What actually predicts a running back's next season

Fantasy running back rankings are mostly bad because they project from the wrong things. Last year's fantasy points. Touchdowns. Highlight runs. None of those repeat well. We built our RB projections around the five signals that actually predict next-year production, weighted by how repeatable each one is. This is what they are and why they matter.

Signal 1: opportunity share

Opportunity share = (carries + targets) ÷ team offensive plays. It's the single most repeatable RB stat we measured — better than yards per touch, better than fantasy points, better than touchdowns. RB fantasy is a volume game, and dividing by team plays removes the pace confound (slow-tempo teams look worse without it).

We scale projections from −20% to +15% based on where each RB lands in this distribution.

Signal 2: schedule of run defenses

Every week, our model looks at the opposing run defense's rank. The aggregate of those weeks — across all 17 games of the projection season — is each RB's schedule strength. Most ranking sites quote a generic team SoS that mixes pass and run defenses; that's noise for RBs, who only care about run defenses.

Bars are projection multipliers as a percent of average. ±5% is small compared to opportunity share but the gap between the easiest and toughest schedules adds up to about a full draft round of fantasy value.

Signal 3: the RB age cliff is real

The published curve is unambiguous: RB production is flat through age 25, drops 7% at 27, 14% at 28, 22% at 29, then 32% at 30.

Our previous projection used a linear 4% per year after 28. The real curve isn't linear — there's almost no decline from 24 to 26, and the bottom falls out at 29-30. The model now uses the real curve.

Signal 4: committees crush ceilings

Two RBs on the same team each getting 30%+ of the snaps does not average to one RB1. It produces two RB3s. We checked this against ten years of data: the median fantasy finish for an RB in a confirmed committee is RB28; for an equivalent solo back it's RB14.

When our pipeline detects two RBs on the same team with similar touch counts in the historical data, both get a 12% projection haircut. The model isn't predicting one of them will lose the job — it's pricing the share uncertainty into both.

Signal 5: who gets the ball at the goal line

Carries inside the 10 yardline score touchdowns at roughly 5× the rate of mid-field carries. A back who owns the goal-line role has higher TD upside than his volume share alone would suggest; a back stuck behind a short-yardage specialist has lower TD upside.

This adjustment is small — ±5% — because goal-line work is volatile year to year. It's more of a tiebreaker between two RBs with similar opportunity share than a primary signal.

How we combine them

Each RB's projection is the credibility-weighted points pace from last season, multiplied by all five adjustments and clipped to a sane range so no single signal can swing a projection more than ±30%. The age curve and opportunity share do most of the work; the other three break ties between players who look similar on volume.

What to do at the draft

1. Trust opportunity share over fantasy points. A high-volume back on a bad team beats a low-volume back on a good team almost every time.
2. Discount age-29 RBs. The cliff is the most reliable pattern in fantasy. Be the one who fades it, not the one who pays for it.
3. Avoid confirmed committees in the first three rounds. Take the workhorse in a worse offense over the lead back in a 60/40 split.
4. Buy goal-line work cheap. Backs whose only job is the 1-yard hammer regularly outscore expectation. They're terrible to roster but great to stream during a TD-friendly stretch.