Homework 5

Prediction Markets: Tapping the Wisdom of Groups
(Due Tues. 10/22, which is one week from now)

Recall that prediction markets try to combine peoples' best judgments using a financial incentive to make sure the judgments really are the best they can make. The topic is timely: the blog entry with the lecture on prediction markets is the first posting to the course blog that has resulted in comments from outside. There is also an academic journal that started in 2007 which is devoted to scholarly examination of prediction markets, their structure, how well they work, etc.

This HW will hopefully be fun, and will help you gain a hands-on understanding of how prediction markets work.

1) (22.5 pts.) Let's assume you have $1,000 to invest in prediction markets. Naturally, you want to make more money, meaning that when you sell your best judgments you get more than the $1,000 you invested. Identify a prediction market company that you want to pretend invest in. We looked at www.intrade.com in class but there are others. Furthermore, www.intrade.com has both a set of real markets where you can invest (or some might claim gamble) real money, and a practice market site where you can invest pretend money. You can pretend invest in the real market without signing up for a login, or you can make a login account with them and use the practice market. Or you can find another prediction market other than intrade.com. The one thing you cannot do is open a real account (as part of this HW) because this course does not endorse, or suggest you to spend real money on, any particular company! For the next question let's assume you are making pretend use of intrade's set of real markets.

2) (22.5 pts.) Find 10 prediction markets that you are interested in or know more about than most others. For example, you might pick the prediction market for whether 2009 will be one of the 5 hottest years on record, the prediction market for whether Michael Jackson's death will result in homicide charges, and some other eight. In fact there are dozens and dozens if not hundreds of markets they have easily accessible from their Web site. Make a list them.

3) (22.5 pts) For each of the 10 markets, decide whether the current probability shown on the graph is higher or lower than your best judgment. List the ones for which you judge the current probability to be too low.

4) Invest your $1,000 in the ones you feel are currently too low, as follows.

• a. (3.5 pts.) Pick one of the markets. Say which one
• b. (3.5 pts.) Figure out the price of a single contract (a contract is a little like a share of stock). That price is: (the probability in percentage points) x ($0.10). So if the probability on the graph is 65%=0.65, then a single contract will cost you 65 x $0.10, or $6.50, to buy. Give this price in your HW.
• c. (3.5 pts.) Decide how many contracts to buy (remember you have $1,000 but will want to spread that over a number of different markets). Note the number in your HW.
• d. (3.5 pts.) Figure out the total amount you are investing in this market. State this.
• e. (8.5 pts.) Subtract that from $1,000 to find out how much you have left to invest in the other markets. Then go through the above steps again using another market. Repeat until you have invested all of the $1,000 in the markets you have chosen to invest in.

5) (10 pts., to be assigned toward the end of the semester. 10 if you make money, but only 5 if you lose money, sorry!) You may sell any of your contracts at any time during the semester, and buy any new contracts you like with the money you got from selling (or just hang on to the money if you prefer). You are not required to do this, if you just want to keep the same contracts. To sell, just see what the contracts you have are worth, using the graphs at the company Web site, and pretend you've sold them. Keep track of what you do by editing your blog entry for this HW. Toward the end of the semester I will ask everyone to sell everything and we'll see how much money you've made or lost!

NOTE: the focus above is on markets where the probability is too low. If the probability is too high, it is possible to bid a lower price, and the company will automatically buy the contracts if the probability goes down enough to match your bid. But that is outside the scope of this HW.