Practical Workflow and Workbook Exercises

You now have all the pieces. Let's put them together into a repeatable workflow for Negative Binomial prop betting, then test your skills with comprehensive exercises.

The 8-Step Negative Binomial Workflow

Key Insight

The Complete Process:

Identify the Prop Type (count-based: TDs, Ks, threes, goals)
Gather Historical Data (15-20 games minimum)
Calculate VMR (if < 1.3, use Poisson and stop here)
Calculate r using method of moments: r = μ² / (Variance - μ)
Adjust μ for Matchup (keep r constant—it reflects player's nature)
Calculate Probabilities using NEGBINOM.DIST
Compare to Market and Calculate EV
Bet if +EV and Within Bankroll (Kelly sizing from Chapter 5)

Workflow Decision Tree

START: Count-based prop (TDs, Ks, 3PM, goals, etc.)
  │
  ▼
Gather 15-20+ games of data
  │
  ▼
Calculate VMR = Variance / Mean
  │
  ├── VMR < 1.3 ──────► Use POISSON
  │                        │
  └── VMR ≥ 1.3 ──────► Calculate r = μ² / (Var - μ)
                           │
                           ▼
                     Adjust μ for matchup
                     (keep r constant)
                           │
                           ▼
                     Calculate p = r / (r + μ)
                           │
                           ▼
                     Use NEGBINOM.DIST for probabilities
                           │
                           ▼
                     Compare to market implied probability
                           │
                           ├── Edge > 0 ──► Calculate Kelly, place bet
                           │
                           └── Edge ≤ 0 ──► No bet

Sport-Agnostic Principles

Regardless of which sport you're betting, these principles apply universally:

1. Overdispersion Is Real and Exploitable

Many players are boom-or-bust, not consistent. Books often assume consistency (Poisson). Your edge comes from identifying overdispersion and using Negative Binomial to get more accurate probabilities for extreme outcomes.

2. Extreme Outcomes Are Where the Value Lives

Negative Binomial's fatter tails mean higher probabilities for 0 occurrences and 3+ occurrences. These are often mispriced, especially 3+ outcomes at long odds (+800 to +1500). The market tends to underprice tails because they use Poisson or don't fully account for variance.

3. Adjust μ, Not r, for Matchups

The dispersion parameter r reflects a player's inherent boom-or-bust nature. It doesn't change game-to-game. Only adjust μ for matchup quality, pace, game script, etc.

4. Process Over Results (Even More Important)

Negative Binomial props have even higher variance than Poisson props. A 3+ TD bet at 11.5% probability will lose 88.5% of the time. You'll have brutal losing streaks. Trust the math over 100+ bets, not 10 bets.

5. Convergence at Middle Outcomes

For moderate μ (0.8-1.2) and moderate r (1.0-2.0), Poisson and Negative Binomial often give similar probabilities for middle outcomes (1-2 occurrences). The edge is in the extremes—0 or 3+.

Cross-Referencing Previous Chapters

Note

Cross-referencing Chapter 5 (Bankroll Management): With precise Negative Binomial probabilities, you can now implement Kelly criterion for boom-or-bust props. An 18.5% probability on a +800 line suggests a Kelly bet of about 8% of bankroll, but most pros use fractional Kelly (25-33%) to reduce variance, bringing it down to 2-3% of bankroll.

Note

Cross-referencing Chapter 8 (Poisson): You now know when to use Poisson (VMR ≈ 1.0) and when to use Negative Binomial (VMR > 1.3). The decision tree is simple: Calculate VMR first, then choose your distribution. This prevents systematic errors in your probability estimates.

Note

Cross-referencing Chapter 7 (Normal Distribution): For high-volume stats (100+ yards, 25+ points), the Normal distribution is still your best tool. Negative Binomial is for low-volume, count-based props where 0, 1, 2, 3 are the common outcomes. Know which tool to use for which situation.

Your Statistical Toolkit Summary

Distribution	Best For	Key Assumption
Normal	High-volume continuous (yards, points)	Symmetric, bell-shaped
Poisson	Consistent count data (VMR ≈ 1)	Variance = Mean
Negative Binomial	Boom-or-bust count data (VMR > 1.3)	Variance > Mean

Workbook Exercise 9.1: Build a Negative Binomial Calculator

Create an Excel sheet for Negative Binomial props with these columns:

Column	Header	Formula/Content
A	Player	Name
B	Stat Type	TDs, 3PM, Ks, etc.
C	Historical Mean	`=AVERAGE(data)`
D	Historical Variance	`=VAR.S(data)`
E	VMR	`=D2/C2`
F	Calculated r	`=C2^2/(D2-C2)`
G	Matchup Adjustment	Multiplier (e.g., 1.10)
H	Adjusted μ	`=C2*G2`
I	p Parameter	`=F2/(F2+H2)`
J	Line	0.5, 1.5, etc.
K	Odds	American odds
L	P(Over)	`=1-NEGBINOM.DIST(INT(J2),F2,I2,TRUE)`
M	Market Implied	`=IF(K2<0,ABS(K2)/(ABS(K2)+100),100/(K2+100))`
N	Edge	`=L2-M2`

Test Case: Boom-or-bust RB, Rushing TDs

Historical Mean: 0.85
Historical Variance: 1.03
VMR: 1.21
r: 4.00
Matchup Adjustment: 1.12 (plus matchup)
Adjusted μ: 0.95
Line: 0.5
Odds: -130 (Over)

Expected Output:

P(1+ TDs) = 56.4%
Market Implied = 56.5%
Edge = -0.1% (no bet)

Workbook Exercise 9.2: Overdispersion Detection Template

📝 Exercise

Instructions

Analyze these three players and determine which distribution to use for each:

Player A (Consistent): 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0

Player B (Moderate): 2, 0, 1, 0, 2, 1, 0, 1, 0, 2, 1, 0, 1, 0, 2, 1, 0, 1, 0, 1

Player C (Extreme Boom-or-Bust): 3, 0, 0, 0, 2, 0, 0, 3, 0, 0, 0, 2, 0, 0, 3, 0, 0, 2, 0, 0

Calculate Mean, Variance, VMR, and r for each.

Workbook Exercise 9.3: Poisson vs. Negative Binomial Comparison

📝 Exercise

Instructions

For a boom-or-bust player with μ = 1.30 and r = 0.67, calculate probabilities using both Poisson and Negative Binomial for k = 0, 1, 2, 3, 4. Identify where the +EV opportunities exist.

Hint: For Negative Binomial, p = r / (r + μ) = 0.67 / (0.67 + 1.30) = 0.340

Workbook Exercise 9.4: Full Prop Analysis Workflow

📝 Exercise

Instructions

Complete a full analysis for a hypothetical backup RB with the following:

Historical Data: μ = 0.75, Variance = 1.20

Market: 2+ Rushing TDs at +800

Matchup: Plus matchup, increase μ by 20%

Bankroll: 100 units

Walk through all 8 steps of the workflow and determine your bet size.

Moving Forward

The Negative Binomial distribution completes your toolkit for count-based props. You now have:

Poisson for consistent players (VMR ≈ 1.0)
Negative Binomial for boom-or-bust players (VMR > 1.3)
Normal for high-volume continuous stats (Chapter 7)

Key Insight

Remember the key principles:

Test for overdispersion using VMR
Use Negative Binomial when VMR > 1.3
Adjust μ for matchups, keep r constant
Extreme outcomes (0 or 3+) are where the value lives
Requires 15-20 games of data for reliable r estimates
Variance is brutal—trust the process over 100+ bets

But props don't exist in isolation. A quarterback's passing touchdowns correlate with his receivers' receiving touchdowns. A team's total points correlates with their star player's scoring props. In Chapter 10, we'll explore correlation and covariance—the mathematical tools for understanding how props move together. This becomes essential for Same Game Parlays and multi-prop strategies.

Negative Binomial Calculator

Try the interactive calculator for this concept

Open Tool