How to build an EA portfolio with software.

You can build a diversified Expert Advisor portfolio in two hours using free tools. Here's the exact process — including the correlation analysis that 99% of retail traders skip, and the worked example that proves the math.

The math behind portfolio construction is one of the most well-understood ideas in finance. Harry Markowitz won a Nobel Prize for it in 1952. Hedge funds have used it for decades. And yet, most retail algorithmic traders ignore it completely — stacking three trend-following EAs on the same currency pair and calling it "diversification."

The good news: with the right software and a few hours of work, you can do this properly. The tools are mostly free. The math is high-school level. The benefit — a smoother equity curve, lower drawdown, and more sustainable returns — is significant.

---

Why software-assisted portfolio building matters

Building a portfolio by intuition seems sensible: pick three EAs that feel different — one on gold, one on indices, one on forex. Done, right?

Wrong. The intuition fails because "feels different" doesn't equal "actually uncorrelated". Two breakout EAs running on different symbols can have 0.8+ correlation if they both ride the same global risk-on/risk-off cycle. An EA on EUR/USD and one on GBP/USD aren't a portfolio — they're the same trade twice. Without quantitative analysis, you can't see this.

Manual correlation calculation across hundreds of daily P&L points, across multiple EAs, becomes impossible. This is exactly the job software is built for.

---

The 6-step process

Here's the sequence used by people who actually build EA portfolios that work. Each step is concrete, repeatable, and uses tools you can access today.

Step 1 — Backtest each EA individually first

Before you can combine strategies, each one needs to be validated on its own. A bad EA doesn't become good by being part of a portfolio — it just contaminates the whole.

For each candidate EA, you need:

• A clean backtest with 99% modelling quality using real tick data
• At least 5 years of history covering multiple market regimes
• 500+ trades for statistical significance
• The Strategy Tester report exported as HTML or XML
• The full trade-by-trade history exported as CSV (this is critical for correlation analysis)

In MetaTrader 5, after running a Strategy Tester pass, right-click on the results and choose "Save Report" → "XLSX". This gives you a spreadsheet with one row per trade, including open/close times and P&L — exactly what you need.

Step 2 — Calculate pairwise correlations

Correlation measures how two strategies move together over time. The values range from -1.0 (perfectly opposite) through 0.0 (independent) to +1.0 (move identically).

What you correlate

Don't correlate individual trades — they happen at different times. Correlate daily (or weekly) P&L instead. For each EA, calculate a single P&L value per day, then check whether the daily P&L of EA A moves in sync with EA B over the same period.

The good news: you don't have to do this calculation manually. Quant Analyzer (covered in the next step) handles the full correlation matrix automatically once you import your strategy reports. The math runs in the background; you just look at the result.

Interpreting the results

• Correlation 0.0 to ±0.2: Excellent. The two strategies are effectively independent — exactly what you want.
• Correlation ±0.2 to ±0.5: Acceptable. Some shared exposure, but most diversification benefit remains.
• Correlation ±0.5 to ±0.8: Problematic. The strategies share too much risk; combining them adds little benefit.
• Correlation above ±0.8: Effectively the same strategy. Don't waste capital on both.

The goal: assemble a portfolio where every pairwise correlation is below 0.3. This is harder than it sounds — most EAs from the same developer have shared logic that produces correlated returns even on different symbols.

Step 3 — The software you actually need: Quant Analyzer

For 99% of retail traders, there's one tool that does the job better than anything else — and the free version is enough.

Quant Analyzer (StrategyQuant family) is purpose-built for MetaTrader strategy analysis. It imports Strategy Tester reports directly, calculates correlation matrices automatically, builds combined portfolio equity curves, and runs Monte Carlo robustness tests — all in a clean interface designed for algorithmic traders, not finance professors.

How to use it for portfolio analysis

1. Download the free version from strategyquant.com/quantanalyzer
2. For each EA, run a backtest in MT5 and save the Strategy Tester report
3. In Quant Analyzer: File → Import Strategies and load each report
4. Open the "Portfolio" tab — the correlation matrix is generated automatically
5. Quant Analyzer also shows the combined equity curve, aggregate drawdown, and portfolio-level metrics in the same view

What used to require hours in Excel — and was error-prone the whole way — now takes about 15 minutes once you have the individual backtest reports ready.

What the free version gives you

• Correlation matrix — automatic, pairwise, across all imported strategies
• Combined portfolio equity curve — see the aggregate result visually
• Portfolio-level metrics — net profit, drawdown, Sharpe, profit factor for the combination
• Monte Carlo simulations — stress-test how the portfolio behaves under randomized trade ordering
• Trade analysis — see which strategies contribute most/least to overall performance

There's a paid tier with more advanced features (custom optimization, walk-forward portfolio analysis, automated robustness tests), but the free tier covers everything you need to build and validate a multi-EA portfolio.

Step 4 — Allocate risk properly (not just equal lot sizes)

The most common mistake retail traders make when running multiple EAs: setting them all to the same lot size.

The problem: each EA has different volatility, different stop-loss sizes, and different trade frequencies. Equal lot sizes don't produce equal risk — they produce wildly unbalanced exposures where one EA dominates the portfolio.

Equal-risk allocation

The correct approach: each EA should contribute roughly equal expected risk to the portfolio. This means scaling position sizes by the inverse of each EA's volatility.

Concretely: if EA A has a typical trade with 30 pips of stop loss and EA B has 80 pips, you don't run both at the same lot size. You run EA B at roughly 30/80 = 0.375x the lot size of EA A. Now each trade risks the same percentage of equity.

Most professional EAs (including the Jara Trading lineup) handle this internally via a RiskPct input. Set every EA in the portfolio to the same RiskPct (e.g., 0.5%), and they'll auto-calculate equivalent position sizes regardless of stop distance.

Step 5 — Simulate the combined portfolio

This is where Quant Analyzer pays off. Now that you have:

• Validated individual backtests for each EA
• Pairwise correlations between them (auto-calculated)
• Risk-adjusted position sizing

You can construct a combined equity curve showing how the portfolio would have performed historically. In Quant Analyzer, this is the "Portfolio" tab — it displays the aggregated equity curve with one click after you've imported your strategies.

The metrics that matter at portfolio level:

• Aggregate Profit Factor — usually higher than any individual EA's PF, due to diversification
• Aggregate Sharpe — should improve significantly if correlations are low
• Portfolio Max Drawdown — much lower than any individual EA's drawdown when correlation is low; this is the main benefit
• Recovery Factor — net profit divided by max drawdown; this metric typically improves the most

Step 6 — Stress-test the worst case

A portfolio that looks good in average conditions can still fail under stress. Before deploying, run these checks:

Worst month / worst week

Find the single worst calendar month in your portfolio history. Is that drawdown survivable for your psychology and account size? If your worst backtest month is -8%, expect real-life worst months to be -12 to -15%.

Correlation breakdown during crises

During 2020 (COVID crash) and 2022 (rate-hike sell-off), strategies that appeared uncorrelated in normal markets often became correlated as everything sold off together. Check your portfolio's behavior specifically during these periods. If your "diversification" disappears when you most need it, you don't actually have a portfolio.

Monte Carlo simulation

Quant Analyzer's Monte Carlo feature randomizes the order of historical trades to simulate "what could have happened with the same strategy but different luck." A robust portfolio shows consistent results across thousands of simulated paths. A fragile portfolio shows huge variance — meaning your historical results were partly luck.

This single feature is one of the strongest reasons to use Quant Analyzer instead of Excel: it tests whether your portfolio is genuinely robust or just got lucky in backtest.

---

A worked example: the Jara Trading portfolio

To make this concrete, here's the analysis for the three EAs published on this site. Each was backtested independently over 2020 to mid-2026 on its respective markets, then combined into a single portfolio.

The candidates

• Gold Breakout Fusion — XAUUSD D1 multi-strategy (5 sub-strategies)
• Range Breakout Fusion — Multi-symbol ORB on XAUUSD, USDJPY, BTCUSD (different sessions)
• New York Breaker — US100 M15 session breakout (NY open)

The three EAs target different markets, different timeframes, and different trading sessions — the maximum number of diversification axes available within the breakout strategy family.

The correlation matrix

All three correlations are essentially zero. This was the design goal: select strategies whose payoffs don't move together. In practice, achieving near-zero correlation between three EAs is rare — and usually only possible when each is genuinely targeting different market mechanics.

The combined portfolio result

With each EA running at 0.5% risk per trade on a starting deposit of €10,000:

The most important number is Maximum Drawdown 5.89%. Each EA on its own has a higher drawdown than the combined portfolio — that's diversification benefit in numbers. The portfolio Sharpe improves materially over any individual EA's Sharpe, exactly as Markowitz's math predicts.

---

Common mistakes to avoid

1. Treating correlated EAs as diversified

Five trend-following EAs on five forex pairs isn't a portfolio — it's the same trade five times. Most retail "portfolios" fall into this trap. Always verify correlation, never assume it from intuition.

2. Equal lots instead of equal risk

As discussed in Step 4, identical lot sizes across EAs with different stop distances produces wildly unbalanced exposures. Use percentage-based risk sizing, not fixed lots.

3. Skipping individual validation

Combining three bad EAs doesn't produce a good portfolio — it produces a worse bad EA. Each component must be validated standalone before portfolio analysis even begins.

4. Ignoring stress periods

Backtests over "normal" periods can hide fatal correlations that emerge during crises. Specifically test 2020 Q1, 2022 H2, and any other periods of extreme volatility for the assets your EAs trade.

5. Over-optimizing the portfolio weights

It's tempting to use software to find the "optimal" portfolio weights that maximize Sharpe. Resist this. Those weights are fit to historical noise as much as to real edge. Equal-risk weighting is more robust than optimization-derived weighting for retail portfolios.

---

Final thought

Portfolio construction is the single highest-leverage skill in retail algorithmic trading. Most traders spend years searching for the "best EA." The few who actually compound capital spend their time on portfolio composition: finding combinations of imperfect strategies that, together, produce something better than any individual component.

The tools are accessible. The math is approachable. The benefit is real and compounds over years. Spend a Sunday afternoon learning this and your equity curve will be smoother for the rest of your trading career.