Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when a Polymarket outcome price and filled volume are accelerating in the same direction.
.claude/skills/superior-trade-probability-momentum/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -51% | 0% |
Use this when someone asks for momentum, trend following, breakout, follow-the-crowd behavior, fast-moving crypto milestones, or rapid probability acceleration.
POST /v3/markets/search.backtestSupported: true.Good fit for this archetype. Filled TradeTick history can validate whether momentum would have produced fill-driven entries, exits, and directional exposure.
Limit: filled-trade backtests do not prove maker queue position, maker rebates, or order-book depth. Treat this as taker-style behavior evidence, not certainty.
Enter YES when recent probability rises more than a threshold over lookback_ticks. Exit when exit_after_ticks is reached, or when momentum stalls if an exit condition is included.
subscribe_trade_ticks(instrument_id) to stream TradeTicks.on_trade_tick as the main event handler.on_order_filled.json{ "lookback_ticks": 5, "momentum_threshold": 0.02, "order_size": 10, "exit_after_ticks": 25, "max_entries": 3 }
| Knob | Effect | |---|---| | lookback_ticks | Larger values reduce noise; smaller values react faster. | | momentum_threshold | Higher values reduce frequency and require stronger momentum. | | order_size | Larger values increase PnL sensitivity and liquidity exposure. | | exit_after_ticks | Shorter values trim quickly; longer values hold trend. | | max_entries | Caps repeated entries in choppy markets. |
"This is a probability momentum strategy: it buys when fills and last-trade probability move together in one direction. It is backtestable with filled TradeTicks, but it does not confirm maker queue priority or spread capture."
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 27,644 | 18,368 | -34% | 1 | 1 | 0% | 4,993 | 3,667 | -27% | 0 | 0 | — |
case-02 | fail→fail | 18,606 | 10,947 | -41% | 1 | 1 | 0% | 3,173 | 2,632 | -17% | 0 | 0 | — |
case-03 | fail→pass | 27,269 | 11,869 | -56% | 1 | 1 | 0% | 4,052 | 2,609 | -36% | 0 | 0 | — |
case-04 | pass→pass | 24,761 | 22,221 | -10% | 1 | 1 | 0% | 3,895 | 3,935 | +1% | 0 | 0 | — |
case-05 | pass→pass | 13,659 | 22,943 | +68% | 1 | 1 | 0% | 2,085 | 4,900 | +135% | 0 | 0 | — |
case-06 | pass→pass | 39,000 | 35,835 | -8% | 1 | 1 | 0% | 6,965 | 7,129 | +2% | 0 | 0 | — |
case-07 | fail→pass | 20,457 | 11,149 | -46% | 1 | 1 | 0% | 2,982 | 2,484 | -17% | 0 | 0 | — |
case-08 | fail→pass | 14,022 | 3,476 | -75% | 1 | 1 | 0% | 2,172 | 1,094 | -50% | 0 | 0 | — |
case-09 | fail→pass | 15,295 | 3,301 | -78% | 1 | 1 | 0% | 2,183 | 1,074 | -51% | 0 | 0 | — |
case-10 | fail→pass | 22,358 | 19,656 | -12% | 1 | 1 | 0% | 3,982 | 3,536 | -11% | 0 | 0 | — |
case-11 | fail→pass | 17,078 | 10,606 | -38% | 1 | 1 | 0% | 2,300 | 2,009 | -13% | 0 | 0 | — |
case-12 | pass→pass | 24,858 | 14,135 | -43% | 1 | 1 | 0% | 3,336 | 2,587 | -22% | 0 | 0 | — |
case-13 | pass→pass | 20,575 | 11,331 | -45% | 1 | 1 | 0% | 2,867 | 2,327 | -19% | 0 | 0 | — |
case-14 | pass→pass | 15,622 | 10,342 | -34% | 1 | 1 | 0% | 2,237 | 1,870 | -16% | 0 | 0 | — |
case-15 | fail→pass | 20,569 | 14,558 | -29% | 1 | 1 | 0% | 3,199 | 2,830 | -12% | 0 | 0 | — |
case-16 | pass→pass | 14,023 | 6,043 | -57% | 1 | 1 | 0% | 2,078 | 1,501 | -28% | 0 | 0 | — |
case-17 | pass→pass | 15,464 | 6,213 | -60% | 1 | 1 | 0% | 2,737 | 1,621 | -41% | 0 | 0 | — |
case-18 | fail→pass | 13,808 | 13,256 | -4% | 1 | 1 | 0% | 2,498 | 2,548 | +2% | 0 | 0 | — |
case-19 | fail→pass | 13,871 | 2,308 | -83% | 1 | 1 | 0% | 2,021 | 990 | -51% | 0 | 0 | — |
case-20 | fail→pass | 14,230 | 3,095 | -78% | 1 | 1 | 0% | 2,078 | 1,025 | -51% | 0 | 0 | — |
case-21 | fail→pass | 9,201 | 3,058 | -67% | 1 | 1 | 0% | 1,390 | 1,117 | -20% | 0 | 0 | — |
case-22 | pass→pass | 9,865 | 7,463 | -24% | 1 | 1 | 0% | 1,567 | 1,749 | +12% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +55 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.