Two weeks of Open Beta: what you reported, and version 1.2.1

We opened the beta on August 8 and said we would review every report and ship fixes quickly. Here is the accounting, and the update that came out of it.

1. Already working — three reports we checked and closed

We checked these against the current build rather than assuming, and said so on each thread. A typo'd factor name used to run instead of failing. This one is fixed and has been for a while. The editor marks an unknown name with an ✗ as you type, and a run that reaches one stops with an error instead of treating the condition as false. The member who reported it lost two days to a backtest of a strategy that was never actually tested. That is the worst kind of failure and it is the one we care most about. Connection status stayed green after a session dropped. The status bar reflects data freshness rather than the socket: a stream with no update for seven seconds shows as stale, alongside feed lag and received ticks per second. A quiet feed and a dead feed do not look the same. Symbol entry was case-sensitive. It is not. Both places you type a symbol normalise as you type.

2. Fixed in 1.2.1 — two that were real

The daily loss limit did not survive a restart. This is the serious one. Your running loss lived only in memory, so restarting mid-session handed you a full fresh allowance on a day you had already spent most of it. A limit that quietly resets is worse than no limit, because you believe you are protected. The daily total is now read back from your saved trade history when auto-trading starts, and if that read fails the app says so rather than silently starting from zero. A skipped buy signal now says why. When the computed order size landed below one unit, a small betting amount against a high-priced symbol. Nothing happened and nothing was said, which is indistinguishable from the strategy not firing. The member who reported it spent time doubting entry conditions that were fine. The reason now appears on screen with the arithmetic behind it, once per symbol rather than on every tick, and backtests report the same count. Both of those are the same defect in different clothes: the software knew something and did not tell you. That is the category we will keep prioritising.

3. Getting 1.2.1

The app updates itself. If you would rather not wait, the installer on the download page is already the new version. Thank you for the reports — including the ones that turned out to be already fixed, because checking them is how we found out our own notes were out of date.