7H Conviction Score

7 Horns' confidence rating for a stock: a blended 0-to-1 score built from five weighted pillars (fundamentals, technicals, the CNN model, sentiment, and valuation) that drives how much weight a name gets in a portfolio.

The 7H Conviction Score is a proprietary quantitative model used to evaluate and rank individual stocks. It synthesizes five distinct analytical pillars into a single, normalized score ranging from 0 to 1. These pillars are weighted according to their historical predictive power and strategic importance: Fundamentals (30%), Technicals (25%), the CNN Model (20%), Sentiment (15%), and Valuation (10%). Each pillar is individually scaled to a value between 0 and 1 before being combined. To ensure the score remains robust even when information is incomplete, the model dynamically adjusts for missing data. If a specific pillar lacks reliable data, it is excluded from the calculation entirely, and the weights of the remaining pillars are proportionally rescaled to sum to 100%. Additionally, a freshness multiplier is applied to penalize stale data, scaling the final score by a factor ranging from 1.0 for real-time data down to 0.75 for older inputs. The final score is then sharpened mathematically to widen the spread between average and high-conviction ideas. Investors use this score primarily to drive portfolio construction and position sizing. Rather than allocating equal capital to all holdings, a portfolio manager can use the conviction scores to weight assets proportionally, ensuring that the highest-scoring stocks receive the largest capital allocations. A high score (closer to 1.0) indicates strong alignment across multiple analytical disciplines, while a low score suggests weak fundamentals, poor technical momentum, or overvaluation, signaling that a stock should be avoided or underweighted.

Suppose a stock has the following pillar scores: Fundamentals = 0.80, Technicals = 0.70, CNN Model = 0.90, Sentiment = 0.60, and Valuation = 0.50. Assuming all data is fresh (multiplier of 1.0), the weighted score is calculated as: (0.80 0.30) + (0.70 0.25) + (0.90 0.20) + (0.60 0.15) + (0.50 0.10) = 0.24 + 0.175 + 0.18 + 0.09 + 0.05 = 0.735. If the Valuation data becomes stale, dropping the multiplier to 0.80 for that pillar, the overall score is adjusted downward to reflect the lower confidence in the inputs.