Are LCS Fans Good Predictors of Outcome?

I have gone through all the past LCS results on lolesports.com from the 2013 Spring Split to the 2014 Spring Promotion Series and looked at how the fans voted, and how the matches actually ended up.

I wanted to compare three different methods:

1. How the fans actually voted
2. How they compare to flipping a coin
3. How they compare to picking the team with the better record

How did the fans vote?

Out of 406 matches with polling data available, the fans were right on 236 of them, for a respectable 58.1% accuracy on picking the correct winner of the match. The news isn't all good though. When the fans thought a team was 75% likely to win a match by the voting, we would expect that the team they voted for would come out ahead three out of four times. Instead, they were right only 52% of the time, suggesting that they dramatically overestimate the relative strengths of teams.

Confidence	# Predictions	# Correct	Actual %
50%-55%	43	22	51%
55%-60%	53	26	49%
60%-65%	47	28	60%
65%-70%	53	26	49%
70%-75%	47	30	64%
75%-80%	44	23	52%
80%-85%	49	35	71%
85%-90%	45	30	67%
90%-95%	25	16	64%

Are they better than random chance?

Thankfully, the fans are better than random chance. There would only be a 0.043% chance of getting 236 correct estimates if you were to flip a coin on each match. So there is definitely some predictive ability. But should we be betting on the fans being right, or can we do better?

Marcel the Monkey's Picks

In baseball stat analysis, Tom Tango came up with a method of estimating player performance called the Marcels, because they were so simple that even a monkey could perform them. So I came up with a similar concept for the LCS, which follows these simple rules for picking a winner:

1. Pick the team with the higher win rate in that split
2. If they are tied, go with the blue side

Using this incredibly simplistic system, I was able to successfully predict the outcome of 288 matches, or 70.9% of them. In other words, the fans are worse than the most basic of predictions that a rational human would make. LCS fans are definitely not a good example of the Wisdom of Crowds.

Methodology

I have included my spreadsheets with the raw data here:

Please feel free to use these for data analysis. They were scraped from lolesports.com using jQuery and the analysis was done in Excel.