The first study we published only counted two reactions at a gamma level: the fade, where you lean against the move at the line, and the breakout, where you go with the first one-minute close through it. A trader who watches these levels every day pointed out the case we had skipped, and it is the one that happens most: the level breaks, price comes back to it from the other side, and it holds there. A resistance broken becomes a support. We added that reaction to the harness and replayed the whole archive again.
How the retest was measured
- 88 SPX sessions replayed on ES and 82 NDX sessions on NQ, one-minute futures bars, the index level mapped with the previous minute’s basis, costs included, the stop checked before the target inside every bar.
- A break is the first one-minute close through the level, with a buffer of 0.10 % of spot. The retest is the first bar that comes back to the level within a window — 30, 60 or 120 minutes — played in the direction of the break: a broken resistance is bought, a broken support is sold.
- Stop 0.10 % of spot, target 1.5 times the stop, a placebo at the same level shifted three to eight strikes, and a 70 / 30 split by date so the second half of the archive could disagree with the first.
What it did
Across every gamma level pooled, the retest won 66 % of its trades on ES with a profit factor of 2.53, and 55 % on NQ with 1.68. The fade at the same levels, over the same sessions, ran 0.91 and 1.18. The placebo retest was 1.20 on ES and 1.47 on NQ — so part of the edge is the pattern itself, a pullback after a break in the direction of the move, and the part that belongs to the level was larger on ES than on NQ.
The window mattered. Retests inside 30 minutes did better than retests inside two hours on both indices; beyond half an hour the return to the level looked more like a coincidence than a reaction. On NQ the volume zero gamma carried 80 of the 208 retests and ran 1.87 against a placebo of 1.22, with both halves of the split above one. Those are the two numbers we trust most in the whole table.
The playbook that came out of it
- Read the regime. Positive net OI gamma: buy the dips at the levels. Negative: sell the rallies. Fading against the regime lost on both indices.
- A one-minute close through a level ends the fade on that level for the day. Do not chase the break — the breakout alone ran under one everywhere.
- Wait for the return. The first bar back at the level within 30 minutes, played in the direction of the break, was the best reaction the archive held.
- Stop 0.10 % of spot, target 1.5 R. Tighter stops printed higher win rates and the placebo rose with them; the gap to the placebo was widest at 1.5 to 2 R.