Member since Jun 20, 2025 · 3 threads · 11 replies
Same strategy, three tools, one weekend: EMA 50/200 cross on EURUSD H1, ATR(14) trailing stop, no trades during red-folder news, 0.5% risk per trade. I counted every manual…
Pine is great for prototyping but execution is DIY. What's the most reliable alert-to-execution bridge people are using in 2026? Webhook to a custom server? A third-party bridge…
I'm building walk-forward optimization into my process. The literature says everything from 4 to 50 windows. Practically: how many in-sample / out-of-sample windows do you run…
On Walk-forward results for a simple EURUSD trend EA, warts and all 15 hours ago
Genuine question on the wandering Donchian length: 40 to 70 across windows would fail my stability bar. How do you distinguish "acceptable parameter drift" from "no robust core"? I have never found a clean rule for where that line sits.
On เครื่องมือ AI ช่วยเทรด อันไหนใช้ได้จริง อันไหนแค่คำโฆษณา 3 days ago
Honest taxonomy after a year of using these daily (posting in English): general LLMs — ChatGPT, Claude, Gemini — are genuinely useful for reviewing MQL code and explaining cryptic compiler errors. They are useless for price prediction, and they hallucinate MQL5 function signatures often enough that…
On Six months of backtest-vs-live divergence logging on 2 EAs — exact numbers 5 days ago
Replicated the news-fill issue on my own mean reverter. My tester fix was real ticks with variable spread, but only around red-folder events — much cheaper than full real-tick-everything, and it closed most of the fantasy-fill gap. I also cross-checked the same period in QuantMogul's walk-forward…
On I built the same strategy in FXDreema, StrategyQuant, and an AI generator — manual fix counts inside 6 days ago
Confirmed — basket logic broke all three of mine too, each in a different and creative way. Next test on my list is a multi-symbol portfolio EA; I expect similar failure and I will log the fix counts here either way.
On Regime change: strategies that worked in 2024 trends dying in chop? 2 weeks ago
I tried Hidden Markov Models on returns: 3 regimes — trend up, trend down, range. Worked okay, but honestly the better answer is boring: run both trend and mean-reversion systems with low correlation and let the combination smooth the equity curve. Ensemble over prediction.
On Monte Carlo on backtest results — what drawdown percentile do you size for? 3 weeks ago
Same, 95th. Also worth running Monte Carlo with trade removal — randomly drop 10% of trades per run. It tests whether your edge depends on a handful of outlier winners. Many backtests quietly do.
On LLMs for market sentiment — practical setups? 4 weeks ago
I score central-bank statements with an LLM (hawkish/dovish on a -2..+2 scale) and feed it to my EA as a regime flag. Cost is trivial, about $3/month; latency is fine for H4. As a standalone signal it was noise. As a filter blocking longs into hawkish surprises, it cut my drawdown meaningfully.…
On Structuring a multi-symbol EA without it becoming spaghetti 4 weeks ago
Two more things that matter at 6+ symbols: do the heavy analysis in OnTimer (once a second) instead of OnTick, and create all indicator handles at init and cache them. Multi-symbol EAs that create handles on the fly leak resources and eventually freeze the tester.
On Walk-forward optimization: how many windows is enough? 1 month ago
The parameter stability chart per window is exactly what QuantMogul's WFO report shows, and it's what convinced me my last "robust" strategy was actually noise. Rejected it before it cost me real money. Rejection is the feature nobody markets.
On Is anyone actually making money with ML-based signals? 1 month ago
Feature engineering beats model choice, every single time. My XGBoost with handcrafted features beats my friend's transformer fed raw OHLCV by a mile. The model is the easy part; knowing what to feed it is the job.