder in a two-man room, and his backup's succession branch reads 8.3.

Rank those by transfer rate and C comes out on top: 8.3 over 10.5 is 0.79, against 0.70 for A and 0.44 for B. Draft off that ranking and you have bought the second-best handcuff on the list, and the rate never tells you which one you missed.

Add the baseline. Set the invented replacement level at 7.0 per game — the back you can add off the wire in a twelve-team league during the week the injury happens. A's backup clears that by 3.5 in the succession branch, C's clears it by 1.3, and B's comes in 1.3 under water, which gives B's handcuff negative value in the exact branch it was drafted for.

Weight the branches. Give each starter an invented 20 percent chance of losing the back half of the season, with nine games left to play when it happens. A's handcuff is worth 0.20 times 3.5 points times 9 games, or about 6.3 points of surplus over the free alternative. C's comes to about 2.3, and B's is worth less than nothing, since no probability rescues a negative.

Read the two numbers in their own units. In the season where A's starter actually goes down, the weekly edge is the full 3.5 points a game; the 6.3 is what a one-in-five chance at that season is worth before anyone gets hurt. C's 2.3 rides the same odds on a much thinner branch edge, and B's negative needs no interpretation. The arithmetic sorted three players who looked interchangeable in the pre-draft chatter, and it sorted them on inputs the chatter almost never mentions: the size of the starter's pie, and how many bodies stand between it and the backup.

The Threshold That Earns A Bench Slot

Two conditions, both of them on the same player. The succession-branch median clears replacement level by roughly three points per game, and the backfield behind the starter is two-deep or thinner.

Both cutoffs are judgment. The three-point line is calibrated to the size of a weekly lineup edge worth a roster spot, the depth line to where committees start splitting the branch, and neither has a backtest standing behind it, which is worth saying out loud every time the numbers appear — including here. The depth condition is at least the one you can verify in August: count the bodies who have taken meaningful carries or passing-down snaps, and count anyone with a defined sub-role as a separate claimant on the vacancy. Two-deep means the starter and his handcuff with no third claimant standing there; a second claimant splits the branch before it reaches your guy.

The option also expires. A handcuff you are still carrying in the back half of the season behind a healthy starter is a bench slot paying rent, and that slot belongs on a live role off the wire.

This is the part of the board our model grades worst, and the reason is structural. The accuracy ledger scores medians against what actually happened, and a succession branch describes a season that mostly did not occur, so there is no clean scoreboard for a conditional that never resolved. Read the public accuracy record knowing it grades the medians hard and the branches barely at all.

How To Act On It In A Draft

Your build sets the budget. A robust-RB roster carrying three early backs has the least bench room and the least need for its own handcuffs, while a zero-RB roster wants them most and has the fewest slots to spend on them. The build shapes are argued through on the zero-RB, hero-RB and robust-RB page.

League size moves the baseline, and the baseline moves every threshold above it. A fourteen-team room does this: the wire back you were treating as free is already rostered, the invented 7.0 replacement level drops toward an equally invented 5.0, and handcuffs that failed the test in a twelve-team league clear it comfortably. Deeper-league draft mechanics have their own page.

Where any of this sits on a current run is a live question, and prose is the wrong surface for live numbers. Pull up the PPR draft board, find the backups sitting behind thin backfields, and check the spread between their upper band and the last startable back.

This page is the bench-slot slice of a draft plan, and it binds only after the build and the format are settled. Read the record first, then decide how hard to lean on any of it, including this piece.