I Tested Every Free Trading Robot in MetaTrader 5. Here's the Cemetery.

Every MetaTrader 5 installation on Earth — roughly fifty million of them — ships with a folder of free, ready-to-run trading robots. Twenty-eight candlestick-pattern expert advisors (every Hanging Man, Dark Cloud and Three White Soldiers, each cross-checked by a different oscillator), plus the two most famous sample robots ever written: Moving Average and MACD Sample.

Millions of new traders have stared at that folder and wondered the same thing: what if the answer was free the whole time?

Nobody ever tests them. So we did — all thirty, under identical, honest conditions. No cherry-picking, no screenshots from a lucky week on a demo account. The platform's own Strategy Tester, real broker spreads, factory-default settings.

How we tested

• Instrument: XAUUSD (gold)
• Timeframe: H1
• Period: January 2024 to July 2026 (2.5 years)
• Starting balance: $10,000 each (about R180,000)
• Costs: real broker spreads and swaps — the part almost every backtest you've ever been shown conveniently leaves out
• Settings: factory defaults, exactly as they arrive on your machine

Thirty robots went in. Here is every result, unedited.

The burial registry

Twenty-nine flatlined or lost. One made real money.

The famous one died worst

MACD Sample is arguably the most widely installed trading algorithm in history. It sits on every terminal, it's the file every YouTube tutorial opens first — and in thirty months of honest testing it converted $10,000 into $41.90. In rand terms: roughly R180,000 into R750.

That's not a drawdown. That's a cremation. There isn't enough left in the account to buy the airtime to complain about it.

The candlestick army fared less dramatically, mostly because the patterns fire so rarely that the robots barely trade. Twenty-eight expert advisors encoding half a century of charting folklore produced a huddle of results within about three percent of zero — statistical noise, minus commission. The same patterns your favourite R2,000 course is selling are sitting in this folder for free, and they are worth exactly what they cost.

The one that lived

And then there's the file everyone scrolls past.

Moving Average.mq5 — the plainest, oldest idea in the entire folder. Price above its average: buy. Price below: sell. No pattern poetry, no oscillator seasoning, no artificial intelligence in the product name. It finished at +13.1%, the only robot in the cemetery that earned more than pocket change.

That is not an accident, and it's the real lesson of this study. Trend-following — riding a move slowly, cutting losers early and letting winners run — is the one idea in that folder backed by decades of academic evidence. It survives because it's hard to live with: it loses more often than it wins, it goes quiet for months, and it demands a patience nobody sells on TikTok. Every flashier strategy in the folder dies of transaction costs and noise. The boring one quietly compounds.

What this means if you trade

1. Free robots aren't strategies — they're tutorials. MetaQuotes ships them as coding examples. Deploying them with real money is like driving the crash-test car home from the lab.

2. If it wasn't tested with real costs, it wasn't tested. Every robot in this study looks completely different once real spreads and swaps are charged. The spread is a toll you pay on every single trade — and most "profitable" backtests you'll ever be shown were never charged rent.

3. Boring survives. If a two-line moving-average rule beats twenty-nine cleverer robots, treat anything sold to you on its cleverness with deep suspicion. Ask to see it tested with real costs, over years, on the actual broker you'll use. If the seller can't show you that, you've learned everything you need to know.

Ask them for their cemetery

At Inkatech we build and validate trading systems, business software and AI tools the same way we ran this study: with receipts. Before we trust any strategy, it has to survive real spreads, real swaps, and years of data it has never seen — and most ideas don't. That's the point. Every bad idea that dies in testing is money that stays in your account.

If someone is selling you a trading robot, a signal group or a "guaranteed" strategy, ask them one question: show me your cemetery. Anyone who has genuinely tested their ideas has buried dozens. If everything they've ever touched is a winner, you're not looking at a track record — you're looking at marketing.

Want software that's actually been tested before it ships? Talk to us at inkatech.co.za — Pietermaritzburg, South Africa.

---

Method note: MetaTrader 5 Strategy Tester, XAUUSD H1, 2024-01-01 to 2026-07-15, $10,000 initial deposit, factory-default inputs, real broker spreads, 1-minute OHLC modelling. Results are a historical simulation on one instrument and one period. They are not trading advice, and past performance — especially this past performance — guarantees nothing.