A few days ago I asked how to tell whether a GEX filter improves trade selection or just reduces trade frequency. The suggestion I got: track the rejected trades as a shadow book using identical entry, exit and cost rules, then compare the accepted and rejected groups, not just the filtered book against the original total. So I did that on data I already had.
Setup
• Real SPY option chain quotes, Jul 29 - Sep 25, 2026.
• One entry per day for each structure, tested separately, 1 contract, short premium only (sold strangles, credit put spreads), $1 per leg commission, mid fills. Bid/ask fills give almost the same numbers.
• GEX and flip from my own model on the entry chain (dealers assumed long calls / short puts).
• Filter = net GEX > 0 and spot above the flip.…