Pietro Giacobazzi
bb670836e1
style: normalize indentation in Orchestrator.mqh via clang-format
...
Whitespace-only: consistent 3-space indentation throughout (was a mix of
3/4/5/6.. spaces). Repo conventions preserved: for(/if( spacing, access
labels, attached braces. Adds .clang-format so the style is reproducible
(run: clang-format -i --style=file <file>).
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-06-14 18:21:44 +00:00
Pietro Giacobazzi
a8b4a7d3ff
Fix sym shadowing warnings + coherent softmax gating
...
- Rename Init() param sym -> symName in IAgent base and overrides
(MAAgent, MomentumAgent, RegimeADX) to stop hiding the global sym.
- Softmax gating: derive temperature from dispersion of the competence
weights ws[] (the same quantity the softmax weights), not from the
variance of signals zs[]; skip disabled agents in the normalization.
- Use |rho| (with weightMin floor) as competence and sign-flip
anti-correlated agents instead of zeroing them, so rho<0 information
is used (inverted) rather than discarded.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-06-14 18:20:19 +00:00
pietro_giacobazzi
3256ebb5c6
Fix InitNeuralNet: non forza m_useNeural=true quando solo TrainMode
2026-06-14 11:47:30 +02:00
pietro_giacobazzi
30e116a2d5
Remove duplicate LogBarHistory in OnTick (gia chiamata in Analyze)
2026-06-14 11:42:27 +02:00
pietro_giacobazzi
281716982a
Fix: TimeHour/Minute/Seconds MQL4 -> MqlDateTime+TimeToStruct MQL5
2026-06-14 11:21:13 +02:00
pietro_giacobazzi
19855af849
Commenti chiari su tutti gli input parameters
2026-06-14 11:18:28 +02:00
pietro_giacobazzi
713460357e
Add m_runTimestamp to all export filenames (YYMMDD_HHMMSS) — ogni backtest genera file unici
2026-06-14 11:15:15 +02:00
pietro_giacobazzi
907e100379
Fix: entryTime perso in completedTrades (era 0), LogBarHistory mai chiamata in OnTick, data.Time(0) -> data.time[0], return lotMin -> 0 in CalcRiskLot, entryTime esportato per open trades in CSV
2026-06-14 10:24:08 +02:00
pietro_giacobazzi
e7d38ec937
Risk-based position sizing: replace Inp_Lot+Inp_RiskFraction with Inp_RiskPerTrade+Inp_RiskTotal, proper CalcRiskLot() using tickValue/tickSize, total risk cap, track lot/riskAmount per trade
2026-06-14 09:55:51 +02:00
pietro_giacobazzi
e2ca6336af
Add CSV exports: BarHistory, TradeHistory, AgentInteraction, DecisionLog, plus .gitignore
2026-06-14 09:48:43 +02:00
pietro_giacobazzi
dcf01418ea
Add TP validation via Inp_TPRiskATR, fix saveInterval in OnTrade, add SL/TP order validation
2026-06-14 09:38:54 +02:00
pietro_giacobazzi
6296cc36b2
Fix remaining Orchestrator references: replace 't' with openTrades[idx] in OnTradeClose loop and active flag
2026-06-13 15:41:06 +02:00
pietro_giacobazzi
a29befcd67
Fix MQL5 compile: rename 'input' params in NeuralNet (keyword conflict), replace Orchestrator pointers with direct array access (MQL5 doesn't support struct pointers)
2026-06-13 15:18:49 +02:00
pietro_giacobazzi
c70c2f28ce
Fix compile errors: NeuralNet input param rename, Orchestrator refs to pointers, EVT_MaxAbsZ call, SYMBOL_MARGIN_INITIAL, remove unused var
2026-06-13 15:07:05 +02:00
pietro_giacobazzi
67c8485373
Default Inp_Lot=0.0 for adaptive sizing
2026-06-13 14:07:56 +02:00
pietro_giacobazzi
5c00f17121
MultiAgentTest v7: zero magic constants, neural orchestrator, multi-position, agent interaction fixes
2026-06-13 13:58:38 +02:00