Kalshi KXHIGHNY 72–73°F Market Research Report

Short disclaimer

This report describes a historical simulation exercise only. No live trading occurred. Past results—especially zero-trade outcomes—do not suggest future profitability. All strategy logic is based on US National Weather Service (NWS) data and Kalshi contract mechanics; results are hypothetical.

Intro / thesis

Weather markets on Kalshi sometimes price daily high-temperature contracts with surprising confidence. The KXHIGHNY 72–73°F bucket market trades a simple binary question: will La Guardia’s (KLGA) official high land inside the 72–73 °F window? My thesis was that market participants occasionally bid YES well above fair value when every available signal—specifically the NWS forecast high and the current observed temperature—points comfortably below the bucket floor. If YES is still expensive while the NWS high-temp forecast is 71 °F or lower and we are not yet at 70 °F on the day, then a NO position provides a defined, short-lived edge. My research tested whether anchoring the entry strictly to a forecast-below-bucket filter (≤71 °F) improved reliability versus looser “observed‑vs‑forecast” temperature-gap logic.

Variant and strategy explanation

All variants share the same core rule set derived from the base DSL:

• Entry (forecast_below_bucket_buy_no): Buy NO contracts only when every condition is true:
  • YES price > 0.50  (market is confident the bucket will hit).
  • Current KLGA temperature < 70 °F.
  • NWS KLGA forecast high ≤ 71 °F (firmly below the 72 °F bucket floor).
  • At least two hours remain until expiry.
• Size: 3 contracts per signal (max position 10; price floor 0.10, ceiling 0.90; max loss 15.0).
• Exits:
  • Profit: Sell all if YES re-prices below 0.30.
  • Temperature breach: Sell all if current KLGA temperature reaches 72 °F or higher (bucket floor is threatened).
  • Stop-loss: Hard exit at –15.0 unrealized P&L.
  • Time stop: Close any remaining position with ≤ 30 minutes to expiry.

The 100 completed variant runs kept the entry logic identical but explored subtle parameter shifts—different look‑back windows, refresh cadences, or exit thresholds—to see if any produced different behavior in historical simulation. Because the base filter (forecast high ≤ 71 °F) is restrictive by design, the research effectively stress-tests when the market ever gave an entry signal at all under real historical NWS data and Kalshi pricing.

Top results

Every top‑ranked variant produced identical zero-activity profiles:

Each successful variant is saved as a runnable Turbine strategy (strategy slug visible in the raw data).

In plain terms: the historical data never triggered a single trade for any of the 100 tested variants. The conjunction of conditions—YES above 0.50 while the official NWS forecast sits at or below 71 °F and the current temperature is still under 70 °F—was apparently too stringent to occur in the backtest window.

Bottom results

The bottom of the ranking looks identical to the top for the same reason. The lowest-ranked variants also generated zero trades, zero P&L, zero drawdown, and zero Sharpe. There is no performance dispersion because no variant ever took a position. Ranking differences are purely an artifact of ID ordering, not of any return or risk metric. All bottom variants likewise carry zero win rates and zero ROI.

Conclusion

The forecast-below-bucket filter (NWS KLGA forecast high ≤ 71 °F) proved to be an extremely restrictive entry gate. Across 100 parameter variants and the full available historical simulation data, the strategy never received a valid entry signal. That does not mean the logic is faulty; it may simply indicate that Kalshi market makers rarely price YES above 0.50 on days when the official forecast is already below the bucket. In other words, the very overconfidence the strategy sought to exploit did not historically appear in the data under these precise constraints.

For a researcher, the key takeaway is that the filter works too well as a gate—it may be filtering out the entire opportunity set. A live or forward test would need a less strict entry trigger (e.g., forecast ≤ 72 °F combined with a smaller YES premium) to see any trade volume. Alternatively, a different weather market or a longer historical dataset might reveal enough instances. The current configuration is conservative to the point of inactivity.

Long disclaimer

This document is a historical strategy research report produced by Turbine’s simulation engine. All results are derived from hypothetical backtesting using NWS KLGA observations and Kalshi market data that were available during the test period. No real money was deployed, and no actual trades were placed.

The zero-trade outcome described here does not imply that the strategy would never trade in the future or that it is fundamentally unsound. It reflects the specific intersection of backtest data, entry filters, and timing assumptions used in this simulation. Real-world factors—data delays, order-book depth, fee schedules, partial fills, NWS forecast revisions, and market microstructure—could materially alter outcomes.

Trading prediction markets involves substantial risk of loss. Past simulation results are not guarantees of future performance. This report does not constitute investment advice, a recommendation to trade any specific contract, or a promise of any particular return. Readers should perform their own diligence and consult a qualified financial advisor before engaging in any live trading.

The strategy DSL and each successful variant have been saved inside the Turbine platform for further testing or paper trading. By accessing this report, you acknowledge that all trading decisions are your own, and you bear full responsibility for any gains or losses.