Reviewed bots.
Clear deployment decisions.

Only 26 annual / 3 recent positions; do not infer a strong repeatable edge or increase risk.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

This BAT entry is optional and only installs when the broker exposes a compatible symbol.

This setting was selected after comparing exit and session variants. It is historical evidence, not a guarantee, and the combined portfolio still needs shared-margin forward observation.

Separate signal/session components and overlapping DMC risk. Optimise only well-supported hypotheses, with untouched validation and comparison to the frozen original.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The untouched year returned +12.08% with PF 1.18, 40.52% wins, 9.96% drawdown and 116 trades. Run the three DMC entries on demo first because their signals can overlap and create correlated portfolio exposure.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Collect independent trades and review correlated US100 exposure. Keep entry rules frozen during observation; small samples are not a reason for aggressive optimisation.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The untouched year returned +3.68% with PF 1.88, 66.67% wins, 2.99% drawdown and 12 trades. Run the three DMC entries on demo first because their signals can overlap and create correlated portfolio exposure.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Refresh exact current dates, confirm zone freshness/entry logic, and collect forward trades. Do not optimise to four observations.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The untouched year returned +6.88% with PF 1.90, 46.67% wins, 3.11% drawdown and 15 trades. Run the three DMC entries on demo first because their signals can overlap and create correlated portfolio exposure.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Conditional keep, not a recent winner. Obtain a current native quarter and demo results before increasing risk; keep the added ADX/DI filter OFF as selected.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The untouched year returned 19.57% with PF 2.77, 70.73% wins and 3.45% drawdown. Monte Carlo return P5 was +15.89%, but historical robustness does not guarantee future performance.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Final user selection: Added ADX/DI filter removed; exact native-tested baseline restored. Existing unrelated filters, risk and exits retained. Retrospective evidence, not independent validation.

Audit point-in-time forecasts, service outages, order fallback, doubled exposure and exact current binary/settings. Do not backtest with forecasts generated using future information.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

First check small targets versus costs and average loss, not extra win-rate filters. Preserve the existing version and test a minimum net payoff / exit hypothesis on untouched windows.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

High win rate, variable sub-1R targets; weak five-year expectancy.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Only a narrow recent PF margin. Five-year equity DD was 24.07% at 1% standalone risk; the annual cache and 3M fresh runs use separate account starts.

DI OFF is a custom unvalidated selection; published DI-ON results do not describe it.

Latest DI-on wide-stop ATR-trail BASE; no experimental ML gate

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The 2.5R exit was selected on the first two years and confirmed on the untouched final year. Historical performance is not a guarantee.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Conditional keep. Cached evidence ends September 5, so a current quarter and live costs are required before promotion.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The conservative 0.75R alternative achieved PF 3.36 and 0.88% drawdown, but only 26 locked-year trades and +4.14% return. The active version is retained because it produced +8.67% across 72 trades; neither result guarantees future performance.

The embedded Markov gate was natively validated and rejected for deployment because it materially reduced return and sample size. Full Safe therefore preserves the stronger Standard inputs for this EA.

Replay current settings at realistic latency, spread and fill assumptions. Both sides can trigger; the per-order risk is not a portfolio cap.

Current Oct 3 placement release requires matching replay; old event-optimised caches intentionally excluded

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Validate current release with realistic costs and historical-vintage event inputs before reconsidering live trading.

Current Oct 3 placement release requires matching replay; old event-optimised caches intentionally excluded

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Stress delayed fills, spread spikes, both orders triggering and market fallback; use current hashes. Silver tight-stop risk is particularly sensitive to costs.

Current Oct 3 placement release requires matching replay; old event-optimised caches intentionally excluded

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Replay current source/SET/helper with point-in-time event inputs, realistic spreads/delays/slippage and two-sided exposure. Old large returns do not validate the current profile.

Current Oct 3 placement release requires matching replay; old event-optimised caches intentionally excluded

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Gross website winners can turn into net losers after costs. Overlapping XAU ORB signals need one combined exposure budget.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

This setting was selected after comparing exit and session variants. It is historical evidence, not a guarantee, and the combined portfolio still needs shared-margin forward observation.

23 annual / 5 recent trades; W/L 3/3 annually. Correlated with other Gold ORBs; no evidence to stack all of them at full risk.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The 52.17% locked-year win rate is historical and based on only 23 trades. This EA is strongly correlated with the other two XAU ORBs, so their planned risks add when signals overlap.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Recent-quarter standalone target-matched evidence not republished here; use matching portfolio slices. Observe forward execution; approval is not statistical certification.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The 10,000-path Monte Carlo return P5 was +0.12% and P95 drawdown was 11.25%. The placed target is 6R, but timed exits reduced the locked-year realized average win-to-loss ratio to about 1.68.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Retained results are the original fixed-one-lot benchmark, not the history-sized reviewed profile. No stop-loss: a future trade can lose more than the selected risk scenario. Retained results are the original fixed-one-lot benchmark, not the history-sized reviewed profile. No stop-loss: a future trade can lose more than the selected risk scenario. Original 1-CFD-lot benchmark differs from BAT sizing. Validate hours prospectively with realistic spread and an actual loss limit.

Research only: the hours were selected on the latest year, so neither recent window is an untouched holdout. These profiles failed the long-history validation gate. Zero-spread quotes occur frequently in the broker history; recorded-cost returns can overstate live performance. Fixed one CFD lot, no protective stop, and delayed/weekend exits are possible. Included in normal BATs by owner request with historical-loss sizing; the displayed results are the original one-lot benchmark, not the new sized portfolio. FTMO/Ava installers and licensed checkout remain excluded.

One CFD lot is broker-specific, not one NQ/ES futures contract. A hedging account is required.

Recent windows overlap the selection period. No independent-broker or prospective validation is available.

Only four recent positions; existing Monte Carlo lower-tail return was negative. Recent profitability is not a pass-rate forecast.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

Only 34 locked-year trades occurred. The 10,000-path block bootstrap had a 78.38% profit probability, -5.24% return P5 and 9.20% P95 drawdown; this is historical evidence with a meaningful downside tail, not a forecast.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Recent-quarter standalone target-matched evidence not republished here; use matching portfolio slices. Observe forward execution; approval is not statistical certification.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

Only 27 locked-year trades were observed. The 10,000-path Monte Carlo return P5 was -0.94%, so this configuration needs forward observation and is not a guarantee.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Recent-quarter standalone target-matched evidence not republished here; use matching portfolio slices. Observe forward execution; approval is not statistical certification.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

Only five trades occurred in the displayed one-year period. The broader 2020-2026 report returned 18.70% with PF 2.16 over fifty-three trades, but that longer result is context rather than a substitute for fresh forward evidence.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Retained results are the original fixed-one-lot benchmark, not the history-sized reviewed profile. No stop-loss: a future trade can lose more than the selected risk scenario. Retained results are the original fixed-one-lot benchmark, not the history-sized reviewed profile. No stop-loss: a future trade can lose more than the selected risk scenario. Displayed returns are the original 1-CFD-lot benchmark, NOT the new historical-loss-sized deployment. Historical maximum loss is not a future risk cap. Stress real spread and protective stops before live use.

Research only: the hours were selected on the latest year, so neither recent window is an untouched holdout. These profiles failed the long-history validation gate. Zero-spread quotes occur frequently in the broker history; recorded-cost returns can overstate live performance. Fixed one CFD lot, no protective stop, and delayed/weekend exits are possible. Included in normal BATs by owner request with historical-loss sizing; the displayed results are the original one-lot benchmark, not the new sized portfolio. FTMO/Ava installers and licensed checkout remain excluded.

One CFD lot is broker-specific, not one NQ/ES futures contract. A hedging account is required.

Recent windows overlap the selection period. No independent-broker or prospective validation is available.

Do not remove a profitable asymmetric payoff just to manufacture more wins. The filter selection was retrospective, not an independent holdout.

DI OFF is a custom unvalidated selection; published DI-ON results do not describe it.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The preceding independent validation year returned +3.84% with PF 1.09 and 11.94% drawdown. Latest-year Monte Carlo return P5 was -7.24%, so this remains a watch allocation.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Final user selection: ADX(14) >=20 + DI direction agreement on completed M15. Only completed bars are used; this gate blocks new entries, never protective exits. Retrospective evidence, not independent validation.

Recent-quarter standalone target-matched evidence not republished here; use matching portfolio slices. Observe forward execution; approval is not statistical certification.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

Only 24 locked-year trades were observed. The 10,000-path Monte Carlo return P5 was -1.17%, so this configuration needs forward observation and is not a guarantee.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Recent-quarter standalone target-matched evidence not republished here; use matching portfolio slices. Observe forward execution; approval is not statistical certification.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

Only 11 locked-year trades were observed. The 10,000-path Monte Carlo return P5 was +0.04%, so this configuration needs forward observation and is not a guarantee.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Do not raise allocation from this retrospective screen alone. Prior research also had a negative bootstrap lower tail.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The locked result contains only 44 trades and its Monte Carlo return P5 was -1.71%. Treat it as a demo forward-test candidate, not a proven production edge.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Observe forward execution. User deployment approval is separate from research qualification; no new backtest was run for this package.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Final user selection: Added ADX/DI filter removed; exact native-tested baseline restored. Existing unrelated filters, risk and exits retained. Retrospective evidence, not independent validation.

Not a high-win-rate core. Refresh the missing September/October data before changing exposure.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The untouched year returned 89.60% with PF 1.89, 43.20% wins and 12.79% drawdown. The 10,000-path Monte Carlo return P5 was +89.28%; historical results still do not guarantee future performance.

Separate modules. Momentum did not beat random direction; the qualified turn-of-month PROP module is not the current BEST trio. Changing to it requires its own approval and replay.

BEST trio exact whole positions; live installer variant/clock may differ. Older untouched holdout PF 0.74 / -20.3%, WATCH_ONLY

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The optimised settings lost 20.3% (PF 0.74) on untouched 2019-2021 history and the deflated Sharpe was 63% against a 95% requirement. Longer-window results include the development years the settings were tuned on.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Not a robust high-win-rate allocation. Retain the code and demo logs; require new independent evidence before live use.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

This BAT entry is optional and only installs when the broker exposes a compatible symbol.

Only 26 locked-year trades occurred and the 10,000-path Monte Carlo return P5 was -14.32%. The model is installed at the user's request but is not classified as a robust core EA.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Audit EM1 versus EM4, direction, session, stop/target and risk rounding separately. The experimental AND flow and range-only flow did not fix the longer-history weakness; do not deploy them automatically.

Fresh Oct 5 native replay of current Safe M15 rules; not the experimental M5 flow

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

This setting was selected after comparing exit and session variants. It is historical evidence, not a guarantee, and the combined portfolio still needs shared-margin forward observation.

Recent four trades alone are too small to prove a breakdown; the longer negative half-year and existing demo-only status support a pause.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The native one-year result confirms the combined implementation, but its most recent six-month slice was negative. Keep this EA on demo until forward evidence shows the recent weakness has ended.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Revisit entry regimes and payoff using frozen walk-forward windows. This finding is for normal 1R, not a test of a different FTMO 0.5R variant.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

User-selected retrospective target; strict cross-window PF screen failed. Not an FTMO pass forecast.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Check range width, breakout side, session and stop/exit attribution. Do not blindly restore DI: the prior long-window review rejected that addition.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

This setting was selected after comparing exit and session variants. It is historical evidence, not a guarantee, and the combined portfolio still needs shared-margin forward observation.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Final user selection: Added ADX/DI filter removed; exact native-tested baseline restored. Existing unrelated filters, risk and exits retained. Retrospective evidence, not independent validation.

Refresh the missing month, review long/short and exit costs separately, then validate any session/regime hypothesis out of sample.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

This BAT entry is optional and only installs when the broker exposes a compatible symbol.

Only 26 trades occurred in this locked year and 44 in the exact three-year audit. Treat this as demo-forward-test evidence, not proof of future performance.

Validate bearish-regime eligibility and exit timing. Avoid adding filters selected solely on the recent bullish Nasdaq sample.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

The untouched year returned +7.07% with PF 1.14 over 73 trades, and the 10,000-path locked-trade Monte Carlo return P5 was -12.01%. Treat this as a demo-forward candidate rather than a proven live-capital edge.

Refresh the missing month, audit signal timing and structural stop/entry costs. Use a frozen longer validation period rather than fitting to four losses.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

Only 24 locked-year trades occurred. The 10,000-path bootstrap return P5 was +8.47% with 6.03% P95 drawdown; this remains historical evidence rather than a guarantee of future performance.

No embedded Markov-gate version has been validated for this EA, so Full Safe installs it with the same locked native inputs as Standard mode.

Audit the Markov gate and signal/event logs before loosening conditions. FTMO uses Standard instead: its separate cached year had 19 trades, +4.89%, PF 1.99 and 52.6% wins; this is not the same mode.

Historical returns are not guaranteed and live execution can differ.

Broker symbol names, spread, slippage and contract size affect results.

Standard had broader trade evidence, but its recent-half PF was 0.93. Keep it as the broad benchmark while collecting demo-forward evidence.