Compounding depends on which capital is available to reinvest. Applying every daily round’s return to an entire account assumes locked NMR can be reused immediately. That assumption does not describe individually funded atomic positions.
Revised September 8, 2026. The previous universal payout formula and causal “five times more NMR” interpretation are withdrawn.
The arithmetic is conditional on reinvestment
A single tranche gaining 20% and then losing 20% ends at 96% of its original size if its full proceeds are reinvested between the two outcomes. Constant-principal positions with those same rates instead have zero combined net payout before costs. Neither calculation applies until the funding and release sequence is specified.
The illustration starts with two 10 NMR positions and assumes both earn 10% each cycle. After four complete cycles, constant allocation has 28 NMR including idle gains; compounding has about 29.28 NMR. These are hypothetical cycles, not four daily openings or an annual forecast.
Score volatility is not payout volatility
Payout weights, clipping, principal and score dependence affect the NMR outcome. Comparing raw MMC variance with lifetime NMR earnings also mixes model size, longevity and cohort survival. Matching average MMC alone does not isolate the effect of volatility.
The configured payoff guide uses both atomic payout scores. A reliable account analysis would then apply each settled outcome to the specific tranche and allocation mode, keeping unavailable capital locked.
I would describe an observed low-volatility earnings premium as an association until funding, tenure and selection effects are tested. The profitability note explains why cumulative pool earnings are not a typical account return.
Method: hypothetical arithmetic with no fees, losses or external funding in the plotted example. The two-step ±20% example is separate and shows why the reinvestment assumption changes the result; no empirical causal premium is estimated.