Short Disclaimer

This report summarizes historical simulation research only. Nothing here constitutes trading advice or a forecast of future performance.

Intro / Thesis

We asked a simple question: Does short-term BTC momentum work better when Coinbase BTC is already trading above its 1‑hour VWAP? The base strategy combines EMA/SMA crossover confirmation with positive momentum filters—but only enters when price holds above that volume‑weighted benchmark. Across 100 systematically varied configurations, we tested whether the VWAP regime filter meaningfully separates winners from losers.

The conclusion is blunt: the framework, in every variant we sweeps, produced deeply negative results. No parameter combination rescued it. The VWAP filter did not create a profitable regime; the momentum setup itself appears unreliable in the Kalshi KXBTC15M market structure.

Variant and Strategy Explanation

All 100 variants descend from a single YAML‑defined base strategy operating on Kalshi’s BTC 15‑minute binary market (KXBTC15M). The core logic is a four‑filter entry:

1. VWAP regime: Coinbase BTC price must be above the 1‑hour VWAP.
2. EMA/SMA crossover: 12‑period EMA (1‑minute bars) must be above 20‑period SMA.
3. Positive momentum: 1‑minute velocity above 0 and 5‑minute change above 0.
4. Entry ceiling: Contract price ≤ 0.70 (prevents buying late‑cycle expensive contracts).
5. Liquidity & time guardrails: Spread ≤ 0.03, time to expiry ≥ 5 minutes.

When all conditions align, the strategy buys YES with a base size of 5 contracts. Exits trigger on:

• Profit target (contract price ≥ 0.85)
• VWAP break (BTC dips ≤ VWAP)
• Momentum flip (velocity < –10 or 5m change < –0.5%)
• Time cutoff (≤ 2 minutes to expiry)
• Max realized loss (–$10.00)

The research design swept key parameters across 100 combinations:

• Momentum velocity thresholds
• VWAP‑distance sensitivity (how far above VWAP counts as “in regime”)
• Entry price ceiling values
• Profit‑target levels
• Time‑to‑expiry cutoffs
• Position‑sizing increments

Every successful variant was automatically saved as a runnable Turbine strategy.

Top Results

Even the “best” variants were losers.

The top‑ranked variants clustered into two groups: one producing –756% ROI with a –$214 max drawdown, the other slightly worse at –808% ROI. Win rates hover near 11%, meaning roughly one in nine trades closed profitably. That win rate is far too low to overcome the near‑binary loss profile of these 15‑minute markets, where losers frequently go to near zero.

Notably, the best drawdown figures (–$214, –$224) are still multiples of the allowed max loss (–$10), indicating the stop‑loss was either hit repeatedly intra‑series or simply unable to prevent cumulative decay across consecutive losing sequences. Sharpe ratios near zero (0.01 to –0.01) confirm there is no risk‑adjusted edge.

The VWAP filter succeeded in reducing entry frequency during hostile regimes—total trades were lower than an unfiltered baseline would imply—but it failed to raise the quality of the entries it did permit.

Bottom Results

The lowest‑ranked variants were worse by wider margin than the headline percentages suggest.

The worst ROI percentages (–1,902% to –1,906%) reflect a much smaller capital base in these particular runs, so absolute dollar losses appear lower than the top group. However, the drawdown (–$99.96) was nearly total account depletion across every bottom‑ranked variant. These configurations executed fewer total trades (1,314–1,316) but achieved essentially the same win rate (~11.5–11.6%) as the top group—that is, the exit and sizing adjustments that made these the “worst” didn’t meaningfully alter the underlying trade quality. They simply concentrated losses into a tighter capital base.

The practical takeaway is that varying the exit timing, profit target, or position sizing did not create separation between good and bad variants. The range from top to bottom is narrow in terms of what matters: all of them lose, and the ranking is largely an artifact of absolute trade count and starting capital.

Conclusion

The VWAP regime filter, combined with EMA/SMA crossover and positive momentum, produced no profitable configuration across 100 systematically varied test variants on Kalshi’s 15‑minute BTC market. Win rates remained locked around 11% regardless of parameter changes, and every variant suffered severe drawdowns.

Why did this fail? The short‑binary timeframe (15‑minute expiry) likely magnifies noise relative to the signal these indicators capture. A five‑minute change and one‑minute velocity filter may work directionally for spot BTC, but in a binary‑options market with bounded contract pricing, transaction costs (spread), and rapid time decay, benign directional moves do not reliably convert to profitable YES positions. The VWAP condition adds a regime filter but cannot overcome the structural problem: even “good” entries are killed by adverse expiry dynamics.

For the research record, this is a clean negative result. It tells us that a momentum + VWAP overlay on ultra‑short‑dated binary markets is unlikely to generate positive expectancy under these specifications.

Long Disclaimer

This document is a historical simulation research report prepared by Turbine’s research function. It describes the results of automated strategy testing on past market data only. The performance metrics—ROI, Sharpe ratio, total PnL, win rate, drawdown—are backward‑looking and do not represent live trading results, nor do they predict future outcomes in any market.

All strategies described were executed in a simulated environment using historical price feeds from Coinbase and Kalshi. No actual capital was risked. The research does not account for real‑world factors including but not limited to exchange latency, partial fills, order book depth, slippage, or changes in market microstructure that may differ materially from the backtesting environment.

Each successful variant was saved as a runnable Turbine strategy for further analysis and monitoring. The existence of a saved strategy does not imply endorsement, profitability, or suitability for live deployment. Users must conduct their own independent evaluation and accept full responsibility for any trading decisions.

Past performance, whether simulated or real, is not indicative of future results. Trading binary options and cryptocurrencies involves substantial risk of complete loss. Do not trade with capital you cannot afford to lose.