Forward test vs backtest: why you need both

EA Trade Tester Demo lab How we test All guides For AI agents

Published 12 November 2026 · methods v2.5 · by TrueTest

A backtest is judged on prices that were already known when the rules were written. A forward test runs the same rules, unchanged, on prices that arrive afterwards. What our tests cannot do: no backtest of ours can say REAL, however good it is, and our forward watch cannot speed up time. It needs weeks or months of new prices before it has enough trades to judge.

Start a free forward watch See a full example first

Drop your .mq5/.mq4 code, a Python bot or an MT5 report on the home page. The first read takes seconds and needs no sign-up; the lab verdict is free too. Only the full report is paid (GBP 49: a PDF to keep plus the interactive report). The price shown is the total price: no VAT is added. TrueTest is not VAT-registered in the UK. We read your code as text and never run it as given; an outside AI service drafts the rule card only if you tick the box (see Security). Not financial advice.

What is the difference between walk-forward and forward?

Test	What it does	What it can still hide
Walk-forward	splits old data into windows: tune on one window, test on the next, move along, repeat	every window is history you can look at; if you run many walk-forward set-ups and keep the best, the luck problem returns one level up
Forward	freezes one version and runs it on prices that did not exist when it was frozen	nothing about the rules, but it needs time and enough trades

Walk-forward is a good habit while you build. It is still a backtest: the person who chose the windows had seen the whole history.

How does our forward watch work?

When you confirm a version in the lab, the newest version you confirmed is frozen and run every day on new prices from the next UTC day; it is never changed, and it is kept with your lab; REAL needs at least 60 forward trades, a t-value above the luck bar, and the forward trades must still beat the same trades entered one day later by a clear margin.

The one-day shift matters. A rule that buys in a rising market makes money without any timing skill, so the forward trades must beat the same signal entered a day later, not only zero.

Why is PROMISING the best verdict a backtest can get?

Because the luck bar can only count the versions somebody reports. A backtest that clears every check is a good candidate, and our verdict for it is PROMISING. New trades cannot have been chosen with hindsight, so they are the one kind of evidence the version count does not undermine.

How long should a forward test run?

Until it has the trades, not a fixed number of weeks. A strategy that trades five times a day reaches 60 trades in a few weeks; one that trades once a week needs more than a year. Judging it early is the same mistake as stopping a backtest at its best month.

Questions people ask

Can I change the settings during the forward watch?

Then it is a new version with its own forward watch. A frozen version is never changed, so its record stays honest.

Does a demo account count as a forward test?

Yes, if the rules were frozen before the trades and every trade is kept. Fills on demo can be kinder than live.

Is a good forward test a promise of profit?

No. It makes luck a less likely explanation for the result so far; markets change. Not financial advice.

Does the forward watch cost anything?

No. It is free, like the lab verdict.