Methodology

Learn how Custom Venom's toxin system works, how we measure accuracy, and how to interpret the data.

What Are Toxins?

Toxins are predictive signals that Custom Venom uses to identify opportunities and risks in player projections. Each toxin represents a specific pattern or condition that has historically correlated with player performance outcomes.

When a toxin is "assigned" to a player, we call this a "fire". The system tracks whether each fire correctly predicted the outcome, allowing us to calculate accuracy rates and continuously improve our models.

Toxin Tiers

Toxins are organized into tiers based on their impact and reliability:

CRITICAL

Highest-impact signals that indicate significant risk or opportunity. These toxins have strong historical accuracy and can dramatically affect projections.

Confidence Low

PRIMARY

Core predictive signals with proven track records. These form the backbone of our projection adjustments.

Terrible MatchupElite Matchup

SECONDARY

Supporting signals that add context and nuance to projections. Valuable when combined with primary toxins.

Weather ImpactUsage TrendingSell HighMatchup Smash

CONTEXTUAL

Situational signals that apply in specific game contexts. Important for DFS lineup decisions.

Game EnvironmentCatch-Up Script

DFS Indicators

Signals specifically designed for daily fantasy optimization. Focus on volume, role, and matchup advantages.

BellcowYAC MonsterDeep ThreatRed Zone Vulture

How Accuracy Is Calculated

Basic Formula

Accuracy = (Correct Predictions ÷ Total Evaluations) × 100%

A prediction is "correct" when the toxin's implied direction matches the actual outcome. For example, a "boom" toxin firing on a player who outperforms their projection is correct.

Wilson Score Confidence Interval

We use the Wilson score interval to calculate confidence bounds on accuracy rates. This method is preferred over simple percentage intervals because it:

  • Provides accurate intervals even with small sample sizes
  • Never produces impossible values (below 0% or above 100%)
  • Accounts for the inherent uncertainty in statistical estimation

Toxins with fewer than 30 evaluations have wider confidence intervals. Use these accuracy rates with appropriate caution.

Interpreting the Data

High Accuracy (≥65%)

The toxin has a strong historical track record. Consider it a reliable signal, especially when sample size is large.

Medium Accuracy (50-65%)

Better than random chance. Valuable when combined with other signals or in specific contexts.

Low Accuracy (<50%)

Historical performance below coin-flip. May indicate the toxin needs recalibration or is context-dependent.

Insufficient Data

Fewer than 30 evaluations means the accuracy rate is preliminary. Weight accordingly.

Position-Specific Performance

Not all toxins perform equally across positions. A toxin that works well for wide receivers might be less predictive for tight ends due to differences in usage patterns, matchup dynamics, and scoring volatility.

Each toxin's detail page shows its "Best" and "Worst" positions based on historical accuracy. Use this information to weight signals appropriately when building lineups.

General Position Patterns

QB

More predictable, lower variance

RB

Volume-dependent, game script sensitive

WR

Matchup-driven, high variance

TE

Usage varies widely, boom/bust

K

Game script and red zone dependent

DST

Opponent-driven, highly volatile

Our Commitment to Transparency

The Toxicology Lab exists because we believe you deserve to see how our predictions perform. We publish all accuracy data—good and bad—so you can make informed decisions about which signals to trust.

Toxins that underperform are continuously evaluated for recalibration or retirement. This isn't just about transparency—it's how we build a system you can actually rely on.