Abacuquant Portfolio

AbacuQuant Portfolio: a multi-configuration Expert Advisor

Up to 48 independent trading configurations. Multiple instruments. One chart. Shared account-level risk control.

A multicurrency Expert Advisor for MetaTrader 5 trading EURUSD, GBPUSD, USDJPY and AUDUSD from a single H1 chart, with validated presets included, a full manual mode, and no martingale or grid anywhere in it.

Most Expert Advisors are built around one strategy, one set of parameters and one instrument. When the market changes character during the day, that single configuration is right for part of the session and wrong for the rest of it.

AbacuQuant Portfolio runs multiple independent configurations at the same time, each designed for a specific combination of market conditions and trading hours. Each has its own trading window, strategy combination, indicator periods and ATR-based stop and target. A configuration designed for the European session operates independently from one designed for a later session.

Risk is not managed in isolation. The EA evaluates exposure across every configuration and instrument it manages, including when it is running on multiple charts.

Understand the trade-off

The shipped presets use a relatively wide stop and a comparatively close target. Across the validated presets the average stop is approximately 5.8 ATR and the average target approximately 1.1 ATR.

This creates a very specific trade distribution:

• A high historical winning-trade percentage, around 95% in the validation results.
• Winning trades that are individually relatively small.
• Occasional losing trades that can be approximately five times the size of a typical winner.

You will see those losses. They are part of the design, not a malfunction, and they are already included in every historical result quoted below.

With this stop and target geometry, the theoretical break-even win rate is approximately 84% before trading costs. The 95% figure should therefore always be evaluated together with the corresponding break-even level. What matters is the margin above that level and whether the margin survives out-of-sample testing.

This is also why the EA does not manufacture a high win rate by holding losing positions indefinitely. Losses are closed at predefined stops rather than being averaged into, hidden, or recovered through increasing position size.

What this EA does not do

• No martingale. Position size does not increase after a loss.
• No grid. Losing positions are not followed by additional positions to recover the loss.
• No averaging down. The EA does not add to a losing position.
• No loss-recovery logic. There is no automatic attempt to win back a previous loss.
• No trade without a predefined stop loss. The stop and target are sent with the opening order.
• No repainting. Trading decisions use completed bars, never the currently forming candle.

If you are looking for an equity curve with no visible losses, this is not that type of system. The objective is not to eliminate losses from the history, but to define how losses are allowed to occur and how account-level exposure is controlled.

Why it is a portfolio engine

Up to 48 independent configurations can run from a single chart.

Each configuration can have its own trading hours, strategy combination, indicator periods, required number of agreeing signals, trading direction, ATR-based stop and target, and position-sizing settings.

This allows different configurations to operate during different parts of the trading day instead of forcing one strategy to trade continuously under changing conditions. The configurations remain independent at the strategy level while the risk engine evaluates the account as a whole.

Validated presets

The EA includes complete, pre-configured portfolios. You do not need to optimize the shipped presets before using them.

• EURUSD — 8 configurations, covering European through US trading hours.
• GBPUSD — 6 configurations, with the widest coverage among the validated instruments.
• USDJPY — 4 configurations, concentrated in the afternoon and evening.
• AUDUSD — 5 configurations, covering European and later sessions.

EURUSD is enabled by default, so the EA is ready to run after being attached to an H1 chart. The other validated instruments remain disabled until you enable them deliberately, individually or combined as a portfolio.

What the configurations are made of

Every configuration draws from the same ten strategy components: order-flow imbalance, Fibonacci breaks, candlestick patterns, support and resistance, Bollinger Bands, RSI, moving averages, MACD, ADX and Stochastic. What defines a configuration is not the list but the combination and the agreement threshold.

There is no single "RSI preset" or "trend preset" here. Across the four validated presets, Bollinger Bands appear in fifteen configurations, Fibonacci in fourteen, ADX in thirteen, MACD in twelve, support and resistance in eleven, moving averages in eleven, Stochastic in ten, RSI in six and candlestick patterns in one. The combinations were found by systematic search over historical data and then put through the acceptance procedure described below.

What each preset measured

Results below are from the out-of-sample forward period, which the selection process never saw. "Margin" is how far the win rate sat above its own break-even line, in percentage points. "t" expresses that margin in standard errors. Values above 2 are conventionally considered statistically significant.

All four clear the same acceptance threshold independently. That is why four instruments are shipped as validated presets rather than relying on a single instrument.

What happens when you run them together

Run as a single portfolio on one account, the combined result was t = 7.60 across 1,540 trades with 158 losses, at a maximum drawdown of 10.34 percent.

The combined result is stronger than any individual preset, the best of which was GBPUSD at 4.42. The four do not simply stack their profits. They trade at different hours, on different currencies, and their losing periods do not always occur at the same time.

The cushion is another way to see the result. Based on the forward-period outcome, it would take approximately nineteen additional losing trades to erase the EURUSD result, sixteen for GBPUSD, and approximately ninety for the four running together.

The shared risk engine also affects portfolio behaviour directly. Each pair traded slightly fewer times inside the portfolio than it did alone — EURUSD 492 against 521, and USDJPY 278 against 315 — because some entries were refused when another instrument was already using the available risk budget.

Exactly how this was measured

• Broker: Darwinex, on a server clock of GMT+2 in winter and GMT+3 in summer.
• Period: 1 January 2020 to 3 August 2026, split into a selection half and an untouched forward half.
• Modelling: "Every tick based on real ticks".
• Deposit: 100,000, a single account shared by all four presets.
• Clock setting: broker winter offset input set to 2, matching that server.

The server-time setting is important. A preset pointed at the wrong trading hours is not the same preset that produced these validation results.

The two experimental presets, and why they ship anyway

XAUUSD and GBPJPY are included, marked experimental, and switched off. They are provided for research rather than presented as validated presets.

Gold: real, but too thin to sell as validated. Seven configurations, forward margin +1.08 over break-even, t = 1.00, drawdown 5.63 percent. A t of 1.00 means the margin sits about one standard error above zero, which is not sufficient evidence to treat the result as a measured standalone edge. The cushion says the same thing: 2.5 additional losing trades would erase the entire forward result, compared with roughly nineteen for EURUSD. The result is not negative. It is simply not measured well enough to carry a claim.

GBPJPY: inconsistent between halves. Six configurations. In the selection half the margin was negative at −0.33, with a t of −0.43 over roughly 37 losses. In the forward half it turned positive at +2.99 with a t of 3.82 — but this was based on only about ten losing trades, below the minimum required by the validation method.

Both instruments are included so users can research them without having to rebuild the configurations. They are for observation, not allocation. If either clears the acceptance procedure later, it can become a validated preset through a free update.

New validated instruments are added at no extra cost while your subscription is active. Candidates go through the same acceptance procedure before being presented as validated.

How a trade is decided

The EA is not a black box. The decision process can be understood from the configuration and observed in the Experts log.

1. The configuration must be active.
Each one operates only during its assigned trading window and ignores the market outside it.

2. The candle must be closed.
Signals are evaluated on completed H1 bars. The forming candle is never used to generate an entry, making the signal logic reproducible on completed bars in both the Strategy Tester and live trading.

3. Multiple signals must agree.
The active components and the number required to agree are defined per configuration. One indicator alone is never enough to open anything.

4. Position size is calculated from the stop.
For risk-based sizing, the EA determines the stop distance first, using current volatility rather than a fixed pip count, then calculates the position size that applies your selected risk percentage over that exact distance. A wider stop therefore produces a smaller position.

5. Account-level limits are checked.
Before sending an order the EA evaluates total open risk, deposit load and equity drawdown. Any one of them can refuse the trade, and refusing is treated as an acceptable outcome.

6. Protection is sent with the order.
The stop loss and take profit are part of the instruction that opens the position, not added a second later. If your connection drops immediately afterwards, the protection is already on the broker's server.

7. No recovery sequence follows a loss.
The EA does not add to losing positions, move the stop further away, or re-enter to recover a previous loss.

Manual mode: build your own configurations

The validated presets are only one way to use the engine. Manual mode gives access to the same underlying framework, with nothing held back.

You define which strategies are active, how many must agree before entry, the trading window down to the minute, indicator periods, allowed direction, maximum positions per configuration, and the ATR-based stop and target distances.

You also select how the position is sized:

• Risk based — position size follows the stop distance and selected risk percentage, the same as the presets.
• Fixed lot — the same lot size regardless of stop distance. The natural choice while researching, because it holds one variable still.
• Proportional — position size scales with account balance.

Broker minimum volume, maximum volume, volume step and symbol volume limits are all respected. Choosing your own sizing changes the sizing method, not the safety rails.

Custom optimization criterion

The EA includes a custom optimization criterion for the Strategy Tester. It does not rank configurations by net profit. It evaluates how far the observed winning percentage sits above the break-even level implied by the stop and target distances, and results built on too few losing trades can be rejected outright.

The reason is straightforward: a very distant stop with a very close target can produce a smooth equity curve and a high profit factor simply because the stop was rarely reached inside the test window. Ranking by profit can select those configurations first. This criterion is designed to reject them.

Select Custom max in the Strategy Tester to use it.

Risk management

Position size follows the stop. For risk-based sizing the lot is calculated from the actual stop distance using the broker's own contract specifications, so a wider stop produces a smaller lot and the money at risk stays constant.

It scales with your account automatically. Double the balance and positions double; halve it and they halve. You do not need to adjust a lot setting for the scaling itself.

What it is not is a fixed lots-per-balance ratio. Because size comes from each configuration's own stop distance, a configuration holding a wide stop takes a smaller position than one holding a tight stop at the same moment, while targeting the same configured monetary risk. Measured across every EURUSD entry over the validation period at the default half-percent setting, the configurations used between 0.12 and 3.07 lots per 100,000 of balance, averaging around 0.90. That variation is the mechanism working: position size moves so the intended monetary risk remains consistent.

The broker minimum lot is not ignored. On smaller accounts, and on instruments with wide stops, the minimum volume a broker allows can be larger than the volume your selected risk percentage calls for. The EA detects this and reports the configuration involved, the broker-enforced volume, the actual risk that volume creates, your configured limit, and the balance that configuration needs to size correctly.

If SkipTradeIfMinLotExceedsRisk is enabled, the configuration refuses those trades instead of exceeding your limit. On an account too small for the preset, that can legitimately mean no trades at all.

Three global risk ceilings

• Total open risk — the potential loss if the stops of all open positions were reached at once.
• Deposit load — the amount of margin that may be committed. This is collateral, not risk of loss.
• Equity drawdown — the permitted decline from the account equity peak.

All three are evaluated across every configuration, instrument and chart the EA manages, not separately for each. Running four instruments therefore does not silently multiply your permitted exposure by four.

Configuration guards also refuse configurations whose stop is tight enough that the spread can dominate the outcome, or whose stop and target sit close enough together that both could fall inside a single bar. Both situations can produce backtests that are difficult to reproduce live. The thresholds are inputs you can inspect.

Can you run this on a small account?

Yes, and rather than simply assert it, it was tested. The whole portfolio was run over the same period on a 500-unit account with every preset enabled, including the two experimental ones. The EA continued to generate trades throughout the test, the margin level never fell below 259 percent, and across roughly 2,300 out-of-sample trades the maximum drawdown was 10.3 percent on balance and 14.4 percent on equity, with 91 percent of trades closing in profit. Position size scaled automatically from the broker minimum upward as the account grew, without changing a lot setting.

Now the part that matters more than the result. Below roughly 2,500 units the risk control cannot fully maintain the configured percentage because of the broker's minimum lot. Understand why before deciding.

Every broker enforces a minimum lot. When half a percent of your account is less than what that minimum lot risks, the minimum is what you get, and the trade risks more than you asked for. Measured on that 500-unit run while the balance was still near its starting point, the typical losing trade cost about 1.0 percent of the account against the 0.5 percent configured. Roughly double. This is a broker-volume constraint, not something an EA can eliminate. What AbacuQuant does is measure it, identify the balance each configuration needs, and let you decide whether to trade it.

The cent account alternative

There is a practical way around the minimum-volume constraint on small capital.

Many brokers offer cent accounts, where balances are denominated in cents and a lot is one hundredth the size of a standard one. Deposit 100 dollars and the account reads 10,000. The money is identical; what changes is position-size granularity. The broker minimum lot can therefore represent a much smaller monetary exposure, allowing the EA to size closer to the risk percentage you actually selected.

No configuration change is required. Every lot in this EA is derived from the broker's own contract specifications rather than from a fixed monetary assumption, so a cent account is handled according to its actual trading conditions.

A cent account does not multiply returns. One hundred dollars remains one hundred dollars and the percentages are unchanged. Its advantage is finer position-size granularity. Confirm that your broker offers cent accounts for the instruments you intend to trade, and verify the setup on a demo before funding.

One thing not to do on a small account

Do not enable the experimental presets. On that same 500-unit test the four validated presets kept their worst single trade to about 3 percent of the account. Gold, in one trade, took 9.9 percent. On a small balance the experimental configurations can concentrate risk where you can least afford it, which is a second reason to leave them off. The EA ships with them disabled.

Broker server time

Trading sessions are defined relative to a reference server clock because different brokers use different server-time offsets. Every validation result quoted in this description depends on the session timing being set correctly.

The EA is shipped with AutoGmtOffset disabled by default. The default configuration uses the standard GMT+2 winter offset, matching the server used for validation, and applies the corresponding seasonal adjustment. You enter the winter figure only.

If your broker uses a different winter server offset, configure it in the inputs before running the validated presets. Your broker publishes this figure, and you can confirm it by comparing the Market Watch clock against GMT.

This default is intentional. Automatic broker-offset detection is not enabled by default because server-offset reporting can behave differently inside the Strategy Tester, and a wrong answer there could quietly move every trading window while everything still appeared to work. A fixed reference offset is predictable and inspectable. Automatic detection remains available for brokers where it reads correctly.

The offset measured, the offset used, the reference and the resulting adjustment in minutes are printed in the Experts log at startup, so you can verify the session conversion before relying on the presets.

Symbol compatibility

Brokers use different names for the same underlying instrument: EURUSD, EURUSD.pro, mEURUSD, and other prefixes and suffixes. The EA resolves the appropriate broker symbol automatically and provides an input for explicit mapping when necessary.

Magic numbers are generated per instrument so that configurations cannot collide. The EA verifies this at startup and refuses to trade if it detects a conflict.

How the presets were validated

Every validated preset passed the same acceptance procedure. Configurations that failed were not included. Testing used real-tick historical data and a forward period that was not used during selection.

1. Statistical validation. The observed win rate must sit above the break-even level implied by the stop and target structure by a margin large enough to be unlikely by chance, measured as a t-statistic on out-of-sample data.
2. Sufficient losing trades. A result built on a handful of losses is not measured, it is guessed at. Below the minimum, it is rejected regardless of how good it looks.
3. Stop validation. Nearly all losing trades must close at the stop. If they do not, the stop was never really tested and the backtest describes a market condition rather than a strategy.

Two instruments researched during development failed these criteria and were kept out of the validated set rather than included simply to make the product appear larger. The same standard applies to every instrument added from here on.

Set-up

1. Attach the EA to an H1 chart. One chart manages every enabled instrument.
2. Enable the presets or configurations you want to run.
3. Set your risk parameters and check the server-time line in the Experts log.

The log reports active configurations, trading hours, stop and target settings, the balance each configuration requires, the time adjustment applied and the symbols resolved on your broker. If something is wrong the EA reports it in plain language rather than failing silently. An on-chart panel shows live status while it runs.

Test it before you pay

Use the free demo in the MetaTrader 5 Strategy Tester. For the shipped presets use H1 and Every tick based on real ticks to reproduce the validation conditions. Set the deposit to whatever you actually intend to fund, including a small account.

You can inspect the trade list yourself and verify everything described above: how stops and targets are placed, how position size changes with stop distance, how the global risk controls behave, that losing positions are never averaged into, the actual winning and losing trade distribution, and the periods when individual configurations are inactive.

Pay particular attention to the losing trades. The historical win rate is high, but individual losses can be substantially larger than individual winners. If that distribution is not compatible with the risk you are willing to accept, this is not the right product for your account.

I would rather you discover that in the Strategy Tester for free than after committing real money.

Support from the developer

A subscription includes the software and direct developer support.

• Support in English and Spanish, from the developer rather than a template.
• Installation assistance — MetaTrader 5, broker configuration, symbol mapping and VPS setup.
• Experts-log assistance — help understanding configuration, risk and broker-compatibility messages.
• Risk-setting guidance — help judging whether a configuration suits your available account size.
• A direct answer when the configuration is not suitable. If your account or broker conditions are not appropriate for a preset, the objective is to tell you clearly rather than sell you a workaround that does not work.

Updates are part of the subscription. New instruments are researched continuously, and those passing the acceptance procedure are added at no additional cost while your subscription remains active. Engine improvements reach you the same way.

Requirements

• MetaTrader 5.
• Hedging account for the multi-configuration presets. The validated presets maintain independent positions on the same instrument, each with its own stop. A netting account combines positions on the same symbol by platform design, so the configurations cannot remain independent and the presets will not behave as validated.
• H1 chart. One chart manages all enabled instruments.
• A broker providing the required instruments and suitable trading conditions. A low-spread account is recommended.
• A VPS is recommended for continuous operation but is not mandatory.

Users with netting accounts can use manual mode with a single configuration, subject to their broker and account conditions.

Account size

Sizing works properly from roughly these balances, measured per preset over the validation period:

USDJPY about 2,100  |  EURUSD about 2,500  |  AUDUSD about 2,500  |  GBPUSD about 4,200

All four together, comfortably: about 4,200. Below that it still trades, but the broker minimum lot pushes real risk above what you set, as explained in the small-account section. On a cent account these figures apply in cents, which is the point of using one.

The EA prints the exact balance each configuration needs at startup, and the demo will tell you for free.

Honest limitations

• This is not a set-and-forget product. The presets are based on historical research and will require revision as market conditions change. That is what the update stream is for.
• Trading is not continuous. Configurations operate only during their defined windows, so periods without trades are normal and are not a malfunction.
• Wide-stop configurations require a larger account to size correctly. Below roughly 2,500 the broker minimum lot can substantially increase real risk per trade unless you use a cent account.
• The historical winning percentage is high, but individual losing trades can be substantially larger than individual winning trades. If a run of losses would make you switch the EA off, this structure will not suit you.
• Every figure in this description was measured on historical data over a specific period, on one broker, at one clock setting. These figures describe what happened, not what will happen.
• Historical backtests and validation results do not guarantee future performance. No Expert Advisor can guarantee profits or eliminate the risk of losing capital.

Trading involves risk of loss. Only use capital you can afford to lose, and evaluate the EA under the broker, account size and trading conditions you actually intend to use.

AbacuQuant Portfolio is built to make its trading and risk structure visible rather than hide it. The Strategy Tester lets you examine the trade distribution, the drawdowns and the behaviour of every configuration before committing real capital.