Release 2.00 at the workshop: five tracks instead of seven, and one rule

On 29 August we met online for three hours and forty minutes. It was a release workshop: the first part covered what changed between 1.27 and 2.00, the second built a pair of set files from scratch on a shared screen. Not a lecture but joint work — participants repeated every step on their own machines, and they chose the direction of the first set themselves.

This is a report, not a manual. The manual runs to 270 pages and lives elsewhere; here we describe what stays with you from that meeting.

Five tracks instead of seven

The loudest change in 2.00 is disarmingly prosaic: MetaTrader allows 1024 input parameters and we ran out of room. To add the new mechanisms, two logic tracks had to go. Five remain.

This is not a limit anyone can argue with — it is hard, and it comes from the platform, not from a design decision of ours. It is worth knowing, because it explains the pace: three weeks passed between 1.27 and 2.00, and the rebuild touched the foundation, not the facade.

The practical consequence for anyone holding older files: if your set did not use the sixth and seventh button, it keeps working unchanged. If it did, those tracks have to be moved to free slots, which is one action in the panel. Any position opened earlier by the sixth or seventh button has to be closed before migrating — the new version will not adopt it.

Beyond that: oscillator extreme modes that end a run of signals coming out of a single low; the choice of whether all filters must agree or one is enough; position size scaling for every track rather than only for the protective mechanism — and in both directions, because a multiplier below one deliberately shrinks the follow-up positions; and a decision and execution log, a full trace of what the tool decided and what the broker did with it.

One rule that holds the whole set together

While building the set, one sentence kept coming back, phrased several ways:

When the market moves with the direction, the tool should work. When it turns, the tool should step out of the position rather than add against the move.

The whole construction we built live serves that second case. Three tracks: the main one holds the primary direction; the opposite one only starts when the market turns, and its job is to walk the position out; the third returns to the main direction when the market does, and shuts down together with the protective mechanism.

It sounds simple. In practice it means a dozen or so parameters that have to agree with each other — which is exactly why we did it on screen, key by key, instead of sending out a finished file.

The second file was built as a mirror of the first. Here came the first thing that surprised some participants: flipping the direction alone is not enough. The oscillator thresholds and the side watched by the protective mechanism have to be reversed too. A file with only the direction changed looks correct and does something entirely different from what it appears to.

Five things that only surfaced in practice

This is the part you cannot write in advance. It surfaces when a dozen people do the same thing at once on their own platforms.

• Sizing from account balance rounds up. With a rule of „one hundredth per hundred", an account past half the threshold starts opening a double position. If the second position should only appear once the account doubles, enter a value higher than half the threshold. One number, and a quiet change in exposure.
• Missing arrows on the chart are usually a missing system font, not an indicator fault. The arrows are not drawn by the platform; they come from a font installed in Windows.
• A button name with non-ASCII characters is not saved. It looks like an ignored click; a plain-ASCII name is enough.
• Virtual exit levels cut both ways. The broker cannot see them, so it cannot trade against them — but on a sharp move the close can land further away than the setting suggests. That is a deliberate trade-off, not an oversight.
• Broker costs belong in the test. A test without commission describes a different world from the one the file will work in. We entered them during the workshop and the run looked different — more on that below.

The same file, different runs

The most interesting observation of the day had nothing to do with configuration. Participants ran the same file over the same period in the strategy tester — and got different runs.

The differences come from the broker and the data: the tick model, how ticks are archived, the spread, the way orders are filled. Historical ticks are often thinned out compared with what the platform sees live; one broker widens the spread differently from another. There was also a question about restricting the tool to quieter sessions — left open, but with a concrete suggestion of where to start observing.

The conclusion came back several times, and we would rather state it plainly: a tester run does not carry over to a live account. The tester is there to check whether the tool does what we asked it to — whether the tracks switch on where they should, and whether it exits when it was meant to exit. Nothing more.

What a text cannot carry

What setting all of this up actually looks like. Which field turns yellow and why you have to click „apply". What the chart shows when one threshold moves. What the moment looks like when the protective mechanism arms the second track, and what happens in the basket a second later.

The recording of the whole meeting is available — three hours forty minutes, with Polish and English subtitles and a chapter list, so you can jump straight to the part you need. Workshop participants have it in their panel at no extra cost. Everyone else can buy access:

See the details and buy access to the recording

We described the full run of the meeting, with its chapter list, separately: the workshop summary.

The next meeting is planned for the last Saturday of September.

O!kurra is an educational tool. Nothing here is investment advice. Trading carries risk — learn the method and test it on a demo account.