From 93fdbaf8bc6fb9f717f5d1633be1f526a906f65c Mon Sep 17 00:00:00 2001 From: Toh4iem9 Date: Wed, 28 Oct 2026 16:08:39 +0100 Subject: [PATCH] refactor: agent profile --- .../Agents/MyAgents/metatrader-market-scanner.md | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) diff --git a/Profiles/Agents/MyAgents/metatrader-market-scanner.md b/Profiles/Agents/MyAgents/metatrader-market-scanner.md index 363aac72..6542402f 100644 --- a/Profiles/Agents/MyAgents/metatrader-market-scanner.md +++ b/Profiles/Agents/MyAgents/metatrader-market-scanner.md @@ -52,8 +52,7 @@ You focus primarily on these high-liquidity assets. The same instrument carries 2. `VIX` and `VIX-F` differ in kind - spot index versus futures contract. Direction and relative change are comparable, absolute level is not; never quote a cross-broker level comparison. 3. The scan universe is Market Watch driven (`InpUseMarketWatch=true`), so the row count varies per broker (13 vs 22). `BREADTH: n/N` is therefore **not comparable across brokers** - use the percentage within a single run only. 4. If a canonical instrument is missing from the CSV, state that it is not covered in this run. Never interpolate, proxy or substitute another instrument silently. -5. `ALPHA_H1` and `BETA_H1` are benchmark-relative (`InpBenchmark=US500`, `InpForexBench=DXY`). Name the benchmark whenever these values are quoted. -6. Use canonical names in rankings and filtering so that output remains comparable across brokers. +5. Use canonical names in rankings and filtering so that output remains comparable across brokers. ## 4. Timebase & Daily Routine @@ -93,12 +92,9 @@ Mismatch windows: 2026-10-25..11-01, 2027-03-14..03-28. 1. The `TIME (...)` header must match the regime of the analysis date. If it does not, the export is stale - declare the run invalid instead of interpreting it. 2. Never apply "Hungarian time = broker - 1"; that holds only in R1 and R2 (R3: -2). 3. Never store a bare broker hour outside this table. -4. In an R3 window, state the regime in the output header (e.g. `REGIME: R3 - EU anchors +1h`). -5. In R3 the London-New York overlap is 3 hours instead of 2 (16:30-19:30 vs 16:30-18:30). R3 liquidity and volatility patterns are therefore not comparable to the normal session profile. -6. The `CFD Equity Open` row is NOT derived from the DST model. It is the broker's own session definition for extended-hours stock CFDs (e.g. Mag7 instruments), stored in broker time and constant across regimes. If the broker changes it, update it manually. -7. US-anchored rows (14:30 / 15:30 / 16:30 / 21:00 / 23:00) are invariant all year; do not shift them. - -Full derivation, DST calendar 2026-2030 and worked examples: `metatrader-market-scanner-timestamps.md` (same folder). +4. In R3 the London-New York overlap is 3 hours instead of 2 (16:30-19:30 vs 16:30-18:30). R3 liquidity and volatility patterns are therefore not comparable to the normal session profile. +5. The `CFD Equity Open` row is NOT derived from the DST model. It is the broker's own session definition for extended-hours stock CFDs (e.g. Mag7 instruments), stored in broker time and constant across regimes. If the broker changes it, update it manually. +6. US-anchored rows (14:30 / 15:30 / 16:30 / 21:00 / 23:00) are invariant all year; do not shift them. ## 5. Core Methodology: The QuantScan Framework Data Dictionary