Keep a demo bot journal you can actually learn from

Disclosure: I develop MT5 and cTrader trading software at ExpertCodeFX.

A demo account can generate plenty of activity without producing a useful evaluation. If nobody records the settings, changes and errors, the final account balance leaves much of the story unexplained.

A simple journal helps connect the bot you intended to test with the behaviour you actually observed.

Give each configuration its own identity

At the start, record the exact software version, preset, broker symbol, timeframe and whether the account is a demo. Include the observation start time and its time zone.

Keep a copy of the applied settings. A preset filename is helpful, but also confirm what the platform shows after loading it. The journal should describe the configuration that ran.

If you later change an input, make a dated entry and treat the changed setup as a separate observation period. Otherwise, a month of results may quietly combine several different experiments.

Write down the question you want answered

“Does it work?” is difficult to evaluate. A more useful question might be whether orders follow the documented entry rule, whether the expected symbol is used or whether a particular error recurs.

Before starting, check that the licence allows the intended demo use. A package restricted to historical testing cannot answer questions about how it handles new prices on an online demo account.

Choose the observation you need and the evidence that would answer it. This prevents a profitable week from distracting you from the original technical question.

Use one short entry for each review

A review entry can contain five things:

1. The time covered and configuration identity.
2. What the documented rules led you to expect.
3. The orders, positions or quiet periods observed.
4. Relevant log messages and any interruption or manual intervention.
5. What remains unexplained and the next controlled check.

For example, a hypothetical note might say: “No entries during this observation window. Instance running; logs show the session filter rejected the observed signals. Configuration unchanged. Retain the logs and check the next eligible session.”

That is more informative than “bot stopped.” It also avoids assuming that no trades means a software fault.

Record interventions without rewriting the past

If you close a position manually, restart the instance or change the account setup, record what happened and when. Keep the original evidence before attempting another test.

An intervention may be necessary, but subsequent results need that context. A clean-looking record that omits manual actions is less useful than a messy but complete one.

Keep private account details out of anything you share. A developer usually needs the build, applied settings and relevant timestamps, not your password or full account identity.

Finish with findings and limits

At the end, separate observations from conclusions. “Orders matched the documented rule in the cases reviewed” is a bounded finding. It does not establish that every future order will behave correctly or that the strategy will make money.

Record unresolved questions, unobserved conditions and any missing evidence. Retain the report, history and logs alongside the journal so someone else can inspect the basis for your findings.

When you review the figures, ExpertCodeFX's backtest-reading checklist is a useful companion for questions about drawdown, costs and report completeness. It concerns historical reports; a demo journal adds observations from the particular online setup you are watching.

A tidy journal does not make a strategy safe or profitable. It makes the evidence easier to understand and the next question easier to choose.

Copy this journal entry

Paste these fields into your own notes. Keep account identifiers and credentials private.

Observation window and time zone:
EA/cBot name and build:
Preset and applied settings:
Broker symbol and timeframe:
Account type: demo
Question being checked:
Expected behaviour:
Observed orders, positions or quiet periods:
Relevant log messages:
Manual interventions and times:
Unresolved questions:
Next controlled check:

Continue your research

Compare bot versions fairly · Read a backtest report · Compare MT5 EAs and cTrader cBots

Read the exact product documentation and current official Store listing before evaluating a package.