Optimizing Expert Advisor Settings for Automated Trading

But buried inside the boilerplate are four technical knobs that actually matter, and the JSE just dropped new algo trading rules worth tracking. Here's the cut that survived my filter.

Execution gates: the only edge retail still has

A 1.5-pip spread ceiling on majors and a 2.0-pip slippage filter are not innovations — they're survival thresholds. If your EA fires through a 3-pip spread during an NFP release, your backtest is fiction. I treat every rejection on these filters as saved drawdown, not missed alpha.

The latency point is sharper than most writers credit. Running the same logic on a home machine introduces 50–200 ms of jitter. On a VPS colocated with the broker, you drop below 2 ms. For a mean-reversion scalper on EUR/USD, that delta is the difference between capturing the spread and donating it to the HFT shop on the other side of the wire. Edge doesn't survive physics.

Position sizing that scales with chaos

The 14-period ATR dynamic sizing framework is the one piece of advice I'd sign. It mechanically compresses size when volatility expands and lets it breathe when the market quiets down. That's not magic — it's risk-parity at the single-trade level. Fixed lots are how retail accounts die in clusters.

The londoninsider.co.uk piece hammers the same drum from a different angle: protecting capital beats chasing returns. Drawdown discipline, adaptive sizing, real-time spread filters — it's the same checklist, just dressed in different clothes. I agree with the conclusion and disagree with the framing. This isn't philosophy. It's arithmetic. A strategy that gives back 20% needs a 25% gain to recover. One that caps drawdowns at 8% needs 8.7%. The second system is alive longer and compounds.

What I won't endorse: any EA that runs a "mathematical recovery" or martingale variant. These are the systems that look brilliant on the equity curve until a single Wednesday news event vaporizes the account. If the source material recovers losses by increasing lot size, it's not a trading system — it's a deferred liquidation.

Correlation: the silent leverage multiplier

The Business Insider Africa piece flags this almost in passing and it's the most important sentence in the entire article. Running EUR/USD and GBP/USD EAs simultaneously is not diversification if both systems load on USD weakness. You're running one position with extra steps. Check the pair-correlation matrix before you stack strategies. Uncorrelated return streams are the entire point of running more than one system.

JSE algo rules: watch, don't panic

African Business reported on June 30 that the Johannesburg Stock Exchange introduced new algorithmic trading rules. No detail in the snippet beyond the headline. South African equity algos are a thin slice of my universe, but new venue rules usually mean new latency profiles, new order-type behavior, and occasional arbitrage windows during the transition. I'll be tracking whether the new framework leaks into FX-adjacent instruments or if it stays contained. Until the full text is public, I'm not adjusting anything — but I am reading it.

The synthesis

Strip the marketing varnish off these two articles and you're left with a checklist I've used for years: execution filters, colocated infrastructure, volatility-scaled sizing, drawdown caps, correlation audit, and zero tolerance for recovery systems. None of it is proprietary. All of it is optional. That's why most accounts don't run it — discipline isn't a download.