Analysis Of 1.8+ Million WallStreetBets Predictions

WallStreetBets is a subreddit that discusses economics and the stock market. Each month, they run predictions-tournaments whereby users attempt to predict things such as stock price movements, inflation and GDP. I've collected data from over 200 000 users, who in total made over 1 800 000 predictions, in order to analyze if they can predict the future - at least better than random chance.

WallStreetBets vs. Monkeys

There's an infamous story about a chimpanzee that threw 10 darts at a dartboard of 133 internet related companies in January of 1999. By year end, these stocks had thoroughly beat the market and outperformed more than 6,000 money managers.

“A blindfolded monkey throwing darts at a newspaper’s financial pages could select a portfolio that would do just as well as one carefully selected by experts.” -Burton Malkiel, author of A Random Walk Down Wall Street.

The monkey in this case, functions as a random input generator. In the metaphorical analysis of WallStreetBets versus monkeys, we're really comparing WallStreetBets to randomness. So the question is whether users can predict the future better than pure raw chance?

Technical details

The prediction tournaments are typically set up like this, with a question prompt like "How will [company] close following their Earnings Report on [date]" and either 3 or 5 answer options:

If there are 5 answer options, the proverbial monkey would be correct about 20% of the time. If there are 3 answer options, the success rate would be about 33.33%.

Results

In 142 predictions, the expected wallstreetbets score is 41.66 - which gives them a success rate at 29.34%. Meanwhile, the monkey who picks completely at random, would score 38.93 out of 142, or 27.42%.

The winner is WallStreetBets!

WallStreetBets vs. Basic Strategy

I'll propose a very basic strategy that connivingly beat monkeys and WallStreetBets alike. The tactic is simple, always choose the middle option. In practice, this means always predicting flat price movement of the 5 (or 3) answer options.

Using this strategy, you would historically be correct 60 out of 142 times, or 42.25% of the time.

WallStreetBets beats monkeys, but not much more

The analysis shows that WallStreetsBets may be better at predicting future stock price movements, GDP and inflation than pure randomness, but fail to beat even very basic strategy.