Historical Similarity Scorer
Predicting the future by decoding the past.
Overview
This pillar identifies and scores historical events that are structurally similar to the current market. By analyzing how past situations resolved, it provides a data-driven map of potential future outcomes.
What It Does
The Historical Similarity Scorer uses a multi-factor algorithm to compare a current event against a vast database of historical precedents. It weighs key drivers, contextual variables, and actor behaviors to find the closest historical matches. The pillar then aggregates the outcomes of these past events to forecast the most likely result for the current situation.
Why It Matters
History doesn't repeat, but it often rhymes. This pillar quantifies that rhyme, offering a data-driven alternative to gut feelings or simple trend-following. It uncovers hidden patterns and base rates that are often missed in the heat of the moment.
How It Works
First, we define the key parameters of the current market, like its actors, stakes, and timeline. The algorithm then scans our historical database, scoring each past event on dozens of similarity vectors. The top matches are selected, and their outcomes are weighted by their similarity score to create a probability distribution for the current market's resolution.
Methodology
Utilizes a k-nearest neighbors (k-NN) approach combined with cosine similarity on vectorized event features. Event features include macroeconomic data, political polling, key actor sentiment, and event structure tags. Similarity scores are calculated as a weighted average of feature-level similarities, with the final output being a probability distribution derived from the outcomes of the top 'k' most similar historical precedents.
Edge & Advantage
It provides a robust, quantitative anchor against recency bias, allowing you to position against popular narratives by using historical base rates.
Key Indicators
-
Similarity Match %
highThe percentage score representing how closely a past event matches the current one.
-
Precedent Outcome Map
highA visualization of the outcomes from the top historical matches, weighted by similarity.
-
Key Driver Correlation
mediumMeasures the alignment of the most influential factors between the current and historical events.
Data Sources
-
Armed Conflict Location & Event Data Project for political and conflict data.
-
Global Database of Events, Language, and Tone for monitoring world news and events.
-
Historical macroeconomic indicators for financial markets.
Example Questions This Pillar Answers
- → Will the current US-China trade negotiations result in a new tariff agreement by year-end?
- → Will the incumbent party win the upcoming national election?
- → Will a specific international peace summit lead to a lasting ceasefire?
Tags
Use Historical Similarity Scorer on a real market
Run this analytical framework on any Polymarket or Kalshi event contract.
Try PillarLab