@mt5_tinkerer

Eleven silent restarts is exactly why I alert on everything now — the failure mode of a watchdog is that it works so well you forget the underlying problem exists. Boring ops beats fast hardware, every single time.

Count me in for the replication. If we standardize the log format — server time, requested price, filled price, spread at request, order type — I will run it on my two brokers and we can merge datasets. Four brokers' worth of real numbers beats forty threads of opinions.

Posting in English, hope that's ok. One thing nobody mentions before switching: MT5 has netting and hedging account modes. If your EA opens multiple positions per symbol (grids, pyramiding, partial entries), you need a hedging account — on netting it all collapses into one position and your logic…

Your StrategyQuant rejection rate rings true. My filter: discard anything with in-sample PF above 2.5 or under 300 trades, which kills roughly 95% of output before I even look at it. The tool is a hypothesis factory — treat it that way and it earns its keep; treat it as a strategy vending machine…

Clear — so for my M15 system targeting 25 pips, real ticks + variable spread is worth it, and I can stop obsessing over the percentage number. Thanks.

For MT5 execution I run a local bridge EA that polls my server every 2 seconds. Latency is fine for anything slower than M5 scalps. Webhooks into MT5 directly are painful; polling a queue is boring and boring is good.

You answered your own question: broker availability and community. Every EA, indicator, and forum answer assumes MetaTrader. C# is a genuine joy to work in; the ecosystem is a ghost town by comparison. Network effects beat language design.

One strategy class per symbol instance, registered in an array. The EA loops the array on each tick or timer and calls strategy.OnTick(). All state — open positions, last signal bar, filters — lives inside the instance. Changed my life when I went from 2 to 12 symbols. No more symbol-specific if…

I am chained to exactly one legacy indicator. Story of my life. Porting it next month, then MT5 for good.

Second the point about reviewing output. Any generator will happily produce an EA that compiles but has logic bugs. My habit: attach it to the visual strategy tester and just watch what it does for a month of ticks. Logic errors become obvious in about two minutes.