# ROT Sizer: H1 on the time machine (v2, 7 Oct 2026, 13:35 Bangkok)

## Read this first
- **v2 changes the answer. With the 20 day history gate (W.R1 as Winners wrote it), H1 makes money in BOTH eras and every +-20% neighbour is positive. It still fails the drop bar: worst drop -78% since 2023.**
- Rule SZ.H1c (md5 fa52ff5ad7ec94f417e98d4e077f9eac, registered 13:15 Bangkok): W.R1 picks (up 35%+ in 7 days, cap under $20M, young, $500k volume, at least `hist` days on the CMC list), only while BTC is up 30 days; 10% per coin, max 10; sell at -10% from entry or 25% under the confirmed high; 90 day cap; bad print guard. Engine v1 md5 1264f404a8402bf249855c114a8c19e8 (Coder's 06:23 UTC update), 1 day delay, 1% per side, holdout locked, `--rebalance daily --drift-tol 0.05`.

| centre cell (cut 10%, trail 25%, 10% per coin) | 2020 to 2022 | 2023 to 8 Jul 2026 |
|---|---|---|
| hist 20 + BTC gate (H1c) | **+378%, worst -73%** | **+292%, worst -78%** |
| hist 7 + BTC gate (H1b) | +5,914%, worst -71% | -10%, worst -92% |
| hist 20, no BTC gate | +80%, worst -92% | -6%, worst -96% |
| hist 7, no BTC gate | +1,462%, worst -94% | +218%, worst -91% |
| hold BTC | -36%, worst -77% | +275%, worst -53% |
| basket v3 | +20% | +6% |

- H1c trades 2023 to now: 514, 22% wins, avg +9.6%, median -14.2%, PF 1.62, top trade 16% of profit. 2020 to 2022: 252, avg +6.3%, PF 1.49.
- **H1c sweep, 27 cells (cut 8/10/12%, slot 8/10/12%, trail 20/25/30%):** 2023 to now +43% to +347%, ALL 27 positive; 2020 to 2022 +55% to +3,054%, ALL 27 positive. Beats hold BTC since 2023 in 5 of 27 cells (best: trail 20%, cut 10%, slot 10%: +347%, worst -82%). Worst drop since 2023: -71% to -91% in every cell.
- **Bar check (pre-registered):** beats BTC and basket v3 at the centre in both eras: PASS (+292% vs +275% is a thin margin). Neighbours positive in both eras: PASS. Worst drop better than -60%: FAIL.
- Honest caveats: the 20 day gate came from Winners before this run, but it is one of several gates tried today (hist 7 and 20, BTC gate on and off: 4 looks). The two gates flip which era wins, so the rule is sensitive to its entry filter. Auditor: please count all SZ.H1, H1b, H1c cells as tries.
- Engine note: Coder's engine changed at 13:23 Bangkok between my runs. Every number above is on the new engine.

## What this means for sizing
- 10 coins at 10% each still drops 78%+ at some point since 2023 (which months: next check). The BTC 30 day gate does not stop it.
- Next test (Task 2, needs the supervisor's go): a book level brake (stop new buys after the book falls 30% from its high, restart on a new BTC 30 day high) and 5% slots. Goal: worst drop better than -60% without losing the 27/27 positive grid.

## Earlier today
- Trade level W.R1 exits: sizer-w1.md. Basket v3 mechanics: sizer-t1.md.
- Carry over (not redone): SW kit book 50% trail tuned; tighter and wider trails, moonbag, re-entry after stop, second ticket on strength killed there.

## Handover
- rot-box /opt/rot/sizer/rules/SZ_H1.py (frozen /opt/rot/db/rulecode/fa52ff5ad7ec94f417e98d4e077f9eac.py). Run with `--params '{"hist":20}' --rebalance daily --drift-tol 0.05`. Note: `--params` is ignored when `--grid` is set; put fixed settings inside the grid.
- Runs: /opt/rot/runs/SZ_H1_fa52ff_20261007_062424 (27 cell sweep), SZ_H1_fa52ff_20261007_062910 (gate check).

Version: v2, 7 Oct 2026 13:35 Bangkok. Latest: http://159.223.88.168/hx-c9ois/rot/sizer-h1.md