Esports Prop Research

LoL and VALORANT player prop research

The live board brings together player stat lines from supported providers. Choose a game and a match, then open a prop to review its line, stat type, and map span alongside the available research. Provider availability changes, so an empty board does not mean there are no upcoming matches.

A Map 1 line and a Maps 1–2 line describe different totals. Research starts by checking that the historical results are comparable to the selected prop, rather than treating every player result as interchangeable.

What to look for in the evidence

Comparable historical results

Check the stat, map span, evidence window, and sample size. A historical hit rate describes the observed sample; it does not establish the chance of the next prop hitting.

Model Projections

Estimates of a player stat based on recent evidence and context. Compare the projection with the provider line, then inspect its assumptions and the evidence supporting it.

Source freshness and coverage

Missing results, small samples, stale feeds, and uncertain player or match linkage limit what the research can tell you. Unavailable evidence stays visible rather than becoming a confident score.

What Research Confidence means

Research Confidence is an uncalibrated estimate of evidence strength and directional support for a Prop Leg. The Primary Research Score is the explicit headline score. Other algorithm scores provide comparison context.

A Research Confidence score of 80 does not mean an 80% chance of winning. High Confidence helps you find research to inspect; it is not a wager instruction or a guarantee. Experimental views and their projection gaps have their own meanings and limitations.

Team form and matchup context

Team Comparison describes historical form and differences between teams. Matchup Context adds supporting evidence to player research. Neither is a stand-alone match prediction, and supporting context does not replace the Primary Research Score.

Research keeps its original evidence

A saved Evaluation Snapshot preserves the score, evidence window, sample size, and algorithm version known when it was saved. Later imports or model changes do not rewrite that research. Recording a result updates its settlement fields, not its captured evidence.

Automatic Prediction Records have a separate lifecycle for model learning. The Prediction Ledger is an administrator workspace; public research does not require an account.

A tool for research and judgment

This is a personal research project, not a sportsbook or a betting bot. Historical hit rates, backtests, Model Projections, and Research Confidence are evidence to examine. They are not calibrated win probabilities, profit promises, or replacements for your judgment.