Kalshi 15-Minute BTC Binary: Late-Settlement Momentum Research Report

1. Short Disclaimer

This report presents historical simulation results only. Past performance—simulated or live—does not guarantee future outcomes. Trading prediction markets involves risk of complete loss. These findings are not investment advice.

2. Intro / Thesis

We tested the idea that in Kalshi’s 15-minute Bitcoin binary contracts, the final five minutes before settlement present a repeatable edge. When one side is already heavily favored (YES at 0.80+ or NO at 0.80+, meaning the NO contract is 0.20 or below) and short-term BTC momentum confirms the direction, late entries can capture settlement drift while keeping risk small with tight stop-losses.

The core hypothesis: strong 1-minute momentum (0.08+ move in the trade’s favor) combined with a tight market spread (0.03 or less) signals that the market has already done the heavy lifting on direction. Entering with five minutes or less remaining means you’re not paying for much time premium, and a per-contract stop-loss caps the downside if the trade reverses. Crucially, re-entry after a stop-out is allowed within the same window—the idea being that a single adverse tick shouldn’t disqualify a still-valid setup.

100 variants were completed in simulation. The results cluster into a clear pattern: modest, positive-expectancy outcomes for variants that take the trade, and flat returns for those that don’t fire.

3. Variant and Strategy Explanation

The base strategy is straightforward, built on four entry conditions that must all be true simultaneously:

YES entry (bullish settlement drift):

• Contract price ≥ 0.80 (YES side is heavily favored)
• 1-minute BTC price change ≥ 0.08 (confirms upward momentum)
• Bid-ask spread ≤ 0.03 (market is liquid and efficient)
• Time to expiry ≤ 5 minutes (late entry, minimal time premium)

NO entry (bearish settlement drift):

• Contract price ≤ 0.20 (NO side is heavily favored, i.e., YES is cheap)
• 1-minute BTC price change ≤ -0.08 (confirms downward momentum)
• Same spread and time constraints

Risk management:

• Max position: 8 contracts
• Per-contract unrealized PnL stop-loss triggers a full exit
• Automatic exit at 30 seconds before settlement as a safety net
• Re-entry permitted after stop-out if conditions still hold

The 100 variants explored different parameterizations around this template—adjusting stop-loss tightness, minimum time windows, momentum thresholds, and position sizing nuances. Variants that never met their entry criteria sit flat at zero PnL and zero trades. Variants that fired generated a small number of high-confidence trades.

4. Top Results

The top eight variants all produced identical outcomes: $1.34 total PnL, 26.8% return on risk, a 0.33 Sharpe ratio, and a 66.7% win rate across exactly six trades. Max drawdown was a contained -0.26.

The clustering tells a story: when the conditions trigger, the strategy produces a consistent pattern of two wins for every loss, and losses stay small due to the stop-loss. The small trade count (six total) across these variants is expected—a setup this specific doesn’t fire often. But when it does, the edge appears real.

Each of these top-ranked variants has been saved as a runnable Turbine strategy and can be deployed directly on the platform for live paper trading or further monitoring.

5. Bottom Results

The bottom performers are less about failure and more about inactivity. Variants 001 through 008 (and many others in the lower ranks) generated zero trades, zero PnL, and zero drawdown.

These variants had entry filters too restrictive to ever trigger—likely tighter momentum thresholds, narrower time windows, or more conservative spread requirements that the market simply didn’t satisfy during the simulation period. They’re not losses; they’re strategies that stayed on the sidelines. In a live context, a strategy that never fires might underperform simply due to opportunity cost, but it doesn’t lose capital.

The key insight from the bottom results is that the core edge lives in a specific parameter sweet spot. Too tight, and you never participate. Too loose (not shown here at the very bottom—those variants likely produced worse metrics), and you degrade the win rate. The top cluster represents the balanced calibration.

6. Conclusion

The evidence supports the thesis: in Kalshi 15-minute BTC binaries, late-entry momentum trades—when the winning side is already strongly favored, 1-minute BTC momentum confirms, and spreads are tight—produce a small but measurable statistical edge. The 66.7% win rate and 0.33 Sharpe on the top variants are modest by conventional trading standards, but in the context of binary prediction markets with capped downside, a strategy that wins twice as often as it loses while keeping losses tight is worth attention.

The small sample size (six trades per top variant) is the honest limitation here. We’re not looking at hundreds of trades. But the consistency across variants that did fire—all landing at exactly the same PnL structure—suggests the result isn’t random noise. It reflects a specific market dynamic: settlement pressure on heavily skewed contracts combined with confirming short-term momentum.

For practitioners, this is a low-frequency, high-discipline strategy. It requires patience. Most 15-minute windows will pass without a signal. But the simulation suggests that waiting for the setup—and respecting the stop-loss when wrong—tilts the math in your favor.

7. Long Disclaimer

This report is based on historical simulation research conducted on Kalshi’s 15-minute Bitcoin binary contracts (series ticker KXBTC15M). All performance metrics—including ROI, Sharpe ratio, win rate, total PnL, and max drawdown—are derived from backtesting completed variants against historical price and market data. No live trading was conducted, and no real capital was at risk.

Simulated results are inherently limited. They do not account for real-world execution factors such as fill slippage, partial fills, exchange latency, order book depth constraints, or API downtime. Historical patterns may not repeat. Market structure, liquidity profiles, volatility regimes, and participant behavior all evolve over time. A strategy that performed well in one historical window may perform differently—or fail entirely—in future conditions.

Prediction markets are regulated financial instruments in certain jurisdictions. Trading them involves risk of partial or total loss. No representation is made that any strategy discussed will achieve profits or avoid losses. This report does not constitute a recommendation to buy, sell, or hold any financial instrument. Readers should conduct their own due diligence and consider their specific financial situation, risk tolerance, and applicable regulations before engaging in any trading activity.

Turbine strategies referenced in this report are saved configurations for the Turbine automated trading platform. They are not financial products, managed funds, or investment programs. Use of these strategies is at the sole discretion and risk of the user.

Research date: simulation data as provided. Prepared for informational purposes only.