About Signal Alchemist.
Our Mission
Signal Alchemist is an educational trading analysis platform that uses quantitative models, machine learning, and technical analysis to help traders identify potential opportunities in the equity and options markets.
We believe that sophisticated analytical tools should be accessible to all traders, not just institutions with million-dollar research budgets. Our platform distills complex multi-factor analysis into clear, actionable signals with transparent scoring.
What We Do
Our platform scans 3,500+ US equities using 390+ technical indicators, fundamental scoring models, and machine learning ensembles. Every signal is scored across multiple analytical dimensions and validated through walk-forward backtesting to prevent overfitting.
Multi-Factor Ensemble
17 independent analysis engines score every ticker across technical, momentum, mean reversion, volatility, risk, sentiment, factor, anomaly detection, microstructure, intermarket correlation, macro regime, statistical arbitrage, network effects, and quantitative signals. A meta-scorer weighs the ensemble output. No single model failure can compromise a signal.
Regime-Adaptive Weighting
Hidden Markov Models classify the market into four regimes: risk-on trending, risk-off volatile, mean-reverting choppy, and low-volatility quiet. Engine weights shift automatically. A momentum-heavy weight set in a trending market becomes a mean-reversion-heavy set in a choppy one. Regime-stratified optimization with train/test validation prevents overfitting to any single market character.
Self-Improving Learning Loop
Models retrain daily using XGBoost, LightGBM, and a Transformer sequence model for multi-timeframe prediction. Optuna hyperparameter optimization searches thousands of configurations. Feature engineering pulls from macro indicators, cross-asset correlations, earnings signals, and microstructure data. Anti-pattern detection flags drawdown conditions before they compound.
Walk-Forward Backtesting
Every model change is validated on unseen data using multi-window walk-forward evaluation. Transaction costs are included. The ticker universe uses historical S&P 500 membership data back to 2015 to eliminate survivorship bias. Monte Carlo bootstrap simulations generate confidence intervals and Kelly criterion position sizing.
Options Intelligence
Algorithmically generated options strategies with live Greeks, probability of profit, interactive P&L diagrams, and defined risk parameters. Covered calls, vertical spreads, iron condors, straddles, and more. Each strategy is scored against the same conviction framework as equities.
Autonomous Quality Assurance
Six QA agents run on schedule: a data quality sentinel with 8 market-hours checks, a calculation verifier running 130+ unit tests daily, an API health monitor, a learning loop health assessor, a frontend-backend parity checker using static analysis, and a performance regression detector with weekly trend analysis. Issues surface before users ever see them.
By the Numbers
What We Are Not
Our platform is an educational tool that provides algorithmically generated analysis. All signals, scores, and projections should be used as one input among many in your own research process. We operate under the publisher's exclusion from the definition of "investment adviser" as provided under Section 202(a)(11)(D) of the Investment Advisers Act of 1940.
Our Approach to Transparency
- Open scoring - Every signal shows its component scores across all nine analytical dimensions, so you can see exactly why a signal was generated.
- Walk-forward validation - Models are tested on unseen data only. No curve-fitting, no data leaking, no cherry-picked results.
- Honest disclaimers - We clearly label all simulated data, synthetic estimates, and hypothetical performance. We don't hide behind fine print.
- No hype - We don't promise guaranteed returns, risk-free trading, or overnight riches. Trading involves substantial risk, and we say so plainly.
Contact Us
For general inquiries: schuyler@trytaest.com
For legal or privacy questions: schuyler@trytaest.com