14.3K models in the Classic tournament. Per-round staking flow, stake concentration, turnover, effective N, and payout ROI analysis.
| # | Model | Stake |
|---|---|---|
| 1 | shatteredx | 106.3K NMR |
| 2 | cloud_predictor | 58.3K NMR |
| 3 | phorex | 37.6K NMR |
| 4 | reputation | 30.0K NMR |
| 5 | aininja | 20.2K NMR |
| 6 | roadrunner | 14.5K NMR |
| 7 | lolabunny | 14.4K NMR |
| 8 | super_model | 14.4K NMR |
| 9 | mrotsdma20 | 13.0K NMR |
| 10 | ummon | 12.4K NMR |
Sortable table of every weekly round since round 400. Columns: total stakers, net change vs prior round, new stakers (entered staking from any prior state), converted stakers (subset whose immediately prior submission was unstaked), unstakers (had stake last round, no longer staked this round), and unstaked submitters (submitted with zero NMR stake this round). Surfaces the in/out flow of staked participation week over week.
A 100% stacked bar chart showing the percentage of total tournament stake held by each of the top 10 models (plus an "other" bucket) each round. Reveals whether a small number of models dominate the tournament's stake or if stake is broadly distributed.
Area chart with four layers showing the cumulative percentage of total stake held by the top 1, 5, 10, and 25 models over time. A rising area means stake is concentrating among fewer models; a shrinking area means stake is distributing more broadly.
Line chart comparing the stake-weighted average MMC against the equal-weight average MMC each round. When the stake-weighted line is above the equal-weight line, models with more NMR staked are outperforming the average — indicating that capital allocation in the tournament correlates with predictive skill.
Area chart showing the annualized return on investment across the tournament, calculated as (total payout / total stake) × 52. Clipped to a ±50% range. Shows the overall profitability of staking NMR in the Numerai tournament over time.
Stacked bar chart for each round showing how many of the top 20 staked models are returning from the previous round versus newly entering the top 20. High turnover indicates a competitive, dynamic leaderboard; low turnover indicates entrenched top models.
Composed chart with the effective number of models (calculated as 1 divided by the Herfindahl-Hirschman Index of stake shares) on the left axis and total model count on the right axis. The effective N measures how many models would exist if stake were equally distributed — a lower effective N relative to total models means stake is highly concentrated.
Composed chart with total NMR staked on the left axis and USD equivalent value on the right axis. Shows both the raw amount of NMR committed to the Classic tournament and its dollar value over time, reflecting both participation trends and NMR price movements.
A 100% stacked area chart showing the proportion of stake held by models with positive MMC versus negative MMC among the top 50 staked models. When the positive area dominates, the largest stakers are contributing to the meta model; when negative grows, large stakers are underperforming.
Line chart comparing the stake-weighted average model age (in years) against the equal-weight average. When the stake-weighted age exceeds the equal-weight age, older established models hold more stake. When it falls below, newer models are capturing more of the staking capital.