# ROT Sizer: H1-ALT-B, book brake and slot size (v1, 7 Oct 2026, 14:00 Bangkok)

## Read this first
- **No cell passes both bars. Smaller slots fix the drop, but then the rule no longer beats hold BTC since 2023. The book brake does not help.**
- **5% x up to 20 coins (no brake):** worst drop -52% since 2023 (passes the -60% bar), but only +154% vs hold BTC +275%. 2020 to 2022: +976%, worst -44%.
- **10% x 10 coins (the PAPER cell):** +485% since 2023 (beats BTC), worst -65% (fails the drop bar).
- **5% x 10 coins:** +90% to +156% since 2023, worst -59% to -62%.
- **Book brake (no new buys 25% or 30% under the book high, lift on a BTC 30 day high): KILLED.** It changes results by a few points either way and makes the 10% cell worse at 25% (-67% vs -65%).
- **New fragility found: the engine's rebalance tolerance moves the PAPER cell a lot.** Same rule, `--drift-tol 0.05` gives +373% / -65% since 2023; `--drift-tol 0.04` gives +485% / -65%. 2020 to 2022 moves from +846% to +499%. Held coins get trimmed back when they move more than the tolerance in a day, so the tolerance acts like a hidden "trim winners" setting. Coder and Auditor: the paper book must use one fixed tolerance, and the rule should hold coins without daily trims.
- Why 5% x 10 failed at first: `--drift-tol 0.05` skips any trade of 5% of the book or less, so 5% buys never filled (all zeros). Rerun with 0.04.

| size, brake (drift-tol 0.04) | 2020 to 2022 | 2023 to 8 Jul 2026 |
|---|---|---|
| 10% x 10, no brake | +499%, worst -36% | **+485%, worst -65%** |
| 10% x 10, brake 25% | +367%, worst -34% | +471%, worst -67% |
| 10% x 10, brake 30% | +463%, worst -36% | +448%, worst -64% |
| 5% x 10, no brake | +165%, worst -30% | +90%, worst -62% |
| 5% x 10, brake 25% | +165%, worst -30% | +99%, worst -61% |
| 5% x 10, brake 30% | +165%, worst -30% | +156%, worst -59% |
| **5% x 20, no brake** | **+976%, worst -44%** | **+154%, worst -52%** |
| 5% x 20, brake 25% | +916%, worst -47% | +164%, worst -51% |
| 5% x 20, brake 30% | +1,017%, worst -42% | +154%, worst -52% |
| hold BTC | -36%, worst -77% | +275%, worst -53% |

- Rule SZ.H1ALT-B (md5 587e1992881eee573b069724e7172cb2, registered 13:42 Bangkok before the run, 9 variants, 2019 to now only, 2017 to 2018 never touched). Engine v1 md5 1264f404a8402bf249855c114a8c19e8, 1 day delay, 1% per side, holdout locked.
- SZ.H1ALT (centre: 10% x 10) moved to PAPER at 13:43 with the Auditor's FIX receipt (PAPER yes, CHAMPION no).

## The plain choice for Ben
- **Want to beat BTC:** 10% per coin, up to 10 coins. Expect drops of about two thirds along the way.
- **Want drops near half:** 5% per coin, up to 20 coins. Expect less than holding BTC since 2023, much more than BTC in 2020 to 2022.
- Both trade only in alt season with BTC rising; the rest of the time the money sits in cash.

Runs: rot-box /opt/rot/runs/SZ_H1ALTB_587e19_20261007_064809 (drift-tol 0.04), SZ_H1ALTB_587e19_20261007_064337 (drift-tol 0.05, 5% cells did not fill).

Version: v1, 7 Oct 2026 14:00 Bangkok. Latest: http://159.223.88.168/hx-c9ois/rot/sizer-h1altb.md