New to Prediction Markets? Four Tips to Get Started.

By Ben Golden on September 06, 2017

Joining a prediction market or prediction pool can be intimidating with difficult questions, ongoing discussions, and established leaders, you're likely to feel unsure of where to jump in. You might think there's no way you could possibly add anything, or that your forecasts couldn't possible be better than anyone else's. I've been there. Before I started working for Cultivate Labs building forecasting software, I first experienced prediction markets as a user. This blog post will highlight some of my tips to overcome early anxieties and get started with forecasting.

Tip #1: Don't Look for the 'Right Answer'

Unlike the factual questions on quizzes in school that had known answers, the questions in prediction markets are typically reasoning-based questions. These questions are asked in prediction markets because no one really knows the answer and sourcing insights from a crowd with pertinent knowledge produces more accurate conclusions. If question authors knew the correct answer one way or the other, they wouldn't need to use a crowdsourced forecasting site to get one. That's where you come in!

Similarly, when you see other site users making forecasts, keep in mind that they're acting under a great deal of uncertainty. They're making guesses, possibly educated ones and possibly not. You don't need to be an expert to make a forecast--it's okay to go on a hunch or educated guess, and it's okay to be wrong sometimes, or even a lot.

Tip #2: Google the Text of the Question and Look for Clues

No matter what question you're looking at, it's worthwhile to copy the text of the question into a search engine and hit enter. The results might surprise you--there's a decent chance that someone else out there is trying to figure out the same thing you are, and talking about it online.

As an example, here's a question we're asking on Alphacast:

Will the NY Yankees win the 2017 American League East Division title?

When I copy that into Google, my results include news stories, some nonsense (the Wikipedia article for the Yankees), and also a really helpful link to a site that literally gives me an answer to the question I typed, in the form a probabilistic forecast. If I didn't have a way to forecast this question before clicking that link, I certainly do after.

It doesn't always work out this nicely. Prediction markets and prediction pools often ask questions that are very obscure, specific to a certain company or event. But even if a question is Will we meet our Q4 revenue target? or Will my friends Paul and Judy get married?, I'd still take the time (probably less than a minute or two) to copy/paste these questions into Google and see what happens.

Tip #3: Pay Attention to What Everyone Else is Doing

If a question has been active for a while and other users have made forecasts, take a look at what they've been doing. Are users' forecasts fairly consistent, or do they vary widely? Has anyone left interesting comments? Do any of the forecasters have strong track records? Are forecasts trending upwards or downward over time? Again, the idea here is to find some clues to help you form an opinion, and seeing how other people think about the problem may help you to do so.

Tip #4: Start with a Small Forecast and Set a Reminder to Update it

If you're unsure about a question, consider making a small forecast--something close to the consensus--and plan to adjust it down the road. When you make a forecast on Cultivate Forecasts, it will save the question in your history, so you'll see it in your Dashboard / My Forecasts section. It will also give you the option to set an email reminder.

Suppose for a particular question, the consensus answer is 25% and you forecast 25%. Your forecast is unlikely to have much effect on your score, whether you're using a prediction market or prediction pool, but now you're in the game. Then you come back two days later, and the consensus has moved to 32%. You've had some more time to think about the question, and maybe you have a stronger opinion. Maybe you think 32% is too high, and that 25% really should be the answer. Or, maybe you think it should be higher than 32%, and change your forecast to 40%. Each time you visit a question, you can look at new forecasts, read other users' comments, do a little research, and update your thinking on the question.

As you learn more, your forecasts likely will improve, both for a particular question, and overall. Hopefully these tips will help you get started!

Ben Golden, Cultivate Labs Engineer

Blog Post By: Ben Golden


prediction markets crowdsourced forecasting