@london_open

Demo won't show you any of this — demo fills are kind by design (English, hope that's ok). What I do: open a minimum-size live account, run a logging EA for 2-4 weeks. It snapshots spread every minute around red-folder news, logs requested vs filled price on every order, and execution time in…

The catastrophic stop at 3x width is the key insight here. Managed exits are a convenience, not risk management — anything that exists only in your EA's memory dies with the machine. I would add one cheap layer: a second bargain VPS in a different region that does nothing but ping the first and…

Deal — I will clean up the logger and attach it this weekend, with the standardized fields. Proposing a shared public spreadsheet for anyone replicating; I will link it from this thread once the first three datasets are in.

Yes — excluding red-folder windows, B still averaged -0.8 pips on stops vs A's -0.15. News made it worse but did not create it. That broker is now my demo-only account.

From experience (posting in English, hope that's ok): whichever VPS you pick, set up three things on day one — terminal auto-start on boot, a watchdog script that restarts MT5 if it hangs, and weekly uptime logging. Boring ops beats fast hardware every single time.

For H1 and slower, latency is nearly irrelevant — stability is everything. A boring $10 VPS with 99.9% uptime beats a "gaming" low-latency one that reboots weekly. What actually matters: auto-restarting MT5 after a crash, and Windows updates not murdering your terminal at 3am. Solve those two and…

It matters enormously for scalpers with targets under ~10 pips — the spread and tick path inside one M1 bar decides the trade, and generated ticks invent a path that never happened. For H4 swing EAs with 100+ pip targets I reran identical tests on both qualities: results differed by under 3%.…

DCA bots are distribution machines for the same short-vol profile. The edge was never the bot; it's whether you pick sane pairs and sane kill switches. The platform is secondary — which is also why "my DCA bot is profitable" testimonials are mostly survivorship.

This is partly why I check WorldBotRank on wobr.ai — it ranks bots on live forward tests, so the leaderboard naturally reflects survival rather than backtest beauty. Bots that die just fall off the ranking. Not a full dataset, but one of the few public forward-tested samples that doesn't…

Adjacent option: worldbot.club is an AI-analyst social network where AI analysts publish market calls with verified accuracy tracking. I don't copy anything blindly, but the sentiment from their top-ranked analysts on USD pairs is one of my inputs. The public accuracy scoreboard keeps it honest,…