Short Disclaimer

This report is a historical simulation study only. It does not constitute trading advice, does not predict future outcomes, and should not be used as the sole basis for any trading decision.

Intro / Thesis

We investigated whether spot BTC crossing above or below its 1‑hour volume‑weighted average price (VWAP) on Coinbase provides a directional edge in Kalshi’s 15‑minute Bitcoin markets (KXBTC15M). The core idea is straightforward: a cross above VWAP signals upward momentum (enter long / buy YES), a cross below signals downward momentum (enter short / buy NO), and the trade is exited when price crosses back through VWAP. This type of VWAP‑based mean‑reversion‑or‑momentum logic is common in traditional futures, and we wanted to see how it translates to a binary‑outcome prediction market.

Variant and Strategy Explanation

Every tested variant shares the same base rule structure:

• Entry long (buy YES): When BTC spot rises above the 1h VWAP, no current position exists, and the market spread is below 0.03.
• Entry short (buy NO): When BTC spot falls below the 1h VWAP, no current position exists, and spread is under 0.03.
• Exit long: If a long position is open and BTC spot drops back below the 1h VWAP, sell all.
• Exit short: If a short position is open and BTC spot moves back above the 1h VWAP, sell all.

Positions are fixed at size 5 with a max exposure of 10 units and a price floor/ceiling of 0.10/0.90. The data source is Coinbase BTC‑USD with both the latest price and 1h VWAP refreshed every 10 seconds. The ensemble of 100 variants explores parameter (and likely minor rule‑timing) variations around this base specification, giving us a good picture of the idea’s robustness.

Top Results

The top‑ranked variants cluster tightly around two similar performance bands:

• Rank 1 – 4 (variants 017‑020): Total PnL of –$525.95, maximum drawdown –$537.36, ROI –2103.8%. Each placed 5,496 trades with a win rate of roughly 9.39%.
• Rank 5 – 8 (variants 037‑040): Total PnL of –$532.80, maximum drawdown –$539.08, ROI –2131.2%. Trade count is slightly higher at 5,552 and the win rate nudges up to 9.40%.

Sharpe ratios are reported as 0 across the board. In every case, even the “best” variant loses money net‑net. The marginally higher win rates in the second group did not translate into better overall PnL, suggesting that the additional trades increased friction or were entered at slightly worse prices.

Bottom Results

The worst‑performing variants are again grouped into two blocks:

• Rank 93 – 96 (variants 061‑064): Total PnL –$536.87, maximum drawdown –$542.66, ROI –10737.4% from 5,666 trades with win rates around 9.60%.
• Rank 97 – 100 (variants 081‑084): Total PnL –$538.08, maximum drawdown –$543.87, ROI –10761.6% from 5,708 trades and a win rate of about 9.64%.

Notice a faint pattern: the worst variants actually have the highest win rates (9.64% vs. 9.39% in the top variants). However, they are also the most active, with up to 5,708 trades versus 5,496. This strongly suggests that more frequent trading erodes profitability further—likely through cumulative spread costs and transaction friction.

Conclusion

The VWAP‑cross strategy, as tested on Kalshi’s KXBTC15M market, shows no directional edge. All 100 variants lose money, with a win rate consistently below 10% and extreme negative ROI. Increasing trade frequency appears to worsen results rather than improve them. The simulation does not support the thesis that BTC’s hourly VWAP cross is a reliable momentum signal for these binary outcomes. Each tested and saved variant remains accessible as a runnable Turbine strategy for further inspection, but the underlying concept—at least within the constraints we explored—did not survive history.

Long Disclaimer

This research is purely a historical simulation exercise conducted on past market data. It does not represent live trading performance and does not account for execution delays, liquidity shifts, or changes in market microstructure that could affect real‑world results. Past simulations are not indicative of future profitability. All trading involves substantial risk of loss, and you should consult a qualified financial professional before engaging in any trading activity. The strategies discussed are mechanical illustrations, not recommendations. Turbine makes no representation that these strategies will perform similarly in the future.