About

Hi, I’m Tommy

Builder of RoboticEA and of the C++ backtesting engine behind it. Writes up the studies here — the MetaTrader 5 parity work, the archetype study, and the instrument-level tests — with the losing years left in.

Tuomas "Tommy" Jokela

Founder, RoboticEA

Get in touch Facebook Instagram

What I work on

The engine

RoboticEA runs on a C++ backtesting engine written from scratch rather than a wrapper around someone else's tester. It charges spread, commission and overnight swap inside the simulation, and it is the same engine that produces every number published on this site.

The parity work

The engine is checked run-for-run against MetaTrader 5's own Strategy Tester on real tick data, because an engine that quietly disagrees with the platform you will actually trade on is worth nothing.

Read the parity study →

The studies

Large measured studies rather than opinion pieces: 594,580 backtests across 34 strategy archetypes, and instrument-level tests like the gold study — published with the losing years and the failed strategies left in.

Read the archetype study →

How I publish numbers

Most trading content is a screenshot of the best week somebody ever had. These are the rules everything on this site is held to, and you can check each one against any page here.

A return is never published without its maximum drawdown beside it.
Every strategy is tested on each year separately, and the losing years stay in the table.
Costs are charged inside the simulation — real recorded spread by default, never a flat assumption on gold or indices.
Numbers in a post are read from the study artifact that produced them, so a re-run updates the post rather than contradicting it.
Sample size is stated. A result from a strategy that took four trades is labelled as one.
No signals, no managed accounts, no performance promises. The software is a research tool.

Writing

Data studyMT5 Strategy Tester accuracy: real ticks vs synthetic, and why backtests disagree100 strategies, MetaTrader 5 and our engine, two tick models. Real ticks cut the median return gap from 8.97 to 6.07 pp, while exact trade matches fell from 43 to 5. Why both are true.Data studyBest EURUSD expert advisor: what happens to the top picks after they are pickedPicked on 2018-2023, the ten best EURUSD strategies compounded a median +961% and then -52% in the years after. What a holdout test over 8,098,335 backtests shows.Data studyBest GBPUSD expert advisor: do gold strategies transfer to cable?5,298,129 backtests, 2018-2026, on real recorded spread. How often a gold-selected strategy makes money on GBPUSD, why the rank correlation misleads, and what a holdout does to the best cable results.Data studyBest gold expert advisor for MT5: what 9 years of XAUUSD backtests actually show186,637 XAUUSD backtests, 2018–2026, on real recorded spread. How many gold strategies are profitable, how many survive every single year, and what a flat-spread assumption hides.Product guideFree expert advisor for MetaTrader 5: generate, backtest and download your own250 free tokens, no card, and MetaTrader 5 export on the free tier. Generate an EA, backtest it year by year, read the MQL5 and download the .mq5 yourself.Data studyDo forex robots actually work? What 700,000 backtests show705,495 backtests. How many trading robots make money, why the winners stop winning when you move them, and what a cheap cost model hides.Data studyBest timeframe for expert advisors: what 88,750 backtests sayTwo instruments, eight years, five timeframes, one stratified sample. The ranking is monotonic — and switching the costs off closes 56% of the gap.Data studyBest forex pairs for expert advisors: all 25 instruments tested73,250 backtests across 25 instruments, 2018–2025, on real recorded spread. What each pair actually charges a strategy, what a pip is worth, and why a “best pair” league table is an artifact.

All posts →

What I don’t do

I don’t manage money, sell signals, run a fund or take a share of anyone’s trading. RoboticEA is software you run yourself: it generates strategies, backtests them and writes a MetaTrader 5 expert advisor you can read before it ever touches an account. Nothing I publish is investment advice, and every result on this site is a historical simulation — past performance does not predict future results.

Come and check the numbers.

Try it free