How Does it Work?
The tool works by converting the inputted scores to standardized Z-Scores, then converting those Z-Scores back into other sports. To do this data was collected for each sport in the tool. Data includes:
-
NFL - Every outcome of every game in NFL History after 1940
-
NBA - Every outcome of every game in NBA history after 1950
-
NHL - Every outcome of every game in NHL history 1918
-
MLB - Every outcome of every game in MLB history since 1871
-
Soccer - Every outcome of every World Cup match since 1930
-
NCAAF - Every NCAA Football game since 1969
Using this data, 4 values were calculated for each sport:
-
The average score of a winning team
-
The average score of a losing team
-
The sample standard deviation of scores for winning teams
-
The sample standard deviation of scores for losing teams
All data was assumed to be normally distributed. The following figures show the probability distributions of winning and losing scores for each of the 6 sports
The inputted scores are then converted to a Z-Score: (inputted score - average corresponding score) / (standard deviation of the corresponding score). This Z-Score is then used to find the associated expected score in another sport.
There are some issues
-
0 points - though it is technically impossible to win with a score of zero, by assuming a normal distribution, the winning team may be assigned a value of 0 (this is visualized in the above figures)
-
Flipping outcomes - at extremely high scores or very close scores, the outcome will change. This is due to the winner "underperforming" based on the distribution and the loser "overachieving"
-
Rounding - I chose to not display factional scores, so some scores are closer or not as close as the simulation actually says. For example, a score of 1.1 - 2.9 and a score of 1.4 to 2.6 will both display as 1-3, though the exected win differential is 0.6 greater in the first case.
This is just for fun, so don't take any results too seriously.
