Returning Production: How Much Does It Really Predict? A Data Check

"Team X returns 80% of its production." You'll see that line every July, usually wielded as proof that a team is about to break out or about to collapse, as if the percentage settled the matter. It doesn't. Returning production is a real signal — I use it — but it's also one of the most over-interpreted numbers of the preseason, and the gap between what it predicts and what fans want it to predict is where most of the bad takes live. So let me draw that line carefully.

What "returning production" means

It's not just counting returning starters. The modern version, popularized by Bill Connelly, weights players by how much they actually contributed — usually measured in expected points added or a similar per-play value — and asks what share of that value is coming back. Returning a backup who played 50 snaps isn't the same as returning a star quarterback who touched the ball every down. Production-weighting captures that; "returning starters" doesn't.

It's typically split by side of the ball, because offense and defense behave differently. And one position dominates the offensive figure: the quarterback. Returning production with a new quarterback is a very different animal from the same percentage with a returning one.

The honest size of the effect

Here's the part that gets lost. Returning production is positively correlated with next-season results — more continuity tends to mean more wins, and brand-new teams tend to take a step back. But the relationship is loose. It's one ingredient among many: recruiting and talent, coaching changes, schedule, injuries, and — increasingly — the transfer portal, which has scrambled the very idea of "returning."

That last point matters more every year. In the portal era, a team can "lose" production to graduation and the draft, then reload through transfers who don't show up in a returning-production number at all. A roster that returns 55% of last year's production but adds three high-usage transfers may be more talented than a team returning 85% of a worse roster. Returning production was built for a world that's disappearing.

Let's measure it ourselves

The right way to settle "how predictive is it, really?" is to plot every team's returning-production share against its change in win total, and look at the strength of the relationship. That requires production data from the CollegeFootballData API, which is the right source for it:

When you do run it, expect a cloud of points that slopes upward but scatters widely — a real but modest relationship, the statistical signature of "matters, but far from everything." If returning production were destiny, the dots would hug the trend line. They won't. They'll be a smear with a gentle upward tilt, and the outliers — the team that returned everyone and collapsed, the rebuilt roster that surged — are where the actual stories live.

How to read returning-production takes

• Ask which positions. 80% returning production with a returning, proven quarterback is a real edge. The same 80% with the offensive line gutted and a new QB is not.
• Account for the portal. A "low returning production" team that won the transfer market isn't rebuilding — it's re-tooling. The number can't see incoming transfers.
• Mind the baseline. Returning a lot of production from a bad team just means you're bringing back the same problems.
• Use it for direction, not magnitude. It's good at flagging "this team should be more (or less) settled than last year." It's bad at telling you the exact win total.

The bottom line

Returning production earns its place in a preseason model — it carries genuine signal, and ignoring it would be a mistake. But it's a coefficient, not a crystal ball. The teams that beat their projections every year are the ones that found value the number couldn't see: a transfer quarterback, a coaching upgrade, a sophomore leap. Treat returning production as the floor under your expectations, then go find the reasons a team will beat it.