Janus
- Experts
-
Alex Amuyunzu Raymond
I’m an experienced MQL5 developer and software engineer specializing in building robust, high-performance algorithmic trading systems for MetaTrader 5. My work goes beyond standard EA development — I design intelligent, scalable solutions that combine market logic, automation, and external - Versão: 1.4
- Ativações: 5
JANUS — The God of Openings Trades the Only Window That Matters.
The first fifteen minutes of the session are the only part of the day where the market tells you the truth. JANUS listens.
The problem with every opening-range EA on the market.
Retail opening-range breakout bots are, on average, break-even after spread and slippage. Independent replication studies — single-author and not peer-reviewed, but reproducible — have shown this repeatedly. If you buy a mechanical ORB EA, you are buying a coin flip with a commission attached. The reason is not that opening ranges do not matter. It is that most opens are noise. A machine that trades every open is a machine that trades noise. The edge, if it exists at all, is conditional.
The hypothesis JANUS is built on.
JANUS does not trade every open. It trades the open only when three independent conditions align:
-
Abnormal opening participation. The tick volume of the opening window versus the median of the same window over the last twenty sessions. If today's open is not busier than the recent norm, the day is skipped.
-
Gap context. The overnight gap measured against daily ATR. A gap that is small relative to ATR is noise; a gap that is large and aligned with the first candle's behaviour is information.
-
First-candle behaviour. The colour of the opening candle and where it closed relative to the opening range midpoint. If the first candle and the gap disagree, JANUS does not take sides. It waits.
Only when these three agree does JANUS arm a stop order beyond the opening range. If they do not, the session is skipped and the reason is written to the decision journal, so you can see exactly why nothing happened.
What JANUS is not.
-
It is not a martingale.
-
It is not a grid.
-
It is not an averaging-down system.
-
It is not a recovery-sized system.
-
It is not a signal service, a copy-trading product or an advisory tool.
-
It does not promise profit. No software can. What it promises is that every order is sent with a hard stop loss, sized by a risk engine you can inspect, and that every decision — trade or skip — is logged in a CSV you can audit.
The risk engine — the actual reason JANUS is worth $899.
Every rival EA on the MQL5 Market that has blown up an account has done so for one of two reasons: oversized lots, or no stop loss. JANUS is designed so that neither failure mode is possible.
-
Fixed-fraction sizing from account equity, independent of leverage. If your broker offers 500:1, JANUS does not care. The risk per trade is the number you set — default 0.5% of equity — and it is enforced before the order is sent.
-
Real tick-value sizing. JANUS never assumes a contract size. It reads SYMBOL_TRADE_TICK_VALUE and SYMBOL_TRADE_TICK_SIZE from your broker and computes the exact money at the stop loss. This means it works on CFD index contracts, on FX majors, on gold, and on stake-per-point spread-bet accounts where a "lot" is one pound per point.
-
Self-test on every symbol. Before trading, JANUS runs a self-test that computes the money loss at a hypothetical stop, using your broker's real tick values. If the self-test fails — unusual symbol, broken broker data — JANUS refuses to trade. It would rather do nothing than do something wrong.
-
Post-fill verification. After every fill, JANUS re-reads the position, recomputes the actual risk at the actual stop, and if the real risk exceeds the intended risk by more than 5%, it closes the position and logs the breach. This is the safety net against a broker who fills at a worse price than expected.
-
Equity drawdown halts. A daily drawdown limit and a total drawdown limit, both measured on equity, both configurable. If the daily limit is hit, JANUS stops opening new trades for the rest of the day. If the total limit is hit, it stops until you reset it.
-
Correlation cap. GER40, UK100, US100 and JP225 are highly correlated. JANUS will not hold two of them at once by default. This single feature removes the most common cause of "the EA had one good strategy but blew up because it opened four correlated positions."
-
Idempotency. One trade per session per symbol, enforced by a session key that survives terminal restarts. A crashed terminal cannot cause a duplicate order. A second instance on the same account cannot double up.
Four market profiles, selectable, with defaults tuned per market.
-
Europe / Spain — GER40, EURUSD, EURGBP, plus neighbours Portugal, France, Italy. Frankfurt/Xetra open at 09:00 CET. ESMA leverage awareness built into defaults. Sizing independent of leverage by design.
-
Germany — GER40 focus, plus DAX overnight drift module. Aware of the June 2026 turbo/CFD rule changes; defaults respect a narrower CFD segment.
-
Japan — JP225 and USDJPY at the Tokyo open, with the no-DST rule set correctly in code. Domestic 25:1 leverage is irrelevant because sizing never uses leverage. No broker is named or promoted anywhere in the Japan profile.
-
United Kingdom — UK100 and GBPUSD at the London open (08:00 GMT/BST), correctly handling the three-week window each year when UK and US DST are misaligned. Handles the spread-bet contract form where volume is a GBP stake per point.
Seven strategy modules, switchable, all OFF by default.
JANUS ships with seven hypotheses, not one product. M1 is the main candidate — conditional opening drive. M2 is the DAX overnight drift. M3 is the USDJPY gotobi day effect (folklore; off by default and probably false). M4 is the failed-breakout fade. M5 is the London open drive for UK100 and GBPUSD. M6 is the European-session fade of the Asian range on EURUSD and EURGBP. M7 is a higher-timeframe trend pullback on XAUUSD and US100.
You enable exactly one per symbol. Every module has a comment block in the source stating its rationale and its known failure conditions.
The kill gate — nothing ships until it passes.
This is not marketing. This is the actual gate a module must clear before it is enabled in a shipped configuration:
-
Net expectancy at 1× costs must be at least +0.10 R per trade above a seeded placebo, with the bootstrap 95% confidence interval lower bound above zero.
-
Still positive at 2× costs.
-
Net profit factor ≥ 1.25 in validation and in holdout.
-
At least 300 total trades, at least 100 in holdout.
-
Positive in at least 70% of calendar years, no single year more than 40% of total profit, positive excluding 2020–2022.
-
At least 70% of neighbouring parameter sets within 70% of the best metric.
-
Monte Carlo 95th-percentile max drawdown within the prop-mode limit at the intended risk.
-
Deflated Sharpe probability above 0.95 for the number of trials.
-
A written economic rationale not contradicted by the data.
Modules that fail are removed per market. A market profile with no passing module is reported as not shippable. This is why JANUS is sold as a tested research harness with a production candidate inside it — not as a finished money machine.
The placebo, built in.
Switch InpPlaceboMode to random and JANUS fires the same entries at the same times with random direction. You run the real version and the placebo version on the same data, subtract, and you have an honest estimate of edge. No other retail EA I know of ships this. It exists because we would rather you discover that a module has no edge than sell you one that does not.
Test instrumentation built into the product.
-
CSV trade log with date, symbol, side, volume, entry, exit, P&L in account currency, R-multiple, module, profile, and the stress-cost values used. Tax-friendly columns.
-
Decision journal: every skip and every entry with the reason.
-
Module summary CSV at test end.
-
Custom OnTester criterion: net expectancy in R with a trade-count penalty, so the Strategy Tester optimizer rewards robust modules, not curve-fit ones.
Localisation EN / ES / DE / JA.
All dashboard strings and reason codes have professional translations in English, Spanish, German and Japanese. They have been written by an engineer, not by a native speaker. If you buy the product to sell in one of those markets, flag the language for review before commercial release; we say this openly rather than shipping a translation we cannot vouch for.
Support, licensing, honesty.
-
Offline, account-locked key. It deters casual copying. It is not military-grade, and we say so.
-
Telegram alerts optional, via one allow-listed URL. We tell you exactly which URL to add in Tools → Options → Expert Advisors.
-
12 months of email support included. What support does not cover: individual trade advice, portfolio allocation, parameter tuning to your account, or profit expectations. Those are not this product.
-
Refund terms: 14 days, subject to the licence not being activated on more than two accounts. Consumer law in Spain, Germany, the UK and Japan overrides this draft.
-
Regulatory notes for each market are provided for your information and are not legal advice. Wording must be reviewed by a lawyer in Spain, Germany, Japan and the UK before publishing.
What JANUS will not do — the honest limits page.
JANUS will lose money in regime changes. It will lose money when a broker's tick data is unreliable. It will lose money on gap-through-stop days where the stop is filled far worse than placed. It will not open trades if the self-test detects an unusual symbol, and that is a feature. It cannot protect you from a decision you make outside the EA — a manual override, an account with 500:1 leverage and a slashed deposit, a position opened by a phone. It is a discipline tool, not a guarantee.
Who JANUS is for.
Traders who have already been burned by a martingale EA and want to see the risk engine before they trust the strategy. Traders who can read a CSV and want to know why a trade was skipped. Traders who understand that "no edge claimed until verified" is the only honest claim a vendor can make. Traders who want a research harness that ships with a production candidate, not a black box with an equity curve screenshot.
Who JANUS is not for.
Anyone looking for a promise. Anyone looking for a signal. Anyone who will disable the master switch and run it on a live account without reading the manual, running the tester, and doing the forward-test protocol. Anyone who wants 10% a month. That product does not exist, and if someone sells it to you, it is a lie.
The one sentence.
JANUS trades the only part of the day that matters — the opening window — but only when the market is telling the truth, and it enforces that discipline with a risk engine you can inspect line by line.
