Losing less is not the same as winning.
A group may lose less simply because it wagers less. Stops alter exposure and timing; they do not make a negative-expectation wager profitable.
Ten groups · 100 policies each
How the policy groups compared
These are modelled policy families, not descriptions or diagnoses of people. Unless stated otherwise, the comparison reflects both behaviour and the exposure it produced.
| Cohort | Median P&L | 5th–95th percentile | Finished ahead | Ruin | Turnover | Across-world median |
|---|---|---|---|---|---|---|
| 01Small bets, frequent100 / 100 policies | -R 935 | -R 3 428 to R 2 756 | 24 / 100 (24,0%) | 0 / 100 (0,0%) | R 903K | -R 1 099-R 1 381 to -R 815 world range |
| 02Small bets, infrequent100 / 100 policies | -R 237 | -R 664 to R 462 | 21 / 100 (21,0%) | 0 / 100 (0,0%) | R 90K | -R 210-R 258 to -R 162 world range |
| 03Medium bets, frequent100 / 100 policies | -R 4 306 | -R 8 435 to R 15 535 | 23 / 100 (23,0%) | 0 / 100 (0,0%) | R 3M | -R 3 937-R 4 498 to -R 3 210 world range |
| 04Medium bets, infrequent100 / 100 policies | -R 969 | -R 3 226 to R 4 265 | 24 / 100 (24,0%) | 0 / 100 (0,0%) | R 565K | -R 841-R 1 088 to -R 607 world range |
| 05Large bets, frequent100 / 100 policies | -R 8 976 | -R 15 208 to R 29 626 | 29 / 100 (29,0%) | 21 / 100 (21,0%) | R 5M | -R 9 662-R 9 896 to -R 8 707 world range |
| 06Large bets, infrequent100 / 100 policies | -R 2 138 | -R 7 315 to R 3 624 | 21 / 100 (21,0%) | 0 / 100 (0,0%) | R 902K | -R 1 728-R 2 287 to -R 1 161 world range |
| 07Loss-chasing100 / 100 policies | -R 6 408 | -R 9 989 to R 16 402 | 27 / 100 (27,0%) | 4 / 100 (4,0%) | R 4M | -R 6 176-R 7 363 to -R 4 643 world range |
| 08Strict limits100 / 100 policies | -R 1 130 | -R 3 528 to R 2 775 | 23 / 100 (23,0%) | 0 / 100 (0,0%) | R 869K | -R 1 236-R 1 558 to -R 916 world range |
| 09Variable / random staking100 / 100 policies | -R 5 218 | -R 9 650 to R 16 109 | 15 / 100 (15,0%) | 0 / 100 (0,0%) | R 2M | -R 3 992-R 4 831 to -R 3 145 world range |
| 10Random control100 / 100 policies | -R 3 178 | -R 8 994 to R 9 544 | 20 / 100 (20,0%) | 1 / 100 (1,0%) | R 2M | -R 2 543-R 3 315 to -R 1 773 world range |
The rankings did not hold.
Across-world results keep each policy fixed while changing chance. They provide the needed context for an apparently exceptional cohort.
Every policy remains counted.
Each rate uses all 100 policies. Capped or inactive policies stay in the denominator.