Win Rate Filters

BTC 15-Minute Strategy: Single-Change Variant Study (Kalshi)

Short Disclaimer

This is historical simulation research on a Kalshi crypto market, not investment advice and not a prediction of future results. Past simulated performance does not mean future profits. Read the full disclaimer at the bottom.

Intro / Thesis

The question here was narrow and practical. The existing BTC 15-minute strategy (series KXBTC15M) has a certain win rate and net P&L, and we wanted to know which one of three small edits actually improves those outcomes versus simply reducing activity. A strategy that trades less and earns less tells you nothing useful. A strategy that trades less and earns more — or earns more on nearly the same volume — is worth understanding.

So we ran four candidates, each with its own complete rules over a 30-day window, each evaluated every 10 seconds:

• Baseline: the current saved strategy, unchanged.
• Tighter spread gate: the spread ceiling on the four mid-band entry rules cut from 0.01 to 0.005.
• Higher tier-three take-profit: the exit threshold on positions above 30 contracts raised from $5.00 to $8.00.
• Higher acceleration floor: the price floor on the two acceleration entry rules (rules 10 and 11) raised from 0.35 to 0.40.

Only one change at a time was applied, so any difference in results is attributable to that change alone within this window.

Variant and Strategy Explanation

For context, the strategy enters small 10-contract positions during the 2-to-7-minute time-to-expiry window when a set of BTC momentum and price conditions line up, and it exits through position-size-tiered profit and stop rules. Exits are the same across all variants except the tier-three change.

The three variants:

1. Spread gate at 0.005. Rules 8, 9, 10, and 11 — the four entry rules that trade the 0.35–0.65 or 0.35–0.80 price band — now require the quoted spread to be 0.005 or tighter instead of 0.01. Rules 12 and 13 keep their 0.01 gate, and rules 14 and 15 keep 0.02. This is meant to favor entries where the market is tight and the fill is closer to fair value.

2. Tier-three take-profit at $8.00. Rule 3, which closes positions above 30 contracts once unrealized P&L clears a threshold, now requires $8.00 instead of $5.00. Everything else, including the tier-three stop at -$15, is unchanged. This lets the largest positions run further before taking profit.

3. Acceleration floor at 0.40. Rules 10 and 11, the acceleration entries, now require price of at least 0.40 rather than 0.35. This is the only rule these two entries touch; the unchanged 0.80 ceiling stays in place. Fewer marginal, low-priced acceleration trades are expected.

The base rules are context only. Each candidate above is a complete, self-contained strategy with its full rule set.

Top Results

Ranked by total net P&L over the 30-day window:

The winner is the tighter spread gate. It earns about $123 more than the baseline on roughly 227 fewer trades, with a win rate nearly two points higher (86.76% vs 84.98%). It also posts the best Sharpe (0.91 vs 0.86). The cost is a deeper max drawdown, -$159.68 versus -$146.59. So the improvement is real within this window, but it did not come for free: the variant accepted more downside along the way.

The spread-gate variant is saved as a runnable Turbine strategy (slug the-best-one-100-dollar-test-d1346e4e6553), as are the other variants below.

Bottom Results

The two bottom-ranked changes are the least convincing.

Acceleration floor at 0.40 is, in this run, indistinguishable from the baseline. Same P&L ($1,899.31), same win rate (84.98%), same trade count (9,111), same Sharpe (0.86), same max drawdown (-$146.59). That is surprising: raising the entry floor should have removed some acceleration trades, so identical trade counts are hard to square with the change. We do not have an explanation for this and are flagging it as unresolved rather than inventing a mechanism. Treat this variant as having no demonstrated effect in this window.

Tier-three take-profit at $8.00 is the weakest result. It earns $1,874.75, about $25 less than the baseline, with a slightly lower win rate (84.82%) and Sharpe (0.84), on essentially the same trade count (9,109). It does have the smallest max drawdown at -$143.59, so letting large winners run longer may trade a little P&L for a little less peak-to-trough pain — but this single change did not improve the bottom line.

One note on ROI figures: the reported ROI percentages (in the thousands) sit oddly against the dollar P&L and trade counts. We have not reconciled how these ROI values were computed and consider the discrepancy unresolved. Judge the variants on net P&L, win rate, and drawdown, which are internally consistent.

Conclusion

Across four candidates over one 30-day window (this explicit batch, no sweep, no permutation test):

• The tighter 0.005 spread gate is the only change that clearly improves both win rate and net P&L rather than simply trading less. It earns more, wins more often, and has the better Sharpe, at the cost of a somewhat larger drawdown.
• The 0.40 acceleration floor shows no measurable difference from the baseline in this run, and the identical trade counts are unexplained. Do not read the tie as proof the change is harmless; it just means this window did not distinguish the two.
• The higher tier-three take-profit modestly underperforms on P&L and win rate, though it has the lowest drawdown.

Each successful variant is saved as its own runnable Turbine strategy. Every variant in this set — the winning spread gate, the baseline, the acceleration-floor variant, and the take-profit variant — ran as a separate strategy and can be run again.

This was a single 30-day window on one market. The winner's edge is real in the data but modest in dollar terms relative to the drawdown it carried, and a different period or risk budget could easily reorder these results. Nothing here should be read as statistical proof or as a statement about future returns.

Long Disclaimer

This report describes historical simulation research conducted on Kalshi's KXBTC15M crypto market using the Turbine strategy framework. All figures are the output of backtested simulations over a 30-day window. Simulation results are inherently limited: they depend on the specific historical period, the modeling assumptions of the simulator, and the market conditions captured in that window. They are not a record of live trading, they do not account for every real-world execution effect, and they do not generalize to other periods, market regimes, or sizes.

Nothing in this report is investment, trading, financial, legal, or tax advice, and nothing should be construed as a recommendation to buy, sell, or hold any instrument. Trading prediction markets involves substantial risk, including the possible loss of all capital deployed. Crypto-related markets can be volatile and illiquid, and prices can move sharply against a position. Past performance — simulated or real — is not indicative of future results, and no representation is made that any strategy described here will be profitable in the future or will avoid losses.

This report reflects a single explicit batch of four variant comparisons. No parameter sweep and no permutation or significance testing was run, so no statistical claim of robustness is made or implied. A single-change design isolates one edit per candidate within this window, but it does not establish causation across markets or time. The ranking of variants is specific to this dataset and should not be extrapolated.

Where metrics appear inconsistent or surprising, this report states that the discrepancy is unresolved rather than offering an explanation not supported by the data. Readers should treat any unreconciled figures with appropriate caution and verify results independently before relying on them.

The author and publisher make no warranty, express or implied, as to the accuracy, completeness, or fitness for purpose of the information herein. They accept no liability for any loss or damage arising from use of this report. You are solely responsible for your own due diligence, for understanding the venue's rules and risks, and for consulting qualified professionals before acting on any information in this document. Nothing here constitutes an offer or solicitation in any jurisdiction where such an offer would be unlawful.

About this Kalshi strategy backtest

What question does Win Rate Filters investigate?

Compare the current BTC 15-minute strategy against three single-change variants — a tighter 0.005 spread gate, a raised tier-three take-profit, and a higher 0.40 acceleration price floor — to see which change actually improves win rate and net P&L rather than merely trading less.

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.