Each design choice maps to a specific failure mode
The biggest product risk is that a public competition accidentally selects the wrong people. The design above is built to reduce that.
Raw PnL contests reward variance. Weekly capital re-rating reduces that.
Paper contests reward fake execution. Five-minute averaged execution and turnover penalties reduce that.
Directional drift can masquerade as skill. Benchmark-relative and sector-adjusted scoring reduce that.
One lucky trade can dominate a short contest. Minimum activity plus week-by-week scoring reduce that.
Too much transparency invites gaming. A public principles layer plus a private algorithm layer reduces that.
Pure formula selection can still produce bad finalists. Human review reduces that.
| Risk | Mitigation |
|---|---|
| Raw PnL contests reward variance | Weekly capital re-rating |
| Paper contests reward fake execution | 5-min averaged execution + turnover penalties |
| Directional drift masquerades as skill | Benchmark-relative + sector-adjusted scoring |
| One lucky trade dominates a short contest | Minimum activity + week-by-week scoring |
| Too much transparency invites gaming | Public principles + private algorithm |
| Pure formula selection produces bad finalists | Human review |