Deep Research

BTC 15m Variants

For the saved BTC 15m momentum-alignment strategy, test whether smaller entries and a wider entry band reduce the per-trade taker fee drag versus the current 10-contract / 0.45-0.55 setup, holding all rules, signals, and exits fixed.

Historical research only. Not investment advice.

Comparison Results

BTC 15m Momentum-Alignment: Entry Size and Band Width Variants

Short disclaimer: This is historical simulation research, not investment advice. Past simulated results do not indicate future performance. Nothing here is a promise or projection of profit.

Intro / thesis

The saved BTC 15-minute momentum-alignment strategy on Kalshi trades a simple idea: when Coinbase BTC-USD shows short-term momentum in one direction, take the matching side of the KXBTC15M contract inside a price band, and exit near expiry or on a stop. That works, in simulation, but it pays taker fees on every entry. Our question was narrow and practical: does shrinking the entry size, widening the entry band, or both, reduce the per-trade taker fee drag enough to matter?

The logic is straightforward. Smaller entries mean less notional per trade, so the fixed per-contract fee takes a smaller bite out of a smaller position — but it also means each win contributes less. A wider band admits more trades at more extreme prices, where the fee is proportionally smaller relative to the potential payout, but it also admits lower-quality entries. This batch ran four variants to test those two dials against the same rules, signals, and exits.

Variant and strategy explanation

All four variants share the identical rule set and evaluation loop. The strategy checks every 10 seconds. Rules are evaluated top to bottom and the first one that fires ends that tick.

  • Rule 1: when time_to_expiry < 5s, sell all.
  • Rule 2: when unrealized_pnl <= -4.5, sell all.
  • Rule 3: when time_to_expiry <= 5m AND price >= band low AND price <= band high AND edge.btc_1m.change_5m > 0 AND edge.btc_1m.velocity_1m > 0, buy yes.
  • Rule 4: when time_to_expiry <= 5m AND price >= band low AND price <= band high AND edge.btc_1m.change_5m < 0 AND edge.btc_1m.velocity_1m < 0, buy no.

The only things that changed across the four candidates were entry size (10 or 5 contracts) and the entry band (0.45–0.55 or 0.35–0.65). Everything else — the momentum signals, the 5-minute entry window, the near-close exit, and the -4.5 stop — was held fixed. Each of the four was run over a 30-day window. One structural note: the base configuration's risk block carries a price ceiling of 0.59, but the buy rules themselves are gated by the band value, so the effective entry filter is whichever band the variant specifies. The wider-band variants therefore allowed entries up to 0.65 in rule logic even though the base risk block lists a lower ceiling. That mismatch is a property of the configuration as written, not a modeling choice we made.

Each successful variant is saved as a runnable Turbine strategy. The four strategy IDs are 67f6d4298806 (wider band, 10 contracts), 8933e4e18afa (smaller and wider), ed3168c2e554 (baseline), and 7fdb8d729a07 (smaller entries).

Top results

Ranked by ROI over the window:

RankVariantROITotal PnLTradesWin rateSharpeMax drawdown
1Wider band: 10 contracts, 0.35–0.654516.86%4516.86621855.0%1.12-312.99
2Smaller and wider: 5 contracts, 0.35–0.654298.37%4298.37766659.7%1.41-148.70

The wider band is the clear driver. Both wider-band variants more than doubled the trade count of the baseline, which is what you would expect when you loosen the price filter from 0.45–0.55 to 0.35–0.65. The winner on raw ROI was the 10-contract wide band, but its drawdown was the worst of the four at -312.99 and its win rate was the lowest at 55.0%. The 5-contract wide band gave up roughly 5% of ROI relative to the winner while cutting max drawdown by more than half and posting the best Sharpe in the batch at 1.41. On a risk-adjusted basis, that is the more interesting result, even though it finishes second on the leaderboard.

Bottom results

RankVariantROITotal PnLTradesWin rateSharpeMax drawdown
3Baseline: 10 contracts, 0.45–0.552781.79%2781.79270359.9%0.99-160.69
4Smaller entries: 5 contracts, 0.45–0.552252.87%2252.87334160.7%1.06-91.40

Here is where the original thesis runs into trouble. If the goal was to reduce fee drag by trading smaller, the smaller-entry variant at the same band did not translate that into better returns — it finished last on ROI. It did have the smallest drawdown in the batch (-91.40) and the best win rate (60.7%), so it is not a bad strategy; it simply earned less. Meanwhile the two results that surprise us are unresolved:

  • The smaller-entry variant at the same 0.45–0.55 band executed 3341 trades versus the baseline's 2703. If all rules are identical except position size, the trade counts should be similar. They are not, and we do not have a confirmed explanation for that discrepancy. We are flagging it rather than inventing one.
  • The wider band roughly doubled trade counts, which is directionally sensible, but the smaller-and-wider variant hit 7666 trades versus the wider-at-10 variant's 6218 — again more trades on the same band and same signals, only differing in size. That gap is also unresolved.

To be plain: the smaller-entry dial did not do what the thesis expected it to do. ROI fell and trade counts moved in ways we cannot currently account for. We are not going to explain those numbers by guessing at an accounting mechanism or window length. They are open questions.

Conclusion

Two takeaways, stated carefully. First, widening the entry band mattered far more than shrinking entry size. Both wide-band variants beat both narrow-band variants on ROI, and the wide-band 5-contract version delivered the best Sharpe with roughly half the drawdown of the wide-band 10-contract version. If any single variant from this batch looks most balanced, it is that one: 4298.37% ROI, 1.41 Sharpe, -148.70 max drawdown, 59.7% win rate. Second, the size-reduction hypothesis is not supported here. The 5-contract narrow-band variant finished behind the baseline on ROI, and the trade-count discrepancies across the batch are unexplained.

A warning on how to read these numbers: this was a four-variant batch, not a sweep or permutation test. Four candidates is not a robustness test, and the winner's 4516.86% ROI over 30 days should not be treated as evidence that this configuration generalizes. Different windows, different fee schedules, or different risk budgets can make returns look very different, and a single 30-day simulation is a thin basis for confidence. The honest next step, if this were pursued further, would be to run the wide-band configuration across multiple disjoint periods and separate fee assumptions before giving the entry-band change any weight. The unresolved trade-count gaps should be investigated first, because they suggest something in the execution path is not behaving the way the rules imply.

Long disclaimer: The results in this report are from historical, backtested simulations run on recorded market data for the Kalshi KXBTC15M contract series and Coinbase BTC-USD reference data. They are provided for research and informational purposes only and do not constitute financial, investment, tax, or legal advice, nor an offer or solicitation to buy or sell any security, contract, or financial instrument. Simulated performance has inherent limitations: it reflects the assumptions, data, and execution model encoded in the simulation engine, it does not reflect actual trading, it does not account for the full set of real-world frictions, and it cannot capture the behavioral and liquidity effects that occur in live markets. Fees, slippage, partial fills, latency, market impact, data errors, exchange rule changes, and outages are among the factors that can cause live results to differ materially from simulated results, and in some cases the difference can be total. Figures such as ROI, Sharpe ratio, win rate, max drawdown, and total PnL are outputs of a specific simulation configuration over a specific window; they are not statistically validated, not adjusted for multiple-comparison bias, and should not be extrapolated to other periods, markets, or position sizes. Any strategy referenced here may lose some or all of the capital deployed. No representation is made that any variant is suitable for any particular trader or account, and no representation is made about the accuracy or completeness of the underlying data. The reader is responsible for independent verification before acting on anything described in this document.

About this Kalshi strategy backtest

What question does BTC 15m Variants investigate?

For the saved BTC 15m momentum-alignment strategy, test whether smaller entries and a wider entry band reduce the per-trade taker fee drag versus the current 10-contract / 0.45-0.55 setup, holding all rules, signals, and exits fixed.

Read the analysis alongside the individual variant results. Check which entry, exit, price-band or risk settings changed, which settings stayed fixed, and whether the report documents enough historical coverage to support its comparison. A different parameter value defines a different simulation; it does not establish a new source of live trading returns.

How many strategy variants are included?

This report lists 4 completed and 0 failed strategy variants. Failed comparisons do not count as successful tests with zero profit. Review trade counts and drawdown alongside net P&L, win rate and Sharpe, and read any coverage or execution caveats in the analysis. A result based on very few trades needs particular care, even when its return looks high.

Does a profitable Kalshi backtest predict live performance?

These are historical simulations. Live results can differ because of liquidity, fees, slippage, latency, market resolution, outages and data quality. Selecting the best result after comparing many variants can also overstate how reliably it will perform on new data. Compare nearby settings, inspect losing variants and test on a separate period before considering a live run. The report is research, not a promise of profit or investment advice.

How can I inspect or test a strategy from this report?

Use an available Open strategy link to inspect that variant. Review its rules, market, date window, fill assumptions, fees and trade history before changing parameters or rerunning it. Confirm which data and execution assumptions the selected market actually supports; a replay does not establish that every historical order could have filled live. For the workflow, read the Kalshi and prediction market backtesting guide. The Kalshi backtesting software page explains the product's supported testing workflow.

This report is generated from historical simulations. Backtests can be wrong or incomplete, and live trading can differ materially because of liquidity, fees, slippage, latency, market resolution, outages, and data quality. Do your own review before running any strategy.