The Payout Factor: Numerai's Hidden Lever

How Numerai's payout factor works, what drives it, and why it shapes staking returns more than any single model score.

Most Numerai participants obsess over MMC and correlation scores, but one variable affects staking returns more directly than any individual metric: the payout factor. It is a tournament-wide multiplier that scales every payout and burn up or down, and rounds in the top quartile (factor 0.57–1.00) post roughly 25x the median per-round return of rounds in the bottom quartile (factor 0.08–0.12).

This article walks through the payout factor's mechanics, history, and impact on returns. For a refresher on how staking itself works, see Is Staking Profitable?.

What Is the Payout Factor?

The payout formula for each round is:

Payout = stake x (0.5 x CORR + 2.0 x MMC) x payout_factor

The payout factor is the final multiplier. At 1.0, payouts run at the "normal" rate. At 0.5, every payout and every burn is halved. Above 1.0 (rare in recent history), returns are amplified in both directions.

Numerai adjusts the factor based on how close total NMR at stake is to an internal target. Too much staked NMR pushes the factor down to cool incentives; too little pulls it up to attract capital. It is a thermostat for tournament participation.

The Payout Factor Over Time

The payout factor has never been a stable constant. Since round 200 it has swung from 1.0 down to near zero as tournament dynamics shifted.

Payout factor and total NMR at stake across Numerai tournament rounds, showing the factor falling from 1.0 toward 0.1 as stake grew past 400K NMR
Payout factor and total NMR at stake across Numerai tournament rounds, showing the factor falling from 1.0 toward 0.1 as stake grew past 400K NMR

The chart plots payout factor against total stake round-by-round. As staked NMR climbed past roughly 400K during the enthusiasm of rounds 300–500, the factor collapsed from 1.0 into the 0.1–0.3 band and has largely stayed there since. Brief recoveries line up with stake withdrawals after drawdowns. For how earned and burned NMR flow each round, see Round Economics; for broader supply context see NMR Token Economics.

What Drives It?

A scatter of every round makes the relationship explicit: higher total stake produces lower payout factors, and the fitted trend is clearly negative.

Scatter plot of payout factor against total NMR staked per round with a red trend line sloping down from 0.75 at zero stake toward 0.0 near 1M NMR
Scatter plot of payout factor against total NMR staked per round with a red trend line sloping down from 0.75 at zero stake toward 0.0 near 1M NMR

The fit is not mechanical — Numerai retains discretion — but the pattern is consistent enough to be useful. Sharp rises in tournament stake generally precede payout factor compression, and periods of withdrawal precede factor increases.

The tournament has, in effect, a self-correcting capacity constraint. High returns attract capital, which compresses future returns. Losses drive capital out, which expands future returns. It is the same dynamic that caps hedge fund capacity: alpha gets arbitraged away as more money chases the same signals.

Impact on Returns

How much does the factor actually matter? Grouping rounds into payout factor quartiles gives a stark answer.

Bar chart of median per-round return by payout factor quartile: roughly 0% for 0.08–0.12, 0.1% for 0.12–0.36, 0.8% for 0.36–0.57, and 2.5% for 0.57–1.00
Bar chart of median per-round return by payout factor quartile: roughly 0% for 0.08–0.12, 0.1% for 0.12–0.36, 0.8% for 0.36–0.57, and 2.5% for 0.57–1.00

Rounds in the top quartile (factor 0.57–1.00) produced a median return near 2.5% per round. The bottom quartile (0.08–0.12) produced essentially zero. Part of this is mechanical — the same score yields a larger payout when the factor is high — but it also reflects the environment. High payout factors tend to coincide with lower competition and more favorable market regimes.

The rolling view confirms the dynamic over time.

Line chart of rolling 20-round median return tracking payout factor across rounds, both dropping together from around 5% and 1.0 in the early 300s to near-zero by round 600
Line chart of rolling 20-round median return tracking payout factor across rounds, both dropping together from around 5% and 1.0 in the early 300s to near-zero by round 600

The rolling 20-round median return tracks the payout factor step for step. When the factor drops, aggregate returns compress even when model quality is constant. A model can perform identically across two periods and still hand the staker wildly different payouts.

Should You Time Your Staking?

In theory, you could pile on stake during high-factor periods and pull back during low ones. In practice, three things get in the way.

Payout factor changes are not announced in advance. You can watch the trend but cannot call the turning points. Stake changes also take rounds to settle, so you cannot flip exposure on a dime. And the rounds with the highest factor often arrive right after painful burns — the environment that earns those returns is genuinely riskier.

A more practical use is to treat the payout factor as context for your model's results. Modest returns during a low-factor period may reflect a strong model on a factor-adjusted basis. Spectacular returns during a high-factor period deserve more skepticism.

Takeaways

The payout factor is the single largest external driver of staking returns. The same model can produce very different payouts depending on when it stakes.

Total stake and payout factor move inversely. More capital in the tournament means smaller payouts per NMR. It is Numerai's built-in capacity constraint.

High payout factor periods follow pain. The factor rises when stake drops, which typically happens after widespread burns. The best entry points line up with the worst sentiment.

Evaluate your model on factor-adjusted returns. A 0.5% return during a 0.10 factor period likely reflects a better model than a 2% return at factor 0.80.