Do Resting Limit Orders Get Better Fills on 0DTE Puts? A Trade-Tape Replay
Can a limit order that waits for a better price do better than crossing the spread on 0DTE puts? An Enterprise study.
Guides, explainers and comparisons on options analytics: dealer positioning, volatility, flow and the FlashAlpha API, for traders and developers.
Looking for tested results? Studies are listed under Backtests.
9 articles and backtests tagged #Execution×
Can a limit order that waits for a better price do better than crossing the spread on 0DTE puts? An Enterprise study.
We sold 0DTE puts on SPY, QQQ and IWM every 30 minutes from 10:00 to 14:00 ET on sell-window sell days, 2018 to 2022, and tested five intraday options-flow signals as skip rules and 15 exits. No flow signal and no exit beat the plain trade; the strike and the fill mattered more.
We repriced 1.35 million backtested short-premium trades on 11 ETFs and stocks from 2018 to 2026 under five fill scenarios priced from archived quotes, measured what the quoted spread and a one-cent slip cost each design, and tested three fill habits with a plan written before the results.
A complete, paper-first reference implementation: poll FlashAlpha dealer positioning and flow signals via the Python SDK and route defined-risk options orders through Interactive Brokers with ib_insync/ib_async - brackets, position caps, a daily kill switch, and the disconnect/partial-fill paths handled in ~100 lines.
The fill model is the largest single free parameter in any options backtest - our published 8-year studies show the same signal swinging from -5% to +62% total return on the fill assumption alone. The four-rung fill ladder, how to calibrate alpha from replayed point-in-time spreads, and a recommended default per strategy family.
Same VRP strategy, same signal, same 8 years, same trades. Change only the fill model and SPY total return swings from -5% to +62%. The "edge" lives in the fill assumption, not the signal.
Single VRP short-put-spread strategy, tuned once on SPY, frozen, deployed unchanged on QQQ/IWM/AMZN/NVDA/SPXW with honest fills. The 70% win rate held; under post-and-wait fills the Sharpe did not, while the companion fill-model study shows mid-fills flip the result - execution, not the premium, decides the outcome.
Seven years of 1-minute SPY chains. 96 grid cells. Honest fills, signal-gated sizing, drawdown breaker. Same cell ran +5,400% with a stop loss and -100% without. Multi-DTE pooling somehow underperformed single-DTE. SL=200% turned out worse than no stop at all. The best signal in 16k trades was a one-line flag that beat my 3-layer composite. The Calmar-leader cell is in the grid - I'm telling you the shape, not the coordinates.
Practical guide to colocation for options algo trading - costs, latency benchmarks, and when co-location actually improves execution vs wasting capital on unnecessary infrastructure.
Studies that test one trading question on historical data and report what happened, grouped by the plan that opens the full results.