Do baseball winning and losing streaks happen more often than pure luck would predict? You look at every game from the 2001 National League season and count how often teams went on streaks of each length.
Then you create two artificial seasons based on chance alone. One uses a Monte Carlo simulation (a coin-flip model). The other uses calculated streak formulas. You compare the real streak counts to the artificial ones using standard deviation.
The real streaks differ from the random models at both 95% and 99% confidence levels. Streaks in baseball are not purely random. Factors like pitcher rotation and playing games in a series likely play a role.
Hypothesis
The hypothesis is that factors other than random chance affect the frequency of win/loss streaks in baseball.
If each baseball game were an independent coin flip, how often should winning streaks occur? A Monte Carlo simulation generates thousands of random seasons to set that baseline. When you compare real streak counts against those predictions at the 95% confidence level, real streaks exceed the random model — revealing that non-random factors, not just chance, are shaping outcomes.
Standard deviation measures how far individual data points sit from the group average — it sets the boundary for what counts as normal random variation. In this project, you compare real baseball streak counts to artificial seasons built on chance alone. If real streaks fall more than two standard deviations from the model’s average, you can say with 95% confidence that something beyond luck is shaping the results.
Statistical significance tells you whether a pattern is too strong to blame on chance. Here, you build artificial seasons based on chance alone and compare them to real baseball streak counts. If real streaks differ from the random models at both 95% and 99% confidence levels, the result is statistically significant — meaning something beyond luck is shaping how teams win and lose in a row.
Method & Materials
You will compare the streak frequencies of the National League in 2001 to two artificial seasons. One will be a Monte Carlo simulation using a coin flip and the other will be a season based on calculated streak frequencies.
You will need data from the National League in 2001, a coin, and a calculator.
MEL Math — hands-on math experiment kits delivered monthly — makes abstract concepts tangible. (Affiliate link)
Our experiment showed that the distribution and frequency of streaks in Major League Baseball is not purely random. Factors contributing deviation from randomness could include pitcher rotation or playing games in series.
Why do this project?
This science project is unique because it looks at a phenomenon that is often taken for granted and tries to explain it.
Also Consider
Experiment variations to consider include looking at different leagues or different years.
Full project details
Additional information and source material for this project are available below.
