The 68-95-99.7 Rule (Empirical Rule)
One of the most powerful tools for understanding variance is the 68-95-99.7 Rule, also known as the Empirical Rule. This rule applies to normally distributed data (which many betting outcomes approximate with sufficient sample size) and gives you an instant mental framework for evaluating results.
Key Insight
The 68-95-99.7 Rule tells you:
- 68% of outcomes fall within 1 standard deviation of the mean
- 95% of outcomes fall within 2 standard deviations of the mean
- 99.7% of outcomes fall within 3 standard deviations of the mean
Visualizing the Rule
This rule helps you set realistic expectations and identify when something unusual is happening. Let's see it in action.
Practical Example: 100 Bets at 52% Win Rate
Let's apply this to our opening scenario from Lesson 1:
Given:
- 100 bets at 52% win rate
- Expected wins: 100 × 0.52 = 52 wins
- Standard deviation: √(100 × 0.52 × 0.48) ≈ 5 wins
Applying the 68-95-99.7 Rule:
| Range | Wins | Probability |
|---|---|---|
| μ ± 1σ | 47-57 wins | 68% of the time |
| μ ± 2σ | 42-62 wins | 95% of the time |
| μ ± 3σ | 37-67 wins | 99.7% of the time |
Note
Going 47-53 (47 wins) falls right within the normal range—it's inside one standard deviation. It's not bad luck; it's expected variance. In fact, you'll finish with 47 or fewer wins about 16% of the time, even with a genuine 52% edge.
The Tail Probabilities
Understanding what lies in the "tails" (outside the standard ranges) is equally important:
| Scenario | Probability |
|---|---|
| More than 1σ below mean | ~16% |
| More than 2σ below mean | ~2.5% |
| More than 3σ below mean | ~0.15% |
| More than 1σ above mean | ~16% |
| More than 2σ above mean | ~2.5% |
| More than 3σ above mean | ~0.15% |
These tail probabilities help you evaluate whether a result is normal variance or something to investigate.
Using the Rule for Player Props
The 68-95-99.7 Rule works beautifully for evaluating player props. Consider an NBA player:
LeBron James (2023-24 season example):
- Average: 25.7 points
- Standard deviation: 6.7 points
| Range | Points | What This Means |
|---|---|---|
| 68% | 19.0 - 32.4 | Most games land here |
| 95% | 12.3 - 39.1 | Almost all games land here |
| 99.7% | 5.6 - 45.8 | Virtually all games land here |
Betting Application: If you see a prop line at 35.5 points, you instantly know this is roughly 1.5 standard deviations above the mean—in the top 7% of expected outcomes. Unless there's a specific reason to expect an outlier game, the Under has mathematical support.
The Rule as a Red Flag Detector
The 68-95-99.7 Rule helps you identify when results are statistically unusual:
✅ Normal Variance (Don't Panic)
- Results within ±1σ: Completely expected (68% probability)
- Results within ±2σ: Normal fluctuation (95% probability)
⚠️ Worth Investigating
- Results beyond ±2σ: Unusual but possible (5% probability)
- This happens 1 in 20 times by pure chance
🚨 Red Flag
- Results beyond ±3σ: Extremely rare (0.3% probability)
- Only 3 in 1,000 samples should be here
- Investigate your model, data, or strategy
Warning
If you consistently see results beyond 2 standard deviations in the same direction, it's likely not variance—something may be wrong with your edge estimate, or the market conditions have changed.
Quick Mental Math Framework
Memorize these approximations for instant analysis:
| Standard Deviations | Approximate % Within | Approximate % Outside |
|---|---|---|
| ±1σ | 68% | 32% (16% each tail) |
| ±1.5σ | 87% | 13% (6.5% each tail) |
| ±2σ | 95% | 5% (2.5% each tail) |
| ±2.5σ | 99% | 1% (0.5% each tail) |
| ±3σ | 99.7% | 0.3% (0.15% each tail) |
Excel Formulas for Range Calculations
Calculate the expected ranges quickly:
# Lower bound (1σ)
=AVERAGE(A1:A100) - STDEV.S(A1:A100)
# Upper bound (1σ)
=AVERAGE(A1:A100) + STDEV.S(A1:A100)
# Lower bound (2σ)
=AVERAGE(A1:A100) - 2*STDEV.S(A1:A100)
# Upper bound (2σ)
=AVERAGE(A1:A100) + 2*STDEV.S(A1:A100)
Expected Range
Range = μ ± (k × σ)=AVERAGE(range) ± k*STDEV.S(range)Where k = 1, 2, or 3 for the respective confidence intervals.
📝 Exercise
Instructions
Apply the 68-95-99.7 Rule to analyze these betting results.
You place 400 bets with a true 53% win rate. The standard deviation is √(400 × 0.53 × 0.47) ≈ 10 wins. You end up with 192 wins (48% win rate). How would you classify this result?
📝 Exercise
Instructions
Use the rule to set expectations for a player prop.
An NBA player averages 22.0 rebounds + assists combined with a standard deviation of 4.5. The prop line is set at 27.5.
Questions:
- How many standard deviations above the mean is this line?
- What percentage of games would you expect them to go OVER 27.5?
- Is this an Under or pass situation?
Key Takeaways
-
68-95-99.7 is your mental framework for instant variance analysis
-
Within 1σ is normal—don't overreact to results in this range
-
Beyond 2σ is unusual—worth investigating but not impossible
-
Beyond 3σ is a red flag—something may be systematically wrong
-
Use it for players too—quickly assess if a prop line makes statistical sense
-
Tail probabilities matter—16% of the time you'll be 1σ below expectation
In the next lesson, we'll apply these concepts specifically to betting outcomes and learn how variance affects your bankroll over time.