TOKYO-EA builds Expert Advisors — rule-based trading software for MetaTrader 5. The current system, BBF_Cycle 4.0, runs four engines on four currency pairs at once: the parts lose, and the portfolio absorbs it. Twenty-three and a half years of testing are published here in full, together with the weaknesses we found. Where it sits: above your money market fund, below your land.
How it works, and where each part stops working
Three mechanisms do the work. Each one has a condition under which it does not help, and those are printed next to it rather than in a footnote.
- Losses are closed, then chased down. Each engine cuts a losing position at a fixed distance and opens a recovery position to earn it back, rather than holding and hoping. Limit: this is not a high win rate. Roughly two days in five close in the red.
- Four engines, four pairs. One pair's bad year is diluted by the other three — that is why the year that lost money in the previous generation finished up in this one. Limit: all four pairs are yen crosses. A violent move in the yen hits all four at once.
- Position size shrinks automatically while the account is falling. The mechanism is built into the software and cannot be switched off. Limit: it reduces drawdown, it does not remove it — the worst fall inside a single year was 23.5%.
The reference numbers
Compounding at 0.135% of account equity (the default) · ¥1,000,000 (≈ $6,500) start · January 2003 – June 2026
The same system on shorter clocks — measured on the ¥100,000 default run, because “annual return” hides how the money moves
These are backtest results — a simulation over historical prices, not a live trading record. They do not predict future results. Compounding assumes the trade size grows with the account without limit; in real accounts, broker lot limits and liquidity mean the later part of such a curve is not achievable as drawn. A smaller account behaves differently and the difference is not in your favour: at ¥100,000 the drawdown floor is set by the platform, not by your setting. The loss statistics above come from that ¥100,000 run and, for the 246-day figure, from the fixed-size run — the exact basis of each is set out here.
Two generations, both still published
What this is — and what it is not
What it is
- An Expert Advisor: software that applies a fixed set of rules on your own MetaTrader 5 account.
- Tested over 23.5 years on eight synchronised symbols, under deliberately pessimistic cost assumptions.
- Documented. Every test report can be downloaded and read in MetaTrader yourself, including the runs where we broke it on purpose.
- A tool for people who do not want to become traders. The rules are executed by software, not by your mood at 2am.
What it is not
- Not a fund, a managed account, or a scheme where you send us money. We never hold client funds.
- Not an AI bot. The rules were written by a person, are fixed, and do not learn from your account.
- Not a system without losses. It has losing days, losing weeks and losing months, and a future year can finish negative.
- Not investment advice. What you do with your money is your decision.
Verify it, don't trust it
A record with no losing year is exactly what a curve-fitted result looks like, so we publish the material needed to test that suspicion: the test conditions, every year, the loss statistics by day, week and month, the settings at which the system breaks, and the original MetaTrader 5 report files.
Where it sits in a portfolio
This is not a replacement for a money market fund, a SACCO or Treasury bills. Those are the stable part of a portfolio and should stay that way. Rule-based trading software belongs in the part you have set aside for higher risk — an allocation, not a home for money you cannot afford to lose.
It also earns in a different currency from your local savings, which changes what it does to a portfolio. That is a question worth thinking through slowly: how it compares with what you already hold.