Mastering Cumulative Geometric Probability- Formulas and Examples
What Is Cumulative Geometric Probability?
Geometric probability deals with repeated independent trials until you get your first success. Cumulative geometric probability asks a different question: what's the probability of getting your first success on or before trial k?
Instead of "what's the chance I succeed exactly on trial 5?" you're asking "what's the chance I succeed by trial 5 at the latest?"
This distinction matters. A lot.
The Core Formula
For a single trial success probability p and failure probability q = 1 - p:
P(X ≤ k) = 1 - qk
That's it. The cumulative probability that your first success occurs on or before trial k equals one minus the probability that you fail k times in a row.
Why This Works
If you fail k times, each failure has probability q. Multiply them together: q × q × ... × q (k times) = qk. The only way to NOT have a success by trial k is to fail every single time. So subtract that from 1, and you get your cumulative probability.
Worked Examples
Example 1: Manufacturing Defects
A factory produces widgets with a 3% defect rate. What's the probability you find your first defective widget within the first 10 inspections?
p = 0.03, q = 0.97, k = 10
P(X ≤ 10) = 1 - (0.97)10
P(X ≤ 10) = 1 - 0.7374 = 0.2626
You have roughly a 26.3% chance of finding a defect within 10 inspections.
Example 2: Free Throw Shooting
A basketball player makes 75% of free throws. What's the probability she makes her first basket by her 4th attempt?
p = 0.75, q = 0.25, k = 4
P(X ≤ 4) = 1 - (0.25)4
P(X ≤ 4) = 1 - 0.0039 = 0.9961
She'll almost certainly make a basket by the 4th try. The chance of missing 4 free throws in a row is under 0.4%.
Example 3: Password Cracking
A hacker tries login attempts with different passwords. Each attempt has a 2% success chance. What's the probability they crack the account within 50 tries?
p = 0.02, q = 0.98, k = 50
P(X ≤ 50) = 1 - (0.98)50
P(X ≤ 50) = 1 - 0.3642 = 0.6358
Over 63% success rate within 50 attempts. This is why rate-limiting exists.
Geometric vs. Cumulative: What's the Difference?
| Type | Question Asked | Formula | Use Case |
|---|---|---|---|
| Geometric (Point) | Success exactly on trial k | P(X = k) = qk-1 × p | Isolated event analysis |
| Cumulative | Success on or before trial k | P(X ≤ k) = 1 - qk | Realistic deadline planning |
The point probability gives you the odds of a specific outcome. The cumulative probability gives you the odds of meeting your deadline.
Expected Value: The Average Trial Number
The expected value of a geometric distribution tells you the average trial number where success occurs:
E(X) = 1/p
For the 75% free throw shooter: E(X) = 1/0.75 = 1.33 attempts on average.
For the 3% defect inspection: E(X) = 1/0.03 = 33.3 inspections on average to find a defect.
Common Mistakes
- Using qk-1 instead of qk — Cumulative counts k failures, not k-1. Double-check your exponent.
- Confusing "≤" with "<" — P(X ≤ k) includes k. P(X < k) excludes it. The formula 1 - qk handles ≤.
- Forgetting that trials must be independent — Geometric probability breaks down if previous outcomes affect future ones.
- Rounding errors — qk compounds quickly. Keep more decimal places during calculation, round only at the end.
Getting Started: How to Calculate Cumulative Geometric Probability
- Identify your success probability p — What single-trial event are you tracking?
- Calculate q = 1 - p — Your failure probability.
- Set your cutoff k — How many trials maximum can you afford?
- Compute qk — Probability of k consecutive failures.
- Subtract from 1 — P(X ≤ k) = 1 - qk
- Interpret — Does the result meet your threshold? If not, increase k or reconsider your approach.
When to Use This
Cumulative geometric probability shows up everywhere:
- Quality control — How many units until you find a defect?
- Reliability engineering — How long until a component fails?
- Security testing — How many login attempts until unauthorized access?
- Medical testing — How many patients until you find one with the condition?
- Sales pipelines — How many cold calls until you book a meeting?
Any scenario where you're running trials until success, and you care about a realistic timeframe, needs the cumulative version. Point probability tells you what might happen. Cumulative tells you what will probably happen.