A Simple Formula for Unknown Probabilities


In today’s data-driven world, we’re constantly making predictions based on limited information. Whether you’re betting on sports, launching a new product, or evaluating medical treatments, the challenge remains the same: how do we estimate probabilities when we don’t have much data? While I typically write about practical applications and real-world analytics, today I’m diving into a fascinating theoretical concept that has profound implications for how we make predictions. It’s a bit more mathematical than my usual topics, but I promise the insights are worth the journey into probability theory.

Enter Laplace’s Brilliant Solution

The Rule of Succession, developed by Pierre-Simon Laplace in the 18th century, offers an elegant solution:

p(success) = sucesses + 1 total trials + 2

This deceptively simple formula provides a powerful way to estimate probabilities with limited data.

A Deceptive Wheel

Let me share what sparked my interest in Laplace’s Rule of Succession. While playing a popular online slot game, I encountered a bonus wheel that appeared to offer a 50/50 chance between doubling my multiplier or keeping it the same. After several spins, I noticed I was getting far fewer multiplier increases than expected, which led me to explore probability estimation.

The Bonus Round

Consider a bonus wheel in a slot machine that has two seemingly equal sections: “Advanced Multiplier” and “Win – End Round” The example shown below is from the Star Trek game at Chumba Casino. The wheel makes it look like the chance of getting the “Advanced Multiplier” is 50/50. However, after spinning this wheel ten times, I only got the advanced multiplier once.

Star Trek Bonus Wheel

Star Trek Bonus Wheel

The Calculations for the Bonus Round

With this being the case, what is the estimated probability of getting this multiplier? Your initial reaction might be to calculate the probability as 1/10 = 10%.

However, this raw probability might be misleading. Using Laplace’s Rule of Succession:

Probability of getting multiplier = (successes + 1)/(trials + 2)
= (1 + 1)/(10 + 2)
= 2/12
= 16.7%

This adjusted probability better reflects the true odds, especially with limited data. Modern slot machines use complex Random Number Generators (RNGs) and often have hidden weightings that affect outcome probabilities. While a wheel might visually suggest equal odds, the actual probability can be significantly different.

The beauty of Laplace’s rule of succession formula is that it provides a more conservative and realistic estimate, particularly useful when dealing with games where true probabilities are intentionally obscured. This mathematical insight can help players make more informed decisions about their gameplay and manage their expectations about bonus features.

Why It Works

The rule adds two pseudocounts (one success and one failure) to account for uncertainty in small sample sizes. This approach:

  1. Prevents probability estimates of 0 or 1.
  2. Provides more conservative estimates when data is limited.
  3. Gradually converges to the observed frequency as sample size increases

The rule is most valuable when dealing with binary outcomes in situations with limited prior knowledge or small sample sizes.

Conclusion

Laplace’s Rule of Succession stands as a testament to the enduring power of elegant mathematical thinking. While simple enough to fit in a single formula, it captures deep insights about probability and uncertainty that remain valuable two centuries later. In an era of big data and complex algorithms, sometimes the most brilliant solutions are the simplest ones.