From strategy to evidence
ALBR, ALSTB and ADB follow the same gates. Their definitions are preserved; backtests and EAs are versioned outputs that can be reproduced.
Locked V1 research baseline
MJME-V1-20261031 preserves the approved definitions and reproducible research. Execution validation and release remain separate.
Before the next stage
- Observed broker bid/ask, independently calibrated clock and GBPJPY metadata
- Compiler/build identity and deterministic EA parity, including restart and shared risk
- Independent forward sample and predeclared performance review
- Owner-reviewed comparison and purpose-specific release
Tracked strategy experiments
5 registered. Each experiment needs a separate version, a stated hypothesis and evaluation plan, then a comparison with V1 on matching data and costs. Measure net performance and risk alongside the number of opportunities.
Keep, reject or revise after review. An experiment never overwrites V1 or automatically becomes a trading release.
Real-time shadow setup
ALBR / ALSTB / ADB source candidates are prepared for MT4 compilation. Use two separate GBPJPY charts for ALSTB and ADB. ALSTB records its three exit variants separately. The package also includes an optional ALBR v1.08 restart-logging repair for its existing pair charts.
Virtual observations only. Confirm the broker clock before session logic starts. Compilation, installation and actual broker execution are not yet verified.
Source package and setup instructions →RMR · Rapid-Move Reversal →
EURJPY · GBPJPY · USDCAD · EURGBP · AUDUSD · Pair-specific London / New York · 15/30
Strategy development
- Research and definitionComplete
- Backtest and risk evidenceIn progress
- EA implementationIn progress
- Research / EA parityPending
- Forward testingPending
- Comparison and reconciliationPending
- Reviewed releasePending
- Versioned iterationPending
Next: Compile and attach RMR1.02 experts using matching presets; keep Levels1.01, confirm input/timeframe reinitialisation and capture named startup diagnostics. Preserve state/counters. Complete forward quote-covered lifecycle reconciliation remains pending.
Research and entry/exit definitions are approved for replay and EA specification. Results acceptance and EA validation remain pending.
Evidence and open decisions
Partial progress records work already done. Completion requires the preceding gates and recorded acceptance; observer startup does not establish demo execution.
- Known raw-feed anomalies and sparse source coverage remain; no corrected candles are invented.
- Broker quote fills, exact compiled binary identity, sizing and supervised demo remain unvalidated.
- Historical first-two ordering and live quote arrival ordering must be reconciled.
Research and definition
- Owner-approved five-pair research definition; execution gates separate.
Backtest and risk evidence
- Inspected raw-feed diagnostic and full source captures; quality/OOS/broker acceptance remain open.
EA implementation
- Order-free five-pair source candidate; native source checks do not establish MetaEditor compilation or terminal execution.
- RMR1.01 shared five-pair virtual EA/indicator source, mismatch-safe presets and setup evidence. 32 native runtime/trace scenarios and186 selected fixtures; MetaEditor compilation, installed identity and forward acceptance pending.
- 1.02 input-restart repair with create-only variable semantics and39 native cases.1.01 compile/initial attach owner-reported;1.02 MetaEditor and terminal restart acceptance pending.
ALBR · Asian London Breakout Retest →
EURUSD · GBPUSD · Europe/London · 20/40
Strategy development
- Research and definitionComplete
- Backtest and risk evidenceIn progress
- EA implementationIn progress
- Research / EA parityPending
- Forward testingIn progress
- Comparison and reconciliationPending
- Reviewed releasePending
- Versioned iterationPending
Next: Compile and attach the automatic CMC-clock replacement from config/mt4/cmc-clock-release-v1.json; confirm the computed offset on fresh quotes, retain old logs and collect matched observations. Actual orders remain disabled.
Research rules are frozen. Executable-contract approval and EA validation are separate.
Evidence and open decisions
Partial progress records work already done. Completion requires the preceding gates and recorded acceptance; observer startup does not establish demo execution.
- Reconcile complete M5 candidate adjudication with real-time bid/ask virtual activation and fills.
- Verify broker seasonal offset, restart event identity and eight-hour exits.
- Recover cost-aware full-session results, independent holdout and risk evidence.
Research and definition
- Owner research-rule freeze recorded; executable release approval remains separate.
Backtest and risk evidence
- Exact recovered historical candidate engine; costs, full-session export and independent acceptance outstanding.
- Local shared preliminary replay with full ledgers and cost sensitivity; hosted research, real execution and untouched OOS acceptance remain pending.
- Owner-approved-definition local M1 price-proxy replay, repeated byte-identically. ALBR unchanged comparator; costs hypothetical, broker fills/EA/OOS acceptance incomplete.
- Exact parent reproduction, 92/92 model clock checks; five isolated exit gaps resolved with provider M5 bounds, corrected time-exit proxy, three separate discretionary estimates. Still preliminary; broker execution and eight conditional additional setups are not certified.
EA implementation
- Exact supplied source, virtual only; installed binary and compilation not independently attested.
- Optional v1.08 source candidate repairs active signal-ID restoration. Exact limited diff preserves virtual trading logic, inputs and v1.07 state namespace. MetaEditor compilation/runtime acceptance pending.
- Versioned automatic CMC US-DST clock replacement; frozen strategy windows preserved. Native clock/runtime regression checks pass; new MetaEditor compilation and fresh-server validation pending.
Forward testing
- Owner-reported observer startup, not validated demo fills.
ADB · Asian Directional Breakout →
GBPJPY · UTC · aligned 20/40 / opposed control 20/40
Strategy development
- Research and definitionComplete
- Backtest and risk evidenceIn progress
- EA implementationIn progress
- Research / EA parityPending
- Forward testingPending
- Comparison and reconciliationPending
- Reviewed releasePending
- Versioned iterationPending
Next: Compile and attach the automatic CMC-clock replacement from config/mt4/cmc-clock-release-v1.json; confirm the computed offset on fresh quotes, retain old logs and collect matched observations. Actual orders remain disabled.
Research and entry/exit definitions are approved for replay and EA specification. Results acceptance and EA validation remain pending.
Evidence and open decisions
Partial progress records work already done. Completion requires the preceding gates and recorded acceptance; observer startup does not establish demo execution.
- Validate quote-side activation/fills, broker costs, sizing, restart and portfolio safety against the frozen definition.
- Inspected historical results are not untouched OOS. Independent forward evidence and EA parity remain outstanding.
Research and definition
- Explicit owner research-specification freeze recorded on 2026-10-30; option A timing confirmed on 2026-10-31. Executable contract and EA approval remain separate.
- Owner explicitly approved the clarified definition for next backtest and EA specification; no EA/demo/live validation inferred.
Backtest and risk evidence
- Exploratory SQL summary, no pinned engine or complete ledger.
- Local shared preliminary replay with full ledgers and cost sensitivity; hosted research, real execution and untouched OOS acceptance remain pending.
- Owner-approved-definition local M1 price-proxy replay, repeated byte-identically. ALBR unchanged comparator; costs hypothetical, broker fills/EA/OOS acceptance incomplete.
- Forensic verification: exact parent reproduction, all-session independent checks, native/IANA clocks, bounded-gap and nominal entry-stop review; separate estimates/conditional setups. Broker execution and independent forward/OOS acceptance remain pending.
- Owner-requested EURJPY/USDJPY diagnostic pair application, not new approved instruments or EA changes. Identical ADB rules/cost scenario; per-pair metrics, coverage and uncertainty retained. No promotion or portfolio aggregation.
EA implementation
- Order-free observer source candidate implements frozen definition. Native API-simulator scenarios pass; MetaEditor compilation, runtime trace and full parity remain unverified.
- Exact source/contract/package hashes; actual orders and all execution-validation flags remain false.
- Versioned automatic CMC US-DST clock replacement; frozen strategy windows preserved. Native clock/runtime regression checks pass; new MetaEditor compilation and fresh-server validation pending.
Monitoring criteria
Operating specification. Automated monitoring and alerts are not yet implemented. Statistical acceptance bands and sample checkpoints must be agreed before forward-performance decisions.
| Area / cadence | Check | Response |
|---|---|---|
| Baseline and experiments Each repository validation and before candidate review | Pinned V1 files match baseline hashes. Each experiment has a parent baseline, separate version, hypothesis, changed fields, predeclared evaluation period/cost/risk criteria and immutable evidence. | Repository validation blocks baseline drift, unregistered/unsupported experiments and experiment records that claim to replace V1 or authorise live trading. |
| Market data After the target 04:00 Europe/Madrid refresh and before replay | No duplicate or invalid OHLC records; required strategy windows complete; provider and database watermarks recorded against market calendar. | Quarantine bad records; block affected-session replay and report stale or missing input. |
| Economic context At source publication and before any as-of join | Verified release timestamp, vintage and observation identity; no revised or later-known data in historical decisions. | Mark unavailable context explicitly; prevent unverified strategy attribution. |
| Research and EA parity Each code/contract release and every reconciled session | Zero unexplained mechanical differences in signals, activation, expiry, exits and position sizing. | Block promotion and retain discrepancy evidence; investigate without modifying frozen rules. |
| Forward operations During observation and at completed-session reconciliation | Known source/build identity, broker UTC offset, heartbeat and terminal availability; complete signal and no-trade logs. | Mark intervals unobserved when evidence is absent; never infer no-trade from missing logs. |
| Execution and portfolio risk Before order submission, on fill, restart and daily rollover | 1% planned equity risk; at most two actual portfolio entries per London day; verified conversion and atomic fail-closed coordination. | Fail closed for order-capable EAs; observers remain order-free and log unrestricted virtual observations separately. |
| Execution costs and outcomes Each completed trade and session | Match signal IDs/tickets; preserve spread, slippage, costs, fills, exits and rejected orders against declared assumptions. | Classify data/feed, clock, rule, broker, risk, operations or unexplained differences; retain all raw records. |
| Performance and release review At a predeclared sample/date checkpoint | Compare like-for-like windows, R, net costs, drawdown and uncertainty. Statistical limits and minimum samples must be specified before observation; currently unset. | No automatic profitability decision or live promotion; owner reviews evidence and any separately versioned iteration. |