Taunusstar democratizes quantitative market analysis by bringing the same machine learning techniques used by top hedge funds directly to individual investors — for free.
The platform runs three independent CatBoost gradient-boosting models, each trained on 100+ engineered features drawn from macroeconomic indicators (FRED), market data (Yahoo Finance), and valuation metrics. Every trading day, these models produce a probability-based LONG or CASH signal for the S&P 500 across three time horizons: 3 months, 6 months, and 12 months.
What makes Taunusstar unique is its multi-model consensus engine. Instead of relying on a single forecast, the platform aggregates signals from all three horizon models into a unified Consensus Strategy (invest only when 2+ models agree) and a Position Sizing Strategy (scale market exposure from 0% to 100% based on model agreement). This approach reduces false signals, manages downside risk, and has been stress-tested through every major market crisis in recent history.
Full transparency is a core principle. Users can explore complete out-of-sample backtest results — cumulative returns, drawdown analysis, monthly return heatmaps, calendar-year breakdowns, rolling Sharpe ratios, and crisis-period performance — all validated using strict walk-forward cross-validation with zero look-ahead bias. Models retrain quarterly to adapt to evolving market conditions without overfitting to short-term noise.
The result: an always-on, data-driven market signal that gives everyday investors the informational edge previously reserved for institutional quant desks.