StratGen Changelog

AI Copilot Conversational Studio: Built-in autonomous quantitative assistant in Settings (Settings -> AI Copilot). Formulate, backtest, walk-forward validate, and export algorithmic trading strategies in plain English without writing code.

Autonomous Tool Dispatching: AI Copilot autonomously compiles C# boolean rules, runs parallel backtest simulations, orchestrates multi-window Walk-Forward matrices, and formats performance metrics live.

Roslyn C# Live Rule Compilation: Validates strategy logic at runtime using the Microsoft Roslyn compiler, automatically self-correcting any syntax or type errors before simulations are launched.

Cloud BYOK (Bring Your Own Key) & Provider Hub: Support for OpenRouter (100+ models), Anthropic Claude (with Anthropic Workspace ID support), OpenAI ChatGPT, and Google Gemini Cloud, alongside the built-in StratGen Azores 1.0 quantitative engine.

Quick Prompt Chips: One-click automated workflows including Build Swing NQ Strategy, Intraday ES Strategy (with strict 15:30 session liquidation), and Energies Strategy.

Required Signals Column: Added a dedicated "Required" checkbox column in the Signals tab to mandate baseline indicators (regime filters, Elliott Wave, Choppiness Index) across all generated strategy combinations.

Configurable Rule Counts: Directly configure # Of Entry Rules and # Of Exit Rules in Strategy Settings to govern rule combination depth.

Intraday Session Close Guardrails: Dedicated Exit On Session Close toggle and exact Session Close Time (e.g. 15:30) prevent overnight margin risk for day-trading setups while allowing swing strategies to run unconstrained.

Dollar-Formatted Drawdown: Strategy Results and chart metrics now display maximum drawdown formatted in real currency dollars (e.g. -$29,942.50).

NinjaTrader 8 Parity: Enforced strict bar-close profit target and stop loss exits to achieve 100% trade execution parity between StratGen simulations and NinjaTrader 8.

Portfolio Persistence: Automated saving of winning candidates and rolling walk-forward matrices directly into the Portfolio/ folder as structured JSON artifacts.

Backtest Optimization Ceiling: Introduced configurable ceiling limits to prevent combinatorial runaway and guarantee responsive search cycles.

Walk-Forward Custom Signal Persistence: Preserved custom MiniSignals across batch walk-forward candidates and out-of-sample windows.

Single-Combo Debug Isolation: Isolated single-combo debug replays to prevent ledger bloat during multi-combo runs.

Donchian & KAMA Alignment: Resolved indicator lookback offsets and warmup gating to ensure exact match with NinjaTrader 8 indicators.

Backtest Engine Redesign: Re-designed the backtest engine to be much more efficient with data flows, backtesting entry logic, and exit logic.

Combined Strategy Fixes: Fixed numerous bugs regarding combined long and short strategies.

Portfolio Monte Carlo: Added Monte Carlo simulation for portfolio strategies.

New Signal Types: Added Donchian and Elliott Wave signals.

Reversal Behavior: Updated opposite signals so they can flip positions instead of waiting for a flat state.

AI Workflow Export: Removed an avoidable stall in the AI workflow export path.

Export File Names: Plain backtest portfolio exports now use normal timestamped names instead of _WF_Stitched_... names.

Signal Readability: Made Donchian and Elliott Wave signals easier to read.

Backtest Stability: Fixed a bug where backtests could hang.

Part 3 VM Wait: Added a safety timeout to the Part 3 VM wait.

Combined Portfolio Summaries: Recomputed missing summary and runtime fields from stitched equity curves instead of leaving stale values from the first source JSON.

Portfolio Adds: Fixed duplicate portfolio adds.

Buy and Hold Curves: Fixed buy and hold mark-to-market equity curves.

Highest and Lowest Params: Fixed a parameter miscount in Highest and Lowest.

Pyramiding Exits: Fixed NinjaTrader pyramiding exits so they scale out correctly.

Interface Fixes: Fixed icons, the overfit popup, and combined return/drawdown calculations.

Incubation Month Setting: Added incubation month as a configurable setting.

Z-Score Added: Added z-score support.

Template Updates: Updated templates to work with the new settings.

Pyramiding Guard: Disabled pyramiding when max entries is set to 1.

Threshold Consistency: Patched the threshold inconsistency issue.

Position Sizing Testing and Pyramiding: Added position sizing testing with multiple modes including fixed dollar risk, percentage equity, and fixed notional, along with pyramiding entry and exit types.

Portfolio Management and Analytics: Added a portfolio feature to save and load strategies, track incubation performance, rename strategies in the portfolio menu, and analyze portfolio correlations.

Monte Carlo Recording: Added Monte Carlo recording support.

Template Compatibility: Updated templates to work with the new settings.

AI Workflow Messaging: Added more info to the AI workflow message.

Strategy Name Normalization: Hardened strategy-name normalization at the source so invalid characters are cleaned before names are persisted or passed into walk-forward.

Combined Result Qualification: Downstream combined-result qualification now follows the selected goal metric and configured threshold instead of being hardcoded to RetDDGoal and Sharpe greater than or equal to 1.

Tick Calculation Fix: Fixed an incorrect tick calculation.

Walk-Forward Continuity: Fixed walk-forward so it is continuous and matches NinjaTrader more closely.

KAMA Code Export Fix: Fixed KAMA code export.

Backtest Loading Fix: Fixed a blank screen issue while loading backtests.

Walk-Forward Result Coverage: Fixed walk-forward missing some trade results and corrected mismatched data in some walk-forward workflows.

AI Workflows: Added AI workflows to automate your research.

AI Workflow Auto-Combine: AI workflows now automatically combine outputted strategy code.

Symbol Coverage Expansion: Added server-side symbols, more server-side symbols, and micro contracts in server-side variables.

Symbol Search: Added symbol search.

Workflow Defaults: Updated recommended default AI workflow settings and added better default WF settings.

Workflow Summaries: Added a better summary for AI workflow outputs.

AIWF Candidate Stability: AIWF now snapshots and restores combo config per candidate run.

Backtest Guardrails: Added a backtest ceiling limit and clamped WF median ret/dd close to 0.

Save Flow Reliability: Added debounce on save changes and one global cancellation token.

Backtest Accuracy Fixes: Fixed timezone issue, bar-shifting backtests, and order fill resolution.

Window Ordering Fix: Fixed window ordering.

Monte Carlo Stability: Fixed Monte Carlo VM leak/runaway worker and added more cleanup on Monte Carlo chart close.

Result Write Contention Fix: Resolved result file write contention (`UnauthorizedAccessException`).

Server Symbol Point Values: Fixed point values on server-side symbols.

History Data Caching: Added caching of historical data to significantly speed up walk-forward analysis and large-scale backtests.

Large Optimization Performance Fixes: Resolved performance bottlenecks affecting bigger optimization runs.

Signal Update Fix: Fixed issue where some signals were not updating correctly during OnData processing.

Expression Engine Upgrade: Replaced NCalc with Roslyn-based evaluation for improved execution speed and scalability.

Expanded Walk-Forward Code Export: Added TradeStation and Python walk-forward code generation support.

Reduced Logging Overhead: Further optimized backtest execution by reducing unnecessary logging.

Optimized Walk-Forward Performance: Improved walk-forward optimization execution speed and responsiveness for faster iteration on parameter sets.

Walk-Forward Finalizing Message Fix: Resolved issue where walk-forward runs could get stuck on a finalizing status message.

European Slippage Decimal Fix: Fixed slippage input parsing so comma decimals (e.g., 1,5) work correctly for European locales.

Walk-Forward Stop Handling Fix: Fixed walk-forward tests not ending properly if you click stop.

UI Polish: Updated the StratGen logo.

Walk-Forward Optimization: Implemented comprehensive walk-forward optimization feature for robust out-of-sample testing. This advanced technique divides your data into multiple in-sample and out-of-sample periods, optimizing parameters on in-sample data and testing on out-of-sample data to validate strategy robustness and reduce overfitting risk.

Walk-Forward Export & Visualization: Added ability to export walk-forward optimization results to CSV format and view comprehensive equity charts showing performance across all walk-forward periods, making it easier to analyze strategy consistency over time.

Walk-Forward Trades Display Fix: Resolved issue where walk-forward trades were not showing properly in the results view, ensuring complete visibility of all trades across walk-forward periods.

News Events Backtesting Fix: Fixed critical issue where news events (CPI, PPI, ISM, Jobs, Housing, Oil EIA Report, Natural Gas EIA Report) were not generating any trades during backtesting, ensuring proper fundamental signal integration.

Monte Carlo OOS Trade Selection Fix: Corrected issue where Monte Carlo simulations were not using the proper trades for out-of-sample (OOS) analysis, improving the accuracy of robustness testing and risk assessment.