Ten Systems.
One Engineering Standard.
Ten algorithmic systems engineered on 100% real tick data — 17,000+ backtested trades, in-sample equity curves published in full, out-of-sample validation on every system — including the two windows that finished negative. Presale open while live validation runs on Darwinex Zero.
The suite at a glance.
Ten systems. Ten configurations, eight distinct edges. Click any card to jump to the full results — in-sample equity curves, out-of-sample validation and all the data behind each system.
In-sample figures: simulated · real tick data · €100k base per system. OOS windows validated on untouched data.
Two systems in live validation.
Horizon Striker and AEGIS have been running on Darwinex Zero with real capital since June 2025. The live track record will be published when the sample is statistically significant — not before, and never cherry-picked.
Precision-engineered for speed. Horizon Striker operates in three high-probability windows per day — London open, New York open, and the late-US reversal zone — when institutional order flow peaks. The core signal is the Fair Value Gap (FVG): a price imbalance left by rapid institutional moves, filtered through higher-timeframe bias, volatility regime, news events, and dynamic trailing exits.
+194% net return over 5 years of in-sample testing on 1,840+ trades, with a profit factor of 1.43 at just 0.75% risk per trade. Out-of-sample validation confirms edge persistence with +22% return on untouched data.
Running on real capital since June 2025. The live track record will be published in full when the sample is statistically significant — we refuse to market a handful of months, good or bad.
Named after the Greek shield — built for protection and solidity. AEGIS identifies order blocks: price zones where institutions placed significant positions. It applies a 5-layer multi-timeframe confirmation system: weekly/daily bias, liquidity clusters, volatility regime, market structure rules, and portfolio-level exposure controls. If any single filter fails, the trade is rejected.
The result: a profit factor of 2.19 with 61.8% win rate across 620+ in-sample trades, and an out-of-sample profit factor of 2.41 — the edge actually strengthened on untouched data. +340% net return at 1.5% risk per trade.
Running on real capital since June 2025. The live track record will be published in full when the sample is statistically significant — we refuse to market a handful of months, good or bad.
Eight new systems. Every result published.
Eight uncorrelated edges, engineered on 100% real tick data and validated on out-of-sample windows the optimiser never touched. Six windows finished positive, two finished negative. All eight are published below, unedited.
RebalX is a statistical deviation engine. It continuously measures how far price has stretched from its volatility-adjusted mean across six markets, and acts only when the deviation is extreme and the boundary conditions — trend regime, session context, momentum exhaustion — make a reversion the probable outcome. Most deviations are ignored. The few that qualify are traded with strict, pre-defined risk.
The in-sample record: +279% over 5.8 years with a 69.5% win rate, a 7.51% maximum drawdown and a Sharpe ratio of 6.83 across 1,755 trades. The out-of-sample window is negative — −6.3% across four strongly directional months — and it is published below exactly as it happened.
A four-month directional squeeze is the known weak regime for mean reversion. 154 trades do not invalidate a 1,755-trade edge — but hiding them would invalidate us. This window is shown unedited.
VWAP Prime trades the benchmark institutions actually execute against: the volume weighted average price. It is built on the methodology of Zarattini & Aziz (2023), “Volume Weighted Average Price: The Holy Grail for Day Trading Systems” (SSRN 4631351), re-engineered with TEL execution know-how — session filters, volatility gates and disciplined intraday exits.
The sample size is the point: 3,286 in-sample trades producing +183% with an 11.11% maximum drawdown. Out-of-sample, the profit factor came in above the in-sample figure — 1.19 versus 1.16 — on 142 trades the optimiser never saw. A modest edge, unusually well-behaved on unseen data.
An OOS profit factor above the in-sample figure is the single most encouraging robustness signal an intraday system can produce. It is still only 4.5 months of data — we treat it accordingly.
AMD Phantom reads the market in three phases: accumulation, manipulation, distribution. Where the crowd chases the breakout, Phantom waits for the manipulation leg — the engineered push that clears resting stops before the real move — and positions for the reversal. The manipulation phase is invisible to most participants. This system is built to see it.
In-sample: a profit factor of 1.67 — the highest of the next-gen suite at the time of writing — across 980 trades on seven markets, with +97% net return and an 8.06% maximum drawdown. Out-of-sample it held up cleanly: +15.0% over ten months, profit factor 1.43, and a drawdown of 6.90% — lower than in-sample.
Ten months, 209 trades, drawdown below in-sample. The strongest OOS window in the suite is published with the same scrutiny as the weakest — that is the standard.
ORB Seven is seven opening range breakout engines in a single EA — Asia, London, New York and four further session opens, including the GDAXI cash open — each with its own calibrated range logic and proprietary filters for volatility, spread and failed-break conditions. The seven engines are the name. They are also what separates this from a generic ORB script.
The 41% win rate is not a weakness; it is the shape of breakout asymmetry — average winners far larger than average losers, compounding to +282% over five years with the lowest in-sample drawdown of the suite (5.99%), a recovery factor of 12.77, and 4,197 trades. The OOS window drew down mid-period and recovered strongly into the close. We show the whole path.
A mid-window drawdown of 14.36% followed by a strong recovery is exactly what a 41%-win-rate breakout profile looks like from the inside. Shown as it happened — endpoint and path both.
Specter is the specialised sibling of AMD Phantom. It hunts liquidity sweeps: the moments when price raids beyond a prior range’s extreme, triggers the stops resting there, and fails. Breakout traders are trapped on the wrong side of the move; Specter fades the failure with tightly defined invalidation and selective, low-frequency entries.
In-sample: a Sharpe ratio of 7.37 — among the highest in the suite — with a 1.38 profit factor across 495 trades on five markets. The OOS window closed at −2.1% on just 34 trades: a sample too small to prove anything in either direction. It is published anyway.
34 trades is noise, not evidence — in either direction. We publish it because selective disclosure is how this industry lies, and because the portfolio-level OOS across all eight next-gen systems is +€92,003.
IDM Momentum trades intraday directional flow on the S&P 500, built on the methodology of Zarattini, Aziz & Barbon (2024), “Beat the Market: An Effective Intraday Momentum Strategy for S&P500 ETF (SPY)” (SSRN 4824172), re-engineered with TEL execution know-how. When price breaks beyond the normal move for that time of day — a time-of-day volatility band computed on a rolling 14-day window — the system reads a genuine demand/supply imbalance and follows it: long above the band, short below. Exits trail on the band and session VWAP; every position is closed at the cash close (16:00 ET). No overnight risk.
In-sample: +143% net, PF 1.38, win rate 61.2%, Sharpe 5.15 across 846 trades, max drawdown 13.2%. Out-of-sample, the edge held over a full 16-month window (Jan 2025–May 2026): +19.8% at PF 1.19 on 200 trades, DD 8.2%. The longest OOS window in the suite, on data the optimiser never touched. One of two systems on this page built on peer-reviewed research.
The academic edge described by Zarattini, Aziz & Barbon (2024) — SSRN 4824172 — held on 16 months of live data the model never saw. That is the test that matters.
DWS Reversal reads the higher-timeframe daily bias first — who controls the market — then waits for the New York open to produce a fake-out against that bias: a move that traps breakout traders on the wrong side. It confirms the trap with a fair value gap, enters on the pullback into the gap with strictly defined risk, and targets the prior day’s liquidity. A counter-trend entry in the service of the dominant trend, across seven index markets.
Across seven markets, DWS produced +100% net in-sample at PF 1.45 — with a maximum drawdown of just 6.3% and a Sharpe ratio of 7.45, the highest in the suite — on 609 trades. Win rate is 45.2%: by design. The system targets the prior day’s liquidity and cuts losers fast, so average winners outrun average losers. Profit factor 1.45, not win rate, is the number that matters. Out-of-sample: +13.0% at PF 1.21 on 152 trades, DD 8.8%.
By design. The system targets the prior day’s liquidity and cuts losers fast, so average wins outrun average losses. Profit factor 1.45 — not win rate — is the number that matters here.
Most breakout systems enter on the break. Sentinel waits for what comes after: price must break the opening range, then come back and hold the level on the retest before any order is placed. That single extra requirement discards the majority of false breaks that bleed generic ORB scripts dry. On top of it sits a market regime filter that decides whether the day is worth trading at all — expansion regimes are traded, compressive and choppy regimes are sat out. Four markets: Nasdaq, S&P 500, GDAXI and USDJPY.
In-sample: +103% net at profit factor 1.34 over 1,000 trades, with a maximum drawdown of 7.36% and a Sharpe ratio of 4.90. The win rate is 37% by design — average winners run €1,102 against €484 average losers, a 2.3:1 ratio that is the whole point of a breakout system. Out-of-sample it produced the strongest result in the suite: +38.5% over sixteen months at profit factor 1.46 — higher than in-sample — on 288 trades, drawdown 8.31%. It is worth being precise about how that window started: the first two months were negative, with equity down roughly 7% into March 2025 before the run resumed.
Sentinel and ORB Seven both live in the breakout family, so they are not fully uncorrelated — we would rather say that than sell you eight edges that are really seven. What separates them: ORB Seven runs seven session opens on a straight break; Sentinel demands the retest and skips entire days on regime.
Ten systems. One standard.
Different edges, different markets, one pipeline. Every row below passed the same walk-forward, out-of-sample and Monte Carlo gauntlet — and every OOS figure is shown, positive or negative.
| System | Style | Core signal | Markets | IS Net | IS PF | IS Max DD | IS Trades | OOS |
|---|---|---|---|---|---|---|---|---|
| Horizon Striker | Intraday scalping | Fair Value Gap | Indices + FX | +194% | 1.43 | 9.62% | 1,840+ | +22% |
| AEGIS | Swing trading | Order Block | Indices + FX + Gold | +340% | 2.19 | 11.4% | 620+ | +41% |
| RebalX | Mean reversion | Statistical deviation | 6 markets | +279% | 1.40 | 7.51% | 1,755 | −6.3% |
| VWAP Prime | Intraday momentum | Institutional VWAP | 1 index | +183% | 1.16 | 11.11% | 3,286 | +8.2% |
| AMD Phantom | Contrarian reversal | Accumulation–manipulation–distribution | 7 markets | +97% | 1.67 | 8.06% | 980 | +15.0% |
| ORB Seven | Session breakout | Opening range × 7 sessions | 5 markets | +282% | 1.25 | 5.99% | 4,197 | +5.9% |
| Specter | Liquidity sweep reversal | Range raid failure | 5 markets | +85% | 1.38 | 11.99% | 495 | −2.1% |
| IDM Momentum | Intraday momentum | Time-of-day volatility band | 1 (S&P 500) | +143% | 1.38 | 13.2% | 846 | +19.8% |
| DWS Reversal | Sweep reversal | Daily bias + fair value gap | 7 markets | +100% | 1.45 | 6.3% | 609 | +13.0% |
| Sentinel | Break & retest breakout | Opening range + retest + regime filter | 4 markets | +103% | 1.34 | 7.36% | 1,000 | +38.5% |
IS Net: simulated · in-sample · €100k base per system. OOS windows were never touched during development or optimisation. Negative OOS results are shown unedited.
Single windows are noise.
The portfolio is the signal.
Eight distinct edges — mean reversion, institutional flow, contrarian reversal, session breakout, liquidity sweep, intraday momentum, daily-bias reversal, break-and-retest breakout — built to not move together. Judged one by one, short OOS windows wobble. Combined, they point one way.
Statistical mean reversion — earns when stretched prices snap back.
Institutional flow — trades around the benchmark institutions execute against.
Contrarian reversal — fades the engineered stop-run before the real move.
Session breakout — rides expansion out of seven global opening ranges.
Liquidity sweep — trades the failure of raids beyond prior ranges.
Intraday momentum — trades the time-of-day imbalance on the S&P 500, flat by the close.
Sweep reversal — fades the session-open trap back toward the prior day’s liquidity.
Break & retest — trades the opening range only after the level holds, and only in the right regime.
154 trades · PF 0.90. Mean reversion against a strongly directional window. The known weak regime, on full display.
34 trades · PF 0.88. A sample too small to mean anything — published all the same.
“Anyone can show you a wall of green curves. We’d rather show you the truth: single OOS windows are noise, the portfolio is the signal.”
Combined OOS across all eight next-gen systems: +€92,003 on a €100k-per-system base · 1,398 trades · every window published.
Every system on this page was engineered, tested and deployed on his own capital before reaching yours. Not a trader turned marketer, not a team selling signals — a single engineer with three degrees in engineering disciplines and more than five years of full-time research dedicated to one question: can systematic edge be built to last on real data, validated honestly, and deployed with genuine skin in the game.
The answer is these ten systems. The in-sample curves are published in full. The out-of-sample windows — including the two negative ones — are published unedited. The live validation is running on Darwinex Zero with real capital. The methodology is the authority.
How every system earns its place.
No shortcuts. Every system must survive a multi-stage gauntlet before touching real capital — in full, without exception.
Walk-Forward Analysis
The strategy is optimised on rolling in-sample windows, then tested on the next unseen period. If the edge degrades across windows, the system is rejected. No single lucky fit survives this process.
Out-of-Sample Validation
20% of all data is quarantined from day one. Never touched during development, optimisation, or parameter selection. The OOS test is the final exam — pass or fail, no retakes.
Monte Carlo Simulation
100+ reshuffled equity paths quantify worst-case drawdown under randomised trade order, slippage variation, and win-rate degradation. If the 95th-percentile drawdown exceeds tolerance, the system is shelved.
No system skips a stage. No exceptions.
Parameter Sensitivity
Key parameters are shifted ±20% to verify the edge is not dependent on a single fragile setting. Robust strategies produce stable results across a parameter neighbourhood, not a single peak.
Live Deployment
After passing all tests, the system goes live on small risk with every trade journalled. Performance is monitored against backtest benchmarks. Divergence beyond 2 standard deviations triggers review.
Choose your risk profile.
Two numbers matter most: expected 5-year return and maximum observed drawdown. Pick the balance that lets you sleep at night. Presets below cover the two flagships; calibrated presets for the eight next-generation systems ship with delivery.
Next-generation presets (RebalX, VWAP Prime, AMD Phantom, ORB Seven, Specter, IDM Momentum, DWS Reversal, Sentinel) are delivered with the EAs, calibrated on the same €100k baseline shown in the tables above. The eight next-generation systems cover a broad spread of strategies and timeframes.
Key rules to follow.
These apply to all setups. Ignoring them is the most common source of issues reported.
One EA per chart
Never run the same EA on multiple charts — order conflicts and unpredictable sizing result.
Verified brokers only
Use Darwinex. Unverified brokers may have wider spreads or different symbol names that degrade performance.
Always use a VPS
Run MetaTrader on a VPS for 24/7 uptime. Home PCs that sleep interrupt open positions.
Journal every trade
Log all results to track performance and identify issues early — critical during the first month.
Keep WebRequest active
The EA requires WebRequest to tradingengineeringlab.com for license validation.
Market Watch first
Add all required symbols to Market Watch before attaching the EA. Missing symbols cause silent errors.
Where these bots run.
Performance figures validated on Darwinex. Symbol names, spreads, and execution conditions confirmed to match backtest expectations.
Before you run these systems.
Read these before reaching out.
One license. Everything included.
No tiers. No upsells. No recurring fees. One license, up to 3 MT5 accounts, all ten Expert Advisors — two flagships and eight next-generation systems.
Stop trading on
hope and hunches.
Ten algorithmic systems, one engineering standard. Validated out-of-sample with every result published — including the negatives. If the methodology matches your approach, reach out: we will walk you through setup, licensing and what to realistically expect.
Personal reply from the engineer — typically within 24–48h.
1 license · up to 3 MT5 accounts · presale ends at live track record publication
Risk Warning: Past results are not a guarantee of future results. Trading financial markets involves substantial risk of loss. The EAs described here are not financial advice. Backtest and out-of-sample results are simulations performed on real tick data and do not represent live trading. Live validation on Darwinex Zero is in progress; the live track record will be published when the sample is statistically significant. Only trade with capital you can afford to lose entirely. Results shown include modelled commissions, spreads and slippage unless explicitly stated otherwise.