Four engines on duty
Every Samurex AI engine is built for one market and measured on that market alone. Each has its own live account, published by an auditor we do not control. Pick a row to read the dossier.
Every figure is read from the engine's live FXBlue account and describes past months. They are not targets and not forecasts.
Samurex Tandem
Two currencies. Two specialists. One subscription.
Most currency robots run one model over whatever pair you point them at. Tandem does the opposite: EUR/USD gets a model tuned to how EUR/USD actually behaves, GBP/USD gets its own, and neither has to compromise for the other. Sterling is jumpier around the London open and reverts differently in the afternoon; the euro is calmer and more range-bound overnight. One set of parameters cannot serve both honestly.
Inside each pair, three strategies take turns depending on the hour. A breakout model works the London open, when the day's real range is usually set. Two reversion models cover the Asian session and the middle of the European afternoon, when price tends to drift back toward where it started. The result is an engine that is doing something sensible in most hours rather than sitting on its hands waiting for one setup a week.
- Instruments
- EUR/USD and GBP/USD — a dedicated model per pair
- Strategies
- London breakout · Asian reversion · midday reversion
- Active hours
- Asian, London and New York sessions
- Risk control
- Fixed stop on every position, sized to about 1% of the account
- Suits
- A first engine, and anyone who wants FX exposure spread across the clock
- Record
- Published live on FXBlue
Illustrative monthly returns
Twelve months, consistent with the published band. Illustration, not a record.
Past performance does not indicate future results.
Live audited account
Read from the linked FXBlue account. Past performance does not indicate future results.
Samurex Ember
One metal, studied properly.
Gold is unusual. It trends hard when the macro story changes and reverts almost mechanically when it does not, and the switch between those two states rarely announces itself. Ember handles this by refusing to pick one identity. It runs a small portfolio of models — some fading the intraday extremes gold reaches during thin hours, others taking the breakout when the London or New York session finally commits — and a higher-timeframe filter decides which of them are allowed to trade today.
The aim is deliberately unglamorous: a smooth curve made of small, frequent, tightly stopped positions rather than one heroic trade a quarter. Ember will miss the top of a big move. It is built not to be on the wrong side of one.
- Instrument
- Gold (XAU/USD) exclusively — one chart, fully automated
- Strategies
- Intraday reversion models plus session-breakout models, gated by trend
- Active hours
- Asian range, then the London and New York liquidity windows
- Risk control
- Uniform small sizing, daily loss ceiling, monthly drawdown halt
- Suits
- Steady compounding where a deep drawdown would be intolerable
- Record
- Published live on FXBlue
Illustrative monthly returns
Twelve months, consistent with the published band. Illustration, not a record.
Past performance does not indicate future results.
Live audited account
Read from the linked FXBlue account. Past performance does not indicate future results.
Samurex Harbor
Blue chips, handled carefully.
Harbor is the conservative member of the line-up and makes no apology for it. It trades large-capitalisation US stocks and the major index ETFs, and it only takes a position when the distance to its stop is small relative to the move it is aiming for. When that asymmetry is not there, it does nothing, and doing nothing is a legitimate outcome for a whole session.
It also refuses to carry risk it cannot manage. Positions are opened and closed inside regular US market hours, so the account is not exposed to an overnight earnings shock or a gap on Monday morning. This is the engine for money that is measured in years, and it behaves accordingly: fewer trades, smaller size, and long stretches where the most interesting thing in the log is a flat day.
- Instruments
- Large-cap US equities and index ETFs — SPY, QQQ and peers
- Selection rule
- Only setups where the reward clearly outweighs the risk taken
- Active hours
- Regular US market hours, 09:30–16:00 ET — nothing held overnight
- Risk control
- Conservative sizing with a tight ceiling on open exposure
- Suits
- Long-horizon capital that values protection over excitement
- Record
- Published live on FXBlue
Illustrative monthly returns
Twelve months, consistent with the published band. Illustration, not a record.
Past performance does not indicate future results.
Live audited account
Read from the linked FXBlue account. Past performance does not indicate future results.
Samurex Mosaic
Nothing rides on one currency.
Mosaic is the argument against specialisation, run properly. Instead of one or two pairs studied in depth, it trades a wide basket of majors and crosses from a single account, so no single currency decides the month. When the euro goes quiet for a fortnight the Australian dollar is usually doing something, and when the yen is the only story in town Mosaic is already positioned in it.
Because the basket is broad, sizing has to be dynamic. Mosaic measures the recent volatility of each pair and scales its exposure to it, so a violent week in one cross does not quietly become the largest risk in the account. Correlation is watched too: two pairs that are effectively the same trade are not permitted to be held as though they were two.
- Instruments
- A broad basket of major and cross pairs from one account
- Approach
- Diversification across pairs, sessions and time zones
- Active hours
- Around the clock, wherever the basket is moving
- Risk control
- Volatility-scaled position sizing plus a correlation cap across the basket
- Suits
- Maximum spread of exposure from a single deployment
- Record
- Published live on FXBlue
Illustrative monthly returns
Twelve months, consistent with the published band. Illustration, not a record.
Past performance does not indicate future results.
Live audited account
Read from the linked FXBlue account. Past performance does not indicate future results.
Which one should you start with?
There is no ranking here, only different shapes of risk. The honest answer usually depends on how you would react to a bad month rather than how you feel about a good one.
You want the widest coverage from one engine
Mosaic. A single deployment spreads across a whole basket of currencies, so no one market can define the month. In exchange, its drawdowns are less predictable than a specialist's, because several pairs can turn at once.
You cannot stomach a deep drawdown
Harbor, or Ember. Harbor holds nothing overnight and passes on anything marginal; Ember keeps position sizes uniform and small and halts itself early. Both trade less than the others, which is the point.
You already trade FX and want it handled properly
Tandem. It is the most conventional of the four and the easiest to sanity-check against your own experience of EUR/USD and GBP/USD, which makes it a good first engine if you intend to keep watching closely.
You have capital for more than one
Run engines that do not trade the same thing. Ember and Harbor share almost no exposure; Tandem and Mosaic overlap in currencies and should be sized with that in mind. We suggest around USD 10,000 of trading capital per engine, held in your own brokerage account.
Pick one. Then go and check its homework.
Every record above is published by an auditor with no reason to flatter us. Read it before you decide, not after.
Your capital stays with your own broker · Every execution independently audited