Short disclaimer

This is a historical simulation report. Past simulated performance does not guarantee future outcomes. These results are not live trading records and carry no forward-looking claims.

Intro / thesis

I tested whether sharp, short-lived panic moves in the KXBTC15M YES price on Kalshi tend to partially reverse within the same 15-minute window. The idea was simple: when YES drops quickly by $0.03–$0.15, step in as a buyer with a fixed contract size, then exit as soon as price recovers by $0.04. The core question was whether fine-tuning the panic threshold matters more than adjusting fade size. Spoiler: nothing worked.

Variant and strategy explanation

The base strategy waited for a sudden price decline relative to recent levels, entered long with a set number of contracts, and closed on a $0.04 bounce. Every completed variant explored variations of three levers: the panic threshold that triggers an entry, the number of contracts to fade with (50, 100, or 200), and the recovery exit target. All variants capped total fades at three and respected a price floor of $0.05 and ceiling of $0.95. Risk limits stayed identical across runs. The 100 completed variants span the full grid of panic_threshold sensitivity and fade_size combinations drawn from the base DSL. Each successful variant is saved as a runnable Turbine strategy, labeled with a unique strategy ID and slug for traceability.

Top results

The “best” variants were all identical in outcome, and they were still losers. The top eight variants share the exact same stats: -$10.72 total PnL, -107.2% ROI, Sharpe of -0.01, 35.2% win rate over 108 trades, and a max drawdown of -46.79. They differ only in internal label and strategy slug — functionally they are carbon copies. There is no meaningful separation among the top ranks. The recovery exit worked just often enough to avoid catastrophe but not enough to cover the losers. Win rate near 35% with a fixed $0.04 exit and no scaling meant the math was stacked against profitability from the start.

Bottom results

The worst variants fared even worse by amplifying the same losing mechanics. The bottom eight entries all posted -$27.23 total PnL, -108.9% ROI, the same -0.01 Sharpe, a win rate of 34.5% over 109 trades, and a max drawdown of -117.11. That drawdown figure exceeding -100% indicates these variants blew through their starting capital in simulation. Larger fade sizes or more sensitive panic thresholds in this group did not improve the reversal capture; they simply increased exposure to the same adverse drift. The slightly lower win rate and identical Sharpe confirm the strategy had no edge under any tested parameter set.

Conclusion

The panic fade premise did not hold up in historical simulation on KXBTC15M. Neither threshold sensitivity nor fade size turned the strategy positive. Every variant, top or bottom, produced negative total PnL and negative ROI. The recovery swing of $0.04 was too small to overcome the base rate of failed reversals, and the fixed exit left no room for the occasional larger snap-back to compensate. A live version of this approach, using these parameter ranges, would not have survived the drawdowns seen here. The research is conclusive on this dataset: fading intra-window panic without additional filters or dynamic exits is not a viable standalone strategy.

Long disclaimer

This report is a historical simulation study conducted for internal research purposes only. No real money was traded, and simulated results do not account for execution slippage, exchange fees, liquidity constraints, or market impact that would exist in live trading. The performance metrics — PnL, ROI, Sharpe ratio, win rate, and drawdown — are derived from backfilled price data and assume perfect fills at observed prices. These numbers can overstate what is achievable in practice. Past simulated performance does not predict future returns, and any strategy that was unprofitable in simulation should be assumed unprofitable going forward unless proven otherwise in live conditions. This document does not constitute trading advice or a recommendation to trade any instrument on Kalshi or elsewhere. All strategy variants remain archived as runnable DSL configurations for audit and reproduction, not as endorsements.