Level 5 Gate Study

Level-5 Entry Study on Kalshi 15-Minute BTC: Timing Window and Strike-Distance Gates

Short disclaimer: This is a historical simulation report, not investment advice. Nothing here is a promise or projection of future results.

Intro / thesis

This study looks at one narrow question inside a single entry template on Kalshi's 15-minute BTC markets (series KXBTC15M). The template — call it the "level-5" entry — only fires when several things line up at once: price is on the correct side of 0, the spot price is far enough from the strike, the 15-minute EMA is on the same side of the strike, the 5-minute BTC momentum agrees, and the 1-minute velocity agrees or is flat. On top of that the entry has to land in a 0.90–0.99 ask band with a spread no wider than 0.02.

The thesis was simple: how much does the timing window and the strike-distance gate matter for this specific filter stack? I held everything else fixed — the 0.35 stops, the spot-crosses-strike exits, the 0.995 take profits, the ask band, the spread cap, 50-contract size and 50-contract max position — and only varied two numbers:

• The 8-minute entry window with the 70-dollar gate (the saved baseline).
• A 5-minute entry window with the same 70-dollar gate.
• The 8-minute window with a wider 80-dollar gate.
• The 8-minute window with an even wider 100-dollar gate.

This was an explicit batch of three variants, all completed. No sweep, no permutation test. The base DSL is context only; each candidate has its own full rules, market, and window.

Variant and strategy explanation

Every variant runs on Kalshi KXBTC15M, selecting the most liquid market. The loop checks every 10 seconds. Position limit 50 contracts; YES price bounds 0.05 to 0.99. Rules are evaluated top-to-bottom; the first matching rule fires and ends that tick.

The shared exit and entry scaffolding, in the order it's evaluated:

1. Stop, YES: if yes_position_size > 0 and yes_best_bid <= 0.35, sell YES.
2. Stop, NO: if no_position_size > 0 and no_best_bid <= 0.35, sell NO.
3. Spot-cross exit, YES: if yes_position_size > 0 and derived.spot_vs_strike < 0, sell YES.
4. Spot-cross exit, NO: if no_position_size > 0 and derived.spot_vs_strike > 0, sell NO.
5. Take profit, YES: if yes_position_size > 0, time_to_expiry <= 15m and > 2m, and yes_best_bid >= 0.995, sell YES.
6. Take profit, NO: symmetric, with no_best_bid >= 0.995.

Then the entry itself. In every variant the YES entry requires: position_size < 50, order_count == 0, edge.btc.price > 0, time_to_expiry <= 8m (or <= 5m in Version 1), derived.spot_vs_strike >= <gate>, derived.ema_vs_strike > 0, edge.btc.change_5m > 0, edge.btc.velocity_1m >= 0, yes_best_ask between 0.90 and 0.99, and spread <= 0.02. The NO entry mirrors it with <, <=, and negative thresholds. Size is 50 contracts, immediate_or_cancel.

The three variants differ only in the two numbers under test:

• Version 1: 5-minute entry window, 70-dollar strike-distance gate.
• Version 2: 8-minute entry window, 80-dollar strike-distance gate.
• Version 3: 8-minute entry window, 100-dollar strike-distance gate.

Each of the three successful variants is saved as a runnable Turbine strategy under its own strategy ID and slug (see the result rows below), so it can be re-run or paper-traded directly without rebuilding the DSL.

Top results

Period was 30 days for all three variants.

Version 3 takes the top spot on PnL and ROI, with the fewest trades of the three and a drawdown roughly on par with Version 2. Version 2 sits just below it on PnL with more trades and a slightly shallower max drawdown. Both 8-minute variants share a Sharpe of 0.40.

Saved strategies:

• version 3: strategy_id 0d1696098cd64e3717c05d5189f1effe, slug level-5-gate-study-0d1696098cd6
• version 2: strategy_id 29b5bd5f3cdb42b5dc35b3d00971bace, slug level-5-gate-study-29b5bd5f3cdb

Bottom results

Saved strategy: strategy_id 6e2bad8d46a829e581a1a85c40064faf, slug level-5-gate-study-6e2bad8d46a8.

Version 1 is the weakest of the three by PnL, ROI, and Sharpe. The interesting wrinkle: it has the highest win rate of the batch (96.02%) yet the lowest total PnL and the deepest max drawdown (-119.51). That combination is inconsistent with a simple "higher win rate is better" reading, and the data here does not cleanly explain it. Average loss size, exit timing, and which exit rules dominate the 5-minute versus 8-minute variants would all need a closer, trade-by-trade look. I'm flagging the discrepancy rather than guessing at a cause.

Two of the three variants genuinely differ in only one number from each other (V2 and V3: 80 vs 100 dollar gate, everything else including window identical), yet they trade at noticeably different volumes (434 vs 381) and produce different PnL. That's expected given the tighter gate filters out more near-strike opportunities, but the exact contribution of trade count versus per-trade edge is not resolved by this batch.

Conclusion

In this 30-day window, the two 8-minute, wider-gate variants beat the 5-minute, 70-dollar baseline on PnL, ROI, and Sharpe, and the 100-dollar gate edged the 80-dollar gate. All three carried drawdowns of roughly the same order of magnitude (about -100 to -120 on a 50-contract book), and win rates are high across the board — nothing here is a low-hit-rate strategy.

What this does not establish:

• A single 30-day window. Different periods can invert the ranking.
• This was an explicit three-variant batch, not a sweep or a permutation test. There is no statistical robustness claim here.
• The highest win rate (Version 1) also has the worst PnL and worst drawdown, and I have not resolved why. That discrepancy is open, not explained.
• Max drawdown and Sharpe here are single-path simulation statistics, not guaranteed envelopes.

If you continue with any of these, the practical next steps are: longer out-of-sample windows, a closer look at trade-level loss distribution (especially on Version 1), and a check on how the broader 80 vs 100 gate difference behaves under different volatility regimes. The three runnable strategies above are the place to start.

Long disclaimer

This document is research output from a historical simulation. It is not financial, investment, legal, or tax advice, and it is not an offer, solicitation, or recommendation to buy or sell any security, derivative, contract, or prediction-market instrument on Kalshi or anywhere else.

All figures — total PnL, ROI, Sharpe, max drawdown, win rate, trade counts — are the product of a backtest over a specific 30-day historical window on a specific set of market data, with specific assumptions about fills, spreads, liquidity, latency, and rule evaluation order. Backtests are simplified models. They routinely overstate what a live strategy would have achieved. Real fills can be worse, partial, or nonexistent; real spreads can widen; real liquidity can vanish; real markets can gap through stop levels. Past performance, simulated or actual, does not predict future results.

Several specifics to keep in mind:

• Fill assumptions. Entries use immediate_or_cancel at the ask band 0.90–0.99 with a 0.02 spread cap. A backtest that assumes a fill at the observed ask may not fill in live trading, particularly at 50 contracts in a thin book.
• Exit assumptions. Stops fire on bids at or below

About this Kalshi strategy backtest

What question does Level 5 Gate Study investigate?

Compare the level-5 YES/NO entry (strike-distance gate plus EMA, 5-minute BTC momentum and 1-minute velocity agreement) at three timing/distance settings: 8 minutes with a 70-dollar spot-vs-strike gate (saved baseline), 5 minutes with the same 70-dollar gate, and 8 minutes with a wider 80-dollar and 100-dollar gate. Everything else — the 0.35 stops, spot-crosses-strike exits, 0.995 take profits, ask band 0.90-0.99, spread cap, 50-contract size and 50-contract max position — stays 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 3 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.