How APEX Scores Every Stock
APEX runs 12 signals across three weighted buckets — Technical (40%), Fundamental (30%), Market Intelligence (30%) — and sums the weighted scores to produce a 0–100 composite. BUY = 75+. HOLD = 35–75. SELL = below 35. No single signal triggers a verdict; confluence across all three buckets is the signal. Data sources: Yahoo Finance, Finnhub, Anthropic Claude API, CBOE.
TECHNICAL BUCKET
40% of scorePrice momentum, trend direction, volatility, and volume — the primary signals for near-term trade timing.
FUNDAMENTAL BUCKET
30% of scoreBusiness quality, valuation, and analyst consensus — validates that technical setups have fundamental backing.
MARKET INTELLIGENCE BUCKET
30% of scoreMacro context, institutional positioning, and market structure — the signals retail traders typically don't have access to.
Scoring Formula
Each signal is normalized to 0–100 on its own scale, then weighted within its bucket. For example: RSI of 45 normalizes to 45/100. Within the Technical bucket (where RSI carries an 18% sub-weight — the highest of any technical signal), this contributes 45 × 0.18 = 8.1 points to the Technical bucket score. The Technical bucket score is then multiplied by 0.40 for its contribution to the composite. Sub-weights were tuned using a historical backtest across 102 stocks and 86,000+ ticker-days.
Score Ranges and Verdicts
Backtest Results
Signal sub-weights were tuned using a real historical dataset: 86,488 ticker-days across 102 US large-cap stocks, spanning Jan 2023–Jun 2026. A train/test split (train: 2023–2024, test: 2025–2026) ensured no look-ahead bias. For each of 29 weight variants, composite scores were correlated with 5-, 10-, and 20-day forward returns. The sensitivity sweep identified 8 signals worth reweighting: RSI (+), Stochastic (+), Bollinger (+), MA Cross (+), MACD (−), Fibonacci (−), ATR (−), VIX (−). The tuned set improved test-period Pearson correlation from −0.023 to +0.003 — a modest but directionally correct result consistent with the inherently noisy short-term equity signal environment.
Trade simulation: 100 entries across US large-cap stocks, 2022–2024 (bear market, recovery, AI bull market).
Data Sources
Frequently Asked Questions
How does APEX calculate a stock score?
APEX scores each of 12 signals individually (0–100), groups them into three weighted buckets — Technical (40%), Fundamental (30%), Market Intelligence (30%) — and sums the weighted scores to produce a 0–100 composite. Within the Technical bucket, RSI carries the highest sub-weight (~18%), followed by Stochastic (~13%). Signal weights are backtest-tuned using 86,000+ historical ticker-days across 102 US large-cap stocks. The final score drives a BUY (75+), HOLD (35–75), or SELL (below 35) verdict.
What does an APEX score of 75 mean?
A score of 75–100 is a BUY signal — the majority of APEX's 12 signals are aligned bullishly across Technical, Fundamental, and Market Intelligence dimensions simultaneously. A score of 35–75 is HOLD — signals are mixed with no clear directional edge. A score below 35 is SELL — the majority of signals are aligned bearishly.
What data sources does APEX use?
APEX uses four primary data sources: (1) Yahoo Finance — real-time price data, OHLCV, fundamental financials (P/E, revenue, margins), and earnings history; (2) Finnhub — market data fallback, analyst ratings, and insider trading (SEC Form 4) data; (3) Anthropic Claude API — AI narrative synthesis, supply chain mapping, exit plan generation, and macro regime interpretation; (4) CBOE/Federal Reserve — VIX data, treasury yield curve data.
How was the APEX scoring system backtested?
APEX runs two types of backtesting. (1) Trade simulation: 100 simulated trades per tier across US large-cap stocks, 2022–2024. Results — Free (3 signals): 54% win rate. Pro (12 signals): 73% win rate. Elite (+ institutional signals): 86% win rate. (2) Historical weight optimization: 86,488 ticker-days across 102 stocks (Jan 2023–Jun 2026), split into train (2023–2024) and held-out test (2025–2026) periods. Signal weights for RSI, Stochastic, Bollinger, MA Cross, MACD, Fibonacci, ATR, and VIX were tuned based on correlation with forward returns. Past simulated performance does not guarantee future results.
Why does APEX use three buckets instead of a flat signal list?
The three-bucket structure ensures that no single dimension of analysis dominates the score. A stock could have strong Technical signals but deteriorating fundamentals — in a flat model, the technical signals would override the fundamental warning. The bucket structure caps the Technical contribution at 40% of the score, ensuring Fundamental and Market Intelligence inputs always have meaningful weight. This reduces false positives during earnings seasons and macro regime shifts.