Luck vs. Skill Decomposer

Overview

This pillar analyzes historical performance to determine how much of a result is due to sustainable skill versus unpredictable luck. It helps you avoid overvaluing lucky streaks and identify truly consistent performers.

What It Does

The pillar employs variance decomposition models to analyze a series of outcomes, like game scores, financial returns, or polling results. It isolates the component of performance that is stable and persistent (skill) from the component that is random and non-repeating (luck). This provides a clearer picture of an entity's true underlying ability to perform consistently.

Why It Matters

Markets often overreact to recent outcomes, mistaking lucky streaks for newfound skill. This pillar provides a statistical edge by quantifying the sustainability of performance, helping you position on true talent rather than fleeting randomness and avoid recency bias.

How It Works

First, the pillar ingests a time series of performance data for a specific entity. Next, it applies a statistical model to calculate the total variance in those outcomes. The model then attributes this variance to a persistent 'skill' factor and a random 'luck' factor, producing a clear ratio between the two.

Methodology

Utilizes a simplified ANOVA (Analysis of Variance) model where total performance variance Var(P) equals Var(Skill) + Var(Luck). The model estimates Var(Skill) by measuring the autocorrelation in performance over a rolling window, typically 12 to 24 periods. Var(Luck) is treated as the residual, unexplained variance. The core Noise-to-Signal ratio is calculated as Var(Luck) divided by Var(Skill).

Edge & Advantage

It cuts through narrative and recency bias by providing a quantitative measure of performance sustainability, allowing you to fade overrated 'hot streaks' and back underrated consistent performers.

Key Indicators

• Noise-to-Signal Ratio

  high

  The ratio of random variance (luck) to persistent variance (skill). A lower ratio indicates more reliable, skill-based performance.

• Variance Contribution Estimate

  high

  The percentage of total performance variance attributed to skill. A higher percentage suggests future results will be more predictable.

• Skill Persistence Coefficient

  medium

  A measure of how much this period's performance predicts the next. A value near 1 suggests high persistence, while a value near 0 suggests performance is random.

Data Sources

• Sports Reference Sites

  Provides game-by-game statistics for teams and players (e.g., Pro-Football-Reference).

• Financial Data APIs

  Offers historical return data for stocks, funds, and managers (e.g., Yahoo Finance, Alpha Vantage).

• Polling Aggregators

  Historical polling data for political candidates and parties (e.g., FiveThirtyEight, RealClearPolitics).

Example Questions This Pillar Answers

• → Will Team X win the championship next season after a surprise victory this year?
• → Will a fund manager who outperformed the market by 20% last year beat it again this year?
• → Is a political candidate's recent surge in the polls a durable trend or a temporary media blip?

Tags