How do you make money forecasting?

In the Operations management class that I teach this semester this week is dedicated to forecasting. While we mostly focus on methods, for example, time series and causal models, it is also important to think about business aspect of forecasting. That is – how do you make money doing it?

Suppose you are interested in predicting an election outcome.Traditionally, companies like Gallup or Rasmussen run surveys and polls, and this can be quite expensive. If you can substantially reduce costs, then you can lower price and compete successfully with bigger companies. How?

It is called prediction markets. The basic idea comes from sporting bets: if you make people bet money on the outcome then the odds ratio can be very telling about its chances.One can take it a step further and make a security whose payoff will be contingent on realization of the outcome. People will then buy and sell it, and its price will be equivalent to the probability of the outcome.

Two prominent companies that implemented this idea are Hollywood stock exchange and Intrade. Their business model is very similar to “hotels without rooms” and “car rentals without cars”, the companies¬† I posted about before. I think the key feature of these innovations is that the companies are able to tap on resources that they do not own. In case of forecasting, by properly incentivising people to share their private information these firms effectively form a forecast which is based on a broad sample and information from multiple sources. All done at a very small cost.


4 thoughts on “How do you make money forecasting?

  1. While I think business models that redistribute client resources are clever and can certainly succeed, I’m doubting the value of this particular model with regards to forecasting.

    The most evident analogy of using a market to get people to volunteer information is the efficient market hypothesis with regards to the stock market. There,hoards players analyze and scrutinize securities; their livelihoods depend on this analysis.However, even then, accurate reflection of information through prices is debated.

    Are there really enough people with private information that do analogous research into these tiny markets for their “bet” to (a) provide valuable information (b) accurately move the price?

    If these things got big, then sure, they could be valuable as forecasting tools that crowd-source research and information gathering. As of now though, they seem too small and too poorly observed to generate any efficient markets. I don’t see the application of any forecast generated by such means.

    That being said, I guess if you do have private information, they could be vehicle to make a quick buck. You wont make much, and there wont be enough of you to provide any valuable information for anyone else, but easy money all the same.

  2. I think this market model is efficient for things that have forecasting feedback (I don’t know if this is the right word). This is due to the link between the price you pay for a share and the probability of the event occurring. Therefore as the event is forecasted to be more likely, you stand to lose more when it fails to occur. Or when an event is forecasted to be less likely, you stand to make more when it does occur. Therefore there is an incentive to buy and sell shares until the price become equal to the probability, total efficiency, and then you do not have any incentive to trade these securities anymore.

    From Intrade: “The market prices of shares also indicate the probability of the event happening. For example, a market price of $3.63 indicates a 36.3% probability the event will actually happen, according to the market. In other words, the market is predicting a 36.3% probability.”

    Prices start from $0 to $10. $0 indicate 0% probability and $10 indicates 100% probability.

    For example, betting that Barack Obama will win the presidency in 2012 has a forecasting feedback because when you buy a share at Intrade, you are also more likely to vote for Barack Obama.

    Also, buying a share for a new release movie will make you more likely to see it therefore boosting the box office price and increasing the probability that the movie is a success. Or knowing that your friends are all going to see this new release also may be an incentive to buy a share unless the price is too high.

    This model won’t work for things that do not have a forecasting feedback. Like stock prices or future predictions that are uncertain (not become more certain with time). Also, too much liquidity and speculation in these markets may create volatility which will wipe away incentives for efficiency.

    • It is possible that individual predictions are correlated with individual actions, I agree. To which degree this can affect the price of prediction is debatable. I think the model can actually work for stock prices, at least for long term forecasting. You can bet that SP 500 closes, say above, 1500 on Dec. 31 2012. It’s a verifiable event – so the payoff will be determined.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s