ForexBestRobots V230 — Fractals + Alligator H1 first confirmed-level breakout

Educational retrospective historical simulation, not a broker statement.
The five unchanged HistData ASCII M1 EURUSD archives are the retained source
of the preceding H4 study. See source/source-manifest.json for provenance and
SHA-256 hashes. This H1 aggregation and strategy replay is new, not the feed.
Fixed source EST is UTC−05:00: add five hours, without DST adjustments.
Only exact duplicate minutes are removed; conflicting duplicates abort.
Do not interpolate missing minutes or empty H1 intervals. Year 2022 initializes
indicators. Entries are restricted to UTC 2023–2025. January 2026 permits exits.

Requires Python 3, numpy and pandas. From this extracted folder run, in order:
python prepare_data.py
python run_research.py
python verify_research.py

Exports are written under assets/ and research/. The frozen protocol and its
SHA-256 precede the results. The reference uses a separate original-file H1
aggregation, floating SMMA, vectorized strict fractals, completed trend episodes
and separate minute execution. Agreement does not imply an independent feed.
Special cases include strict/equal/unconfirmed fractals, first failed breakout
consuming an episode, eighth-versus-ninth-bar eligibility, gaps and time exits.

The strict central fractal is usable only after two observed right-hand candles
close. F labels the earlier central interval; K labels the second right candle.
At the third ordered Alligator close, or later, lock the first newly confirmed
eligible directional level beyond displayed Teeth. Keep it for at most eight
subsequent observed H1 bars. Only its first breakout may qualify; a newer
fractal does not replace it, and a failed closing confirmation consumes it.
These locked-level and lifetime rules are explicit study choices, not a claim
that MetaTrader specifies this complete trading system. No optimisation was
performed after the fixed protocol. The later 2025 segment is retrospective,
not a blind out-of-sample test or a live forward result.

Minute stamps locate observations within minute intervals, not exact ticks.
Bid OHLC does not reconstruct a complete executable Bid/Ask path. Data gaps
can hide crossings; 2023 has notably fewer minutes than 2024–2025. Partial bars
remain. Simulated last chart candles exclude the remaining exit-minute range.
An adverse stop gap uses the next observed Open; a favorable target gap is
capped at target. With both levels inside one minute, stop is assumed first.
At the timeout Open, opening level gaps take precedence; later range is unused.

1R is the initial Bid-to-stop distance. Sum R is not account percentage return.
Round-trip and per-UTC-midnight charges are allowances, not measured spread,
slippage or swaps. No swap income, triple rollover, lot, leverage or margin
model. Base +2.51R is thin; stress −11.84R and later 2025 −1.25R remain visible.

Definitions: official MetaTrader 4 Fractals / Alligator help and HistData
detailed file specification. All public chart and trade values are derived
from these retained source records; no synthetic Forex candles are used.
