My husband was looking at World Series stats with me last night and we have noticed the following fact:
When a team wins the first game of the world series they win the whole series 61% of the time. But, if the series goes to seven games, then the team that won the first game loses 62% of the time!
The question? Why is this? It seems to go against intuition.
Here is the full break-down.
In 4-game series the game-1 winner has won the series 100.% of the time
(18 out of 18).
In 5-game series the game-1 winner has won the series 82.6% of the time
(19 out of 23).
In 6-game series the game-1 winner has won the series 43.5% of the time
(10 out of 23).
In 7-game series the game-1 winner has won the series 38.2% of the time
(13 out of 34).
Now, we ended up staying up very late last night looking for reasons for this. I got so annoyed I wrote a program in PHP to simulate the situation. In it two evenly matched teams would play a world series 500,000 times.
Here is the full break-down from the model.
In 4-game series the game-1 winner has won the series 100% of the time.
In 5-game series the game-1 winner has won the series 75% of the time
In 6-game series the game-1 winner has won the series 60% of the time
In 7-game series the game-1 winner has won the series 50% of the time
So, maybe the 38.2% is just noise? We only have 34 data points.
Here is the model:
http://www.futurebird.com/...
(It takes about 20 sec. to load.)
A won 249877 times
B won 250123 times
Percent A wins: 49.9754
This shows the teams are evenly matched.
Here are all of the ways for A to win:
A won first then A won in 4 as a percent of all series won by A: 12.6%
A won first then A won in 5 as a percent of all series won by A: 18.7%
A won first then A won in 6 as a percent of all series won by A: 18.8%
A won first then A won in 7 as a percent of all series won by A: 15.4%
Total: 65.5%
B won first then A won in 4 as a percent of all series won by A: 0%
B won first then A won in 5 as a percent of all series won by A: 6.2%
B won first then A won in 6 as a percent of all series won by A: 12.5%
B won first then A won in 7 as a percent of all series won by A: 15.8%
Total: 34.5%
This helps verify the first stat:
When a team wins the first game of the world series they win the whole series 61% of the time.
It should be 65.5%, but this is close enough.
Still, what about:
But, if the series goes to seven games, then the team that won the first game loses 62% of the time.
My model says this should be more like 50%. Could it be psychological factors? Statistical noise? What do you think?