Data Sources and Methodology

The Core Principle

Reliable decisions require reliable inputs.

That means: Fresh data, Verified sources, Transparent logic.

Not: "The model said so."

What Actually Matters

There are three things that determine whether your data is usable:

1) Quality

Is the data accurate? Is it coming from a trusted source? Is it consistent across books / feeds?

Bad examples: Mismatched odds, Missing player roles, Incorrect line assignments.

2) Freshness

How recent is the data? Has anything changed since it was pulled?

This is where most people get wrecked: Injury updates, Line movement, Starting lineup changes.

A "good" bet 2 hours ago can be a terrible bet now.

3) Context

Raw data without context is dangerous.

Example: A player averages 4 shots per game… but his minutes just dropped, his role changed, his opponent suppresses volume.

Same number → completely different meaning.

Check Freshness Before You Bet

Before locking anything in, ask: Are lineups confirmed? Are goalies/starters confirmed? Has the line already moved?

Your system already enforces this: Unconfirmed inputs → downgrade or PASS. Missing key data → unsafe for plays.

If you skip this step, you're betting outdated information.

Avoid Black-Box Thinking

If you can't explain why a play exists or what variables are driving it, you shouldn't trust it.

Good models: Show drivers, Show assumptions, Show uncertainty.

Bad models: "Trust me, it's a play."

Prefer Explainable Signals

Strong signals: Role (minutes, usage, TOI), Matchup dynamics, Pace / environment, Price vs projection.

Not: Vague "trends", Cherry-picked stats, Narrative-based reasoning.

Document Blind Spots

Every model has weaknesses. Sharp approach: Call them out, Adjust for them, Sometimes PASS because of them.

Examples: Unknown starting goalie, Player minutes uncertainty, Missing market data.