① 6-Month Relative Strength Percentile (RS)

Measures how a stock’s 6-month price return ranks against all other US equities. A score of 0.99 means
the stock has outperformed 99% of the entire market. Stocks ranked below the 85th percentile are excluded
before any other factor is applied. Only persistent price leaders pass this first gate.

② Volatility Contraction Pattern (VCP)

Detects price consolidation structures where volatility is progressively contracting — a pattern popularized
by Mark Minervini and used by institutional desks to identify stocks coiling before a potential expansion.
Scored by measuring the sequential reduction in trading range over 4–8 week windows.

③ Volume Dry-Up Score

Quantifies whether sellers are stepping back from a stock. During healthy consolidations, volume contracts
as supply exhausts. A high Volume Dry-Up score indicates the stock is moving sideways or slightly lower
on decreasing volume — a supply vacuum that historically precedes demand-driven breakouts.

④ On-Balance Volume (OBV) Accumulation Slope

Tracks whether institutional money is flowing into or out of a stock by comparing up-volume days to
down-volume days over a 20-day rolling window. A rising OBV slope during price consolidation is one of
the strongest signals of quiet institutional accumulation — smart money buying before price reflects it.

⑤ Bear Market Shield (Residual Resilience Score)

Measures how much better (or worse) a stock performs relative to the S&P 500 during broad market
drawdowns exceeding 5%. This is computed as the residual from a rolling market-beta regression,
isolating the stock’s alpha component during stress periods. High-scoring stocks tend to hold or advance
while the market falls — a rare quality that defines true leadership.

⑥ Institutional Crowding Index

Estimates how saturated institutional positioning already is in a given stock, using short interest,
options market structure, and historical 13F filing trend data. Stocks with very high crowding scores
carry elevated reversal risk and receive a penalty in the composite. Early-stage institutional discovery
— low crowding with rising OBV — is the most attractive combination.

⑦ Gradient-Boosted ML Signal Layer

A gradient-boosted decision tree model trained on 3 years of daily OHLCV price history across the full
US equity universe. The model learns non-linear interactions between the 6 factors above and outputs a
probability-weighted signal. It is retrained quarterly using walk-forward validation to prevent
overfitting and is weighted at 15% of the final composite score.