Deadband and haircut
How a winning close is scaled before the win fee. Small wins are trimmed hardest.
The gain is calculated as if the exit price were entryPrice / 50,000 (0.2 bps) worse. A win smaller than that pays nothing. This removes the incentive to trade on price noise.
Haircut
adjustedGain = rawGain × scale
scale = (1 − baseRate) / (1 + term1 + term2)
term1 = 1 / (move × rateMultiplier)
term2 = referenceNotional / (10^6 × move × positionMultiplier)
move = |exitPrice − entryPrice| / entryPrice (exit after the deadband)
| Parameter | Value |
|---|---|
| baseRate | 10% |
| rateMultiplier | 15,000 |
| referenceNotional | $100,000 |
| positionMultiplier | one value per leverage tier, to be calibrated (Papertrade: 814.598 for BTC, 483.979 for ETH) |
As the move grows, both terms shrink and the trader keeps more, approaching 90%. As the move shrinks toward zero, the scaled gain collapses toward zero. The percentage does not depend on position size: a $100 close and a $1M close on the same move are scaled the same.
Share of a win kept after deadband and haircut, using Papertrade's BTC multiplier as a stand-in:
| Move in your favour | Kept |
|---|---|
| 1% | ~88% |
| 0.3% | ~84% |
| 0.1% | ~74% |
| 0.02% | ~40% |
| 0.005% | ~7% |