Signal Interpretation Guide
Understand every metric, score, and indicator on your Signal Alchemist dashboard. This guide explains how signals are generated, scored, and how to use them in your trading decisions.
🎯 Composite Score
The composite score is the single most important number on each signal card. It represents the overall strength of a trading signal on a 0-100 scale, synthesized from multiple independent analysis engines.
Scores are calculated by blending outputs from 17 independent analysis engines, each specializing in a different dimension of market behavior. Each engine contributes a sub-score which is weighted by the current market regime and the engine's historical accuracy for the given ticker.
Score Ranges
🏅 Tier System
Every signal is classified into one of three tiers based on its composite score and overall quality. Tiers determine visual priority on the dashboard and align with your subscription's signal limit.
Highest-conviction signals. Strong multi-engine agreement with favorable risk/reward. These are the signals institutional desks would prioritize.
Solid signals with good conviction. Most engines agree but there may be one dissenting factor. Good for portfolio diversification.
Moderate signals with mixed conviction. Consider smaller positions and closer stop losses. Best suited for swing trades.
Subscription Access
📊 Confidence Score
The confidence percentage (shown as "Conf" in the signal footer) measures how reliable this specific signal pattern has been historically. It is distinct from the composite score.
Confidence is derived from backtested performance of the specific strategy type (e.g., "Momentum Breakout" or "Mean Reversion Oversold") under the current market regime. A high confidence with a low composite score means the strategy is proven but the current setup is weak. A high composite with low confidence means the setup looks good but the strategy has limited track record.
⚖️ Risk/Reward Ratio
The R:R ratio (shown as "R:R" in the signal footer) measures how much potential profit you stand to gain relative to how much you could lose if the stop loss is hit.
A R:R of 3.0x means for every $1 you risk (distance from entry to stop loss), you can potentially gain $3 (distance from entry to target). Professional traders typically require a minimum R:R of 2.0x before entering a trade.
↕️ Signal Direction
Each signal specifies a directional bias: Long or Short. This tells you which side of the trade the analysis engines favor.
The analysis expects the stock price to rise. Buy shares or call options. Entry, stop loss, and targets are set for an upward move.
The analysis expects the stock price to fall. Consider short selling or put options. Stop loss is above the entry, targets are below.
📏 Price Levels
Every signal includes three critical price levels that define your trade plan: where to enter, where to cut losses, and where to take profits.
Entry Range
The recommended price range for entering the position. Signals provide a low-high range rather than a single price, giving you flexibility to enter on dips within the zone. Avoid chasing entries above the range.
Stop Loss
The price at which you should exit to limit losses. This level is calculated based on volatility (ATR), support/resistance levels, and the strategy's risk tolerance. A hard stop loss at this level protects your capital from adverse moves.
Target
The primary profit target. This is calculated from resistance levels, measured moves, and risk/reward optimization. Some signals include a second target for those who want to scale out of positions in stages.
🔄 Signal Persistence
Persistence badges track how long a signal has remained active across consecutive scans. Signals that persist across multiple scans demonstrate sustained conviction - the analysis engines continue to see opportunity even as conditions evolve.
The score delta next to the persistence badge shows how the composite score has changed since the signal first appeared. A positive delta (green) means conviction is strengthening; negative (amber) means the signal may be weakening.
🌡️ Market Regime
The market regime widget in the dashboard sidebar shows the current macro environment. Signal Alchemist adjusts engine weights based on the detected regime to optimize signal quality for current conditions.
Momentum and trend-following engines are weighted higher. Expect more LONG signals with wider targets.
Volatility and risk management engines dominate. Signals favor defensive plays, tighter stops, and smaller positions.
Mean reversion engines are prioritized. Signals target oversold bounces and overbought reversals.
Balanced engine weights with emphasis on breakout detection. Position sizes may be larger due to lower risk.
💰 Position Sizing
When you set your portfolio size and risk-per-trade percentage in the dashboard sidebar, Signal Alchemist calculates personalized position sizes for every signal.
Position sizing follows a risk-based approach: the number of shares is calculated so that if the stop loss is hit, your loss equals your risk-per-trade amount. This ensures consistent risk across all positions regardless of stock price or volatility.
🧪 Analysis Engines
Signal Alchemist runs 17 independent analysis engines on every ticker across 3,500+ US equities. Each engine specializes in a different dimension of market analysis. Their outputs are combined into the composite score using regime-adaptive weights.
RSI, MACD, Bollinger Bands, Keltner Channels, ATR, ADX, stochastic, Williams %R, MFI, Donchian, and 40+ more indicators across multiple timeframes.
Trend strength, directional movement, price acceleration, and relative performance vs sector and market.
Oversold/overbought extremes using z-scores, Bollinger %B, distance from moving averages, and reversion probability.
Implied vs realized volatility, volatility regime, contraction/expansion cycles, and breakout probability.
Drawdown analysis, Value at Risk, tail risk metrics, correlation exposure, and portfolio-level risk contribution.
Social media sentiment scoring, macro sentiment (yield curve, VIX, risk-on/risk-off), and earnings sentiment.
Value, growth, quality, and profitability factor exposure using Piotroski F-Score, Altman Z-Score, DuPont decomposition.
Statistical outlier identification across price, volume, and fundamental data using isolation forests and z-score methods.
Volume profile analysis, VWAP behavior, bid-ask dynamics, and institutional flow signatures.
Cross-asset correlations between equities, bonds, commodities, currencies, and sector ETFs.
Yield curve regime, Fed policy indicators, recession probability, sector rotation signals.
XGBoost, LightGBM, and Random Forest models trained on engineered features with daily retraining.
Supply chain relationships, sector contagion, and correlation clustering.
Pairs trading signals, cointegration tests, and spread analysis.
Multi-source data aggregation with fallback providers (yfinance, Alpha Vantage, Polygon.io).
Cross-engine signal correlation, redundancy detection, and conviction calibration.
Quantitative strategy signals including momentum factor, value factor, and quality screens.
Engine weights adapt dynamically based on the current market regime. In trending markets, momentum engines receive higher weight. In choppy markets, mean reversion engines are prioritized. This adaptive weighting improves signal quality across all market conditions.
🔬 The Learning Loop
Signal Alchemist does not run static rules. The system retrains daily after market close, so every signal you see reflects the most current market behavior.
All ML models retrain using the latest market data after each trading session closes. Yesterday's price action directly informs today's signals.
Every model change is validated on data the model has never seen. This prevents overfitting and ensures signals generalize to live markets.
Optuna searches thousands of parameter configurations for each model, selecting the combination with the best out-of-sample performance.
Engine weights are optimized separately for each market regime (trending, volatile, choppy, quiet). The system adjusts how much each engine matters based on current conditions.
Confidence intervals on every signal are generated through simulation. You see the probability distribution, not just a point estimate.
The system flags conditions historically associated with drawdowns. If a signal looks good on paper but matches a pattern that has burned traders before, the score is adjusted downward.
🛠️ Free Tools
These tools are available to everyone, no paid subscription required.
DCF Calculator
Discounted cash flow valuation for any US equity. Adjust the growth rate, discount rate, and terminal multiple to model different scenarios. Shows intrinsic value estimate vs current price.
Stock Comparison
Side-by-side comparison of two stocks across all 17 scoring dimensions. Interactive radar charts show where each stock is strong and where it is weak. Useful for deciding between two positions.