Back to Selecting the Right Distribution
Chapter 10Calculator

Sport-Specific Guide

Step-by-step guide to choosing the right model

Sport-Specific Distribution Guide

Now that you understand the theory behind Normal, Poisson, Negative Binomial, ZIP, and Hurdle distributions, let's apply them to real prop types across major sports. This lesson provides a practical reference you can use when analyzing any prop.

The Master Decision Framework

Before diving into sport-specific guidance, here's the universal workflow:

1. Is it continuous or count-based?
   ├── High-volume continuous (50+ yards, 15+ points) → Normal
   └── Low-volume counts (0, 1, 2, 3 outcomes) → Continue...

2. Calculate VMR
   ├── VMR ≤ 1.3 → Poisson candidate
   └── VMR > 1.3 → Negative Binomial candidate

3. Check for excess zeros
   ├── Observed zeros match prediction → Use distribution from Step 2
   └── Observed zeros >> prediction → Continue...

4. Identify zero source
   ├── Structural (no opportunity) → ZIP
   └── Any vs. none (player active) → Hurdle

NFL Distribution Guide

High-Volume Props → Normal

Prop TypeDistributionRationale
Passing yardsNormalHigh volume, roughly symmetric
Rushing yards (bell cows)NormalConsistent touches, ~100+ yard range
Receiving yards (WR1/WR2)Normal50+ yard expected outcomes

Normal P(Over)

P(Over) = 1 - Φ((line - μ) / σ)
Excel: =1-NORM.DIST(line,mean,stdev,TRUE)

Low-Volume Count Props → Poisson or Negative Binomial

Prop TypeDistributionVMR Check
Passing touchdownsPoisson or NBCheck VMR; most QBs ≈ 1.0
Rushing touchdowns (starter)PoissonUsually consistent, VMR ≈ 1
Receptions (high-volume WR)Poisson5+ targets expected, stable

Zero-Heavy Props → ZIP or Hurdle

Prop TypeDistributionRationale
Backup RB receptionsZIPSome games blocker-only (structural zero)
Pocket QB rushing yardsZIPGame plan determines scramble opportunity
TE targets (TE2)HurdlePlaying but may not be targeted
Rushing TDs (backup RB)HurdleActive but goal-line work varies

Key Insight

NFL Rule of Thumb: If the player's role varies game-to-game (sometimes involved, sometimes scripted out), lean toward ZIP. If the player is always active but the stat is just rare, lean toward Hurdle.


NBA Distribution Guide

High-Volume Props → Normal

Prop TypeDistributionRationale
Points (starters)Normal15+ expected, symmetric distribution
Rebounds (bigs)Normal8+ expected, high volume
Assists (point guards)Normal6+ expected, consistent opportunity
PRA combosNormalHigh combined volume

Low-Volume Count Props → Poisson or Negative Binomial

Prop TypeDistributionVMR Check
3-pointers made (shooter)PoissonConsistent attempts, VMR ≈ 1
StealsPoissonRare but stable rate
TurnoversPoissonUsually consistent

Zero-Heavy Props → ZIP or Hurdle

Prop TypeDistributionRationale
Bench player pointsHurdleEither gets minutes or doesn't
Blocks (guards)HurdleZero common, but playing full minutes
3PM (role player in platooned matchup)ZIPMay be benched entirely vs certain lineups
Assists (non-PG)HurdlePlaying but assist opportunities rare

Tip

NBA Rule of Thumb: NBA props are often higher-volume than NFL, so Normal is your default for star player stats. Reserve ZIP/Hurdle for bench players or rare events like blocks.


MLB Distribution Guide

Continuous/High-Volume Props → Normal (with caution)

Prop TypeDistributionNotes
Total basesNormalWith large enough sample
Hits (high BA hitter)NormalIf 2+ expected

Warning

MLB props often have lower expected values than other sports. A hitter expecting 1.2 hits per game is better modeled with Poisson than Normal, despite hits being "common."

Count Props → Poisson or Negative Binomial

Prop TypeDistributionVMR Check
Strikeouts (pitcher)PoissonUsually consistent, VMR ≈ 1
Hits allowedPoissonStable rate per inning
Earned runsNegative BinomialOften clustered (big innings)

Zero-Heavy Props → ZIP or Hurdle

Prop TypeDistributionRationale
RBIsHurdleZero common, but player had at-bats
Stolen basesHurdleMany games with zero, but player was on base
Home runsHurdleRare event, player active
Strikeouts (opener/reliever)HurdleMay only face 3-4 batters
Hits (bottom of order)HurdleMay get pulled for pinch hitter

Key Insight

MLB Rule of Thumb: Most MLB player props are low-volume count data. Default to Poisson unless you see overdispersion (VMR > 1.3) or excess zeros. For RBIs and stolen bases, Hurdle is almost always appropriate.


NHL Distribution Guide

Moderate-Volume Props → Normal or Poisson

Prop TypeDistributionRationale
Shots on goal (star forward)Normal3+ expected, reasonably symmetric
Blocked shots (defenseman)PoissonCount data, usually consistent
HitsPoissonConsistent opportunity each game

Zero-Heavy Props → Hurdle (Almost Always)

Prop TypeDistributionRationale
GoalsHurdleZero very common (~70-80% of games), but players are on ice
AssistsHurdleSimilar to goals—rare but active players
Points (G+A)HurdleZero common for non-stars
Penalty minutesHurdleSome players rarely take penalties, but opportunity exists
Power play pointsHurdlePP time varies, but when on PP they're active

Tip

NHL Rule of Thumb: Goals and assists are classic Hurdle candidates. The player is playing—they're just rare events. Don't confuse "rare" with "no opportunity."


Quick Reference: Distribution Selector

Use this tool to help determine the right distribution for your prop:

Negative Binomial Calculator

Try the interactive calculator for this concept

Open Tool

The Five-Question Checklist

Before betting any prop, answer these questions:

  1. Is this high-volume continuous data? (50+ yards, 15+ points)

    • Yes → Normal
    • No → Continue
  2. What's the VMR?

    • VMR ≤ 1.3 → Poisson
    • VMR > 1.3 → Negative Binomial
  3. Are there excess zeros? (Compare observed to e^(-λ))

    • No excess → Use distribution from Q2
    • Yes excess → Continue
  4. Why are there zeros?

    • Player sometimes has no opportunity → ZIP
    • Player always active, stat just rare → Hurdle
  5. Does my mean projection match my π and λ assumptions?

    • E[Y] should equal (1-π) × λ for ZIP/Hurdle

📝 Exercise

Instructions

Exercise: Match the Prop to the Distribution

For each prop, select the most appropriate distribution based on the guidance in this lesson.

Patrick Mahomes passing yards (expected: 285 yards, σ = 55 yards)

Connor McDavid goals (expected: 0.55 per game, zero in 68% of games)

Backup TE receptions (expected: 1.2 per game, VMR = 1.1, but he's a blocker-only in 40% of games)

MLB pitcher strikeouts (expected: 6.2 per game, VMR = 1.05, consistent starter)

An NBA backup center's blocks (expected: 0.8 per game, zero in 55% of games, but plays 12 minutes every game)

Running back rushing TDs (expected: 0.85, VMR = 1.75, games range from 0-3 TDs)