Note on a Coin Tossing Problem Posed by Daniel Litt

Bruce Levin

2024-09-20

Abstract:We present an analysis of a coin-tossing problem posed by Daniel Litt which has generated some popular interest. We demonstrate a recursive identity which leads to relatively simple formulas for the excess number of wins for one player over the other together with its increments as the number of coin tosses increases.

Combinatorics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **In the coin - tossing game played by Alice and Bob, who is more likely to win?**

Specifically, the rules of the game are as follows:
- Alice and Bob toss a fair coin \(n\) times.
- If heads come up twice in a row (HH), Alice gets 1 point.
- If a head is followed by a tail (HT), Bob gets 1 point.
- The core of the problem is: for any number of coin - tosses \(n>2\), is Bob more likely to win than Alice?

### Main Objectives of the Paper
The paper proves the following conclusions through mathematical analysis:
1. **For any number of coin - tosses \(n > 2\), Bob's probability of winning is greater than Alice's.**
2. This result is proved by the recurrence formula and the definition of the "heady close - call" sequence.

### Key Concepts
1. **Define the Score Difference**:
- Given a binary sequence \(x=(x_1,x_2,\dots,x_n)\) of length \(n\), where 1 represents heads and 0 represents tails.
- Define \(S(x)\) to represent Alice's score minus Bob's score.
\[
S(x)=\sum_{i = 1}^{n-1}[I(x_i = 1,x_{i+1}=1)-I(x_i = 1,x_{i+1}=0)]
\]
- If \(S(x)>0\), Alice wins; if \(S(x)<0\), Bob wins; if \(S(x) = 0\), it is a draw.

2. **Define the Difference in the Number of Winning Sequences**:
- \(D_n\) represents the number of sequences of length \(n\) in which the number of Bob - winning sequences minus the number of Alice - winning sequences.
\[
D_n=\sum_{S(x)<0}1-\sum_{S(x)>0}1
\]

3. **Recurrence Relation**:
- The paper proves the recurrence relations between \(D_n\) and its increment \(\Delta_n=D_{n + 1}-D_n\):
\[
\Delta_n=h_{2}(n)-h_{4}(n)
\]
where \(h_2(n)\) and \(h_4(n)\) represent the numbers of certain specific types of sequences (called "heady close - call" sequences) respectively.

4. **Final Conclusion**:
- Through the recurrence formula, the paper proves that for all \(n\geq3\), \(D_n>0\), that is, the number of Bob - winning sequences is always greater than that of Alice.

### Summary
The paper solves a probability problem proposed by Daniel Litt and proves through mathematical derivation that in the coin - tossing game, Bob is more likely to win than Alice. This conclusion is not only applicable to a specific number of coin - tosses (such as 100 times), but also applicable to any \(n>2\) situation.