Reference Class Robustness Scorer

Overview

This pillar assesses the statistical foundation of a market by analyzing the availability of comparable historical data. It helps you identify and avoid markets where a lack of precedent makes forecasting more of a gamble than a calculated risk.

What It Does

The scorer systematically searches for a 'reference class', which is a set of similar past events relevant to the current market. It evaluates the size and quality of this historical dataset to determine if there is enough information to form a reliable probability. This process flags markets built on thin evidence or unprecedented circumstances.

Why It Matters

Its primary value is in risk management. By quantifying the robustness of the available data, it prevents you from over-allocating capital to markets that are statistically unpredictable. This pillar provides the crucial context of 'do we know enough to even make a prediction?'.

How It Works

First, the pillar deconstructs a market question into its core components. It then queries historical databases for analogous events, forming a potential reference class. Next, it scores the relevance of each analog and checks for common biases, like survivorship bias. Finally, it calculates a single Robustness Score indicating the data's suitability for forecasting.

Methodology

The Robustness Score is calculated using a weighted formula: Score = (log(N) * Relevance_Score) - Bias_Penalty. 'N' is the number of valid historical analogs. 'Relevance_Score' (0-1) is a proprietary measure of similarity between past and current events. 'Bias_Penalty' is a deduction applied if significant data biases, such as survivorship or selection bias, are detected.

Edge & Advantage

This pillar provides a disciplined, data-driven check against emotional or narrative-based trading. It forces you to consider the statistical ground truth before entering a position on a novel or speculative event.

Key Indicators

• Sample Size (N) Sufficiency

  high

  Measures the number of comparable historical events found. A low 'N' indicates insufficient data.

• Historical Analog Relevance

  high

  Scores how closely the past events match the conditions of the current market.

• Survivorship Bias Check

  medium

  Detects if the historical data is skewed by only including successful examples, ignoring failures.

Data Sources

• Historical Event Databases

  Aggregated data from news archives, academic papers, and encyclopedic sources about past events.

• Internal Market Archives

  Data from previously resolved prediction markets on the platform.

Example Questions This Pillar Answers

• → Will a US-based quantum computing company achieve a market cap over $1 trillion by 2035?
• → Will a third-party candidate win a US presidential election before 2040?
• → Will a movie based on a video game IP win the Oscar for Best Picture this decade?

Tags