Predicting the Future with Futures

Commentator Brian recently pointed to a speech by Michael Crichton which quoted a saying attributed to Mark Twain: "I’ve seen a heap of trouble in my life, and most of it never came to pass." As Eliezer reminded us, predictions of the future are seldom accurate and seldom remembered. And as Twain’s quote indicates, predictions tend to be over-dramatized and exaggerate the difficulties and challenges that lie ahead.

Given these problems, it is helpful to realize that in many cases we can extract information about the future from a source which can be expected to be reasonably free of bias, and in many cases provides a track record that can be used to judge its accuracy: futures markets. I won’t try here to summarize or justify the reasons why such markets can be expected to provide among the best available sources of predictions, but rather below the fold I offer some predictions on current events and situations derived from various markets. is one of the most widely used real-money futures markets for non-financial topics. You can bet on questions as diverse as the next U.S. Presidential election or who will be voted off American Idol this week. Unfortunately many of its markets are sparsely traded and don’t provide clear predictions, but we can still extract some useful information.

One current major international flashpoint is Iran. Intrade asks whether there will be an air strike against Iran by the U.S. and/or Israel. The odds are seen as relatively low: about a 10% chance before the end of June, 15% by the end of September, and 25% by the end of the year. It’s notable that even at the peak of the crisis involving Iran’s capture of British troops, the odds never got above 30%. These figures can serve as a good antidote to the sometimes hysterical predictions one reads online.

Or perhaps you are worried about Bird Flu. Will it come to the U.S.? (Apologies for the America-centricism, but apparently many Intraders are from the United States.) I notice that the media often do not clearly distinguish Bird Flu in birds vs in humans. I had to read the contract carefully to determine that this is about Bird Flu in birds. Even so, the odds are low, in the range 10-30% by the end of the year. Bird Flu is something of a seasonal obsession in the media, so if you start hearing more about it, this market can be a good source of unbiased information.

Recession is another topic much in the news. Will the U.S. economy continue to grow? Or will the current slowdown turn into a full blown recession? The normal financial markets shed some light here, but Intrade has a specific bet on the probability of U.S. recession before the end of 2007. Encouragingly (for those who don’t like recessions) the odds are only 17%. This is one topic where the betting markets disagree with polls. Today the Los Angeles Times reports poll results indicating a much higher belief in a recession this year. 60% said a recession was somewhat or very likely within the next year. The Times reports, "Most experts, including Federal Reserve Chairman Ben S. Bernanke, say there’s little chance of a recession." So in this case the markets agree with the experts and not with polls. It will be interesting to see which is correct.

Another area of disagreement relates to real estate prices. Respondents were asked to predict the change in housing values six months ahead. Results were: Increase: 32%, Decrease: 16%, Remain about the same: 51%. The Chicago Mercantile Exchange runs a real estate futures market (free registration required) which lets traders bet on the future of housing prices. Over the next six months, the index is predicted to drop from 218 to 214 to 212, about 3 percent. This prediction of a modest decrease contradicts the more bullish poll results. At least it serves as an anodyne to the constant drumbeat of "imminent crash" one finds in the more bearish corners of the net.

Financial markets provide much useful information, but it often requires interpretation before it can translate into sensible predictions. A market on the Chicago Board of Trade can be used to predict the interest rate policies of the U.S. Federal Reserve Board in advance, but the nature of the bidding is somewhat complex. The Cleveland Federal Reserve Bank turns market prices into easy to read graphs of market consensus about future Fed actions. Presently, interest rates are expected to remain unchanged at 5.25% over the next three Fed meetings, through August (that’s as far out as the graphs go). And the market is very convinced of this, seeing at least an 80% chance of stable rates all the way out. Then I found a Reuters article from Monday suggesting that the same markets when extrapolated to the end of 2007 predict an easing to about 5%. This is up from last week’s prediction of a reduction to about 4.8% at that time, because of stronger than expected financial news last Friday. The markets apparently see a small reduction in interest rates over the next year as the most likely policy, confirming low expectations of a recession.

Commodity futures markets are useful in providing simple and straightforward information about future prices. There is much worry these days about how high gasoline prices will get this summer. Reformulated unleaded gasoline futures are traded on the Nymex market and provide some hints. Now, these are wholesale prices, so you have to add about sixty cents per gallon to get retail estimates. The good news is that prices are expected to drop slowly for the rest of the year, from $2.15 in the near month to $2.04 for September delivery. And if you look at option prices to derive risk-neutral probability estimates, the chances of prices above $2.30 or $2.40, which would correspond to about $3.00 retail, are only 20-30% this summer. Markets seem very confident that we won’t see sky high gasoline prices in the U.S this year. This is another result which seems far more optimistic than the mood online, and is again more in accordance with expert predictions that gas prices will remain well below $3.

The last market I could mention is the venerable Foresight Exchange, the original Idea Futures market set up in response to Robin Hanson’s writings on the subject. FX is a play-money market unlike the others I have discussed here, but there is some evidence that play-money markets do as well as real-money ones. I have played the FX game almost since its inception in 1994, and during that time my score has remained stubbornly near 1.0, meaning that my wins and losses have almost perfectly balanced out. In other words, I could have done just as well by flipping a coin in making my predictions. This is part of why I am so enthusiastic about our discussions of modesty. Playing this game for so long has given me a much more sobering view of my own predictive abilities.

FX is worth a whole post in itself, with many surprising and interesting phenomena which have occured over the years. So I will wait to discuss it until I have time to give it the space it deserves.

Finally, about American Idol – Intrade has Melinda Dolittle as a shoe-in, with an outright majority at 53% odds, 3 times better than the next contestant.

GD Star Rating
Tagged as:
Trackback URL:
  • Why do you say that futures markets can be ‘expected to be reasonably free of bias?’

  • Anna

    Hal: Finally, about American Idol -Intrade has Melinda Dolittle as a shoe-in, with an outright majority
    at 53% odds, 3 times better than the next contestant.

    When weighing the odds I would figure that Melinda, the guy with the funky hair and the replica of Mr. Timberlink:) have more of a chance but how to assume the odds aren’t already decided? I might have an opinion to what is the majority of thought but does it really matter? Probably not but it’s at least interesting to know “what” is the majority of favorite opinion.


  • sa

    great post. it might be relevant to say here that with the advent of internet and electronic archive search, compiling an expert accuracy score much easier than it used to be. this is esp true for equity/fixed income/commodities strategists who don’t trade but are always on tv and in the news.

  • Bruce, that is a good question and it is probably an exaggeration to say that betting markets are “free of bias”. Any kind of institution can be subject to some biases or other. The difference with betting markets is that participants are putting money behind their claims, so that any biases can be exploited by others at the expense of those making the biased claims. And this exploitation will tend to reduce the bias in the market. The main point is that this is unlike the other kinds of futuristic predictions described by Eliezer, where there are virtually no incentives to be unbiased or penalties for even rather wildly incorrect predictions.

  • Something one needs to be careful about here is what is known in international finance as the “peso problem,” which can arise if an underlying distribution is skewed in one direction. Thus in small samples observed means will not equal population means if one does not observe an extreme tail event.

    The original example was studied by Kenneth Rogoff in his Ph.D. thesis at MIT in 1979. The “peso problem” was that the forward market for the Mexican peso was consistently undepredicting the actually realized, eventual spot values of said peso. They were regularly forecasting devaluations that did not happen.

    Rogoff’s solution, consistent with rational expectations, was to note that the distribution on peso movements was skewed to the downside, that it rarely was revalued upwards against the US dollar, but from time to time it would sharply devalued downwards. So, the ratex agent would be factoring in this small probability of this large devaluation, which would lower the forward market valuation of the peso. However, since these large devaluations only happened once in a while, one would regularly see the spot value of the peso doing better than what was forecast in the forward market.

  • sa

    interesting comment by barkley above. i have question for him-this seems similar to the volatility smile/skew that regularly appears in options prices of certain assets(like brazilian coffee/us equities after 1987);faced with such a price action how can one take advantage of this bias?

  • The bias Barkley points out perhaps explains one of the discrepancies I noticed between polling and markets, the prediction of future real estate prices. CME housing futures are trending down while the poll had more people predicting increased than decreased prices. It could be that this represents a consensus of a good probability of a modest price increase combined with a low but significant probability of a major crash. Or putting it another way, although more people think prices will go up than down, the ones who think they are going down expect a bigger change than the ones who think they are going up. So perhaps there is more consistency here than I thought.

    It’s frustrating to me that pollsters so seldom ask people to predict the future. Pollsters are more interested in how people feel about current affairs than in what they think will happen, it seems.

  • zzz

    sa, you’re right: option prices allow one to calculate the whole probability density of future prices, including the probability of sharp devaluations. But this is a rational prediction rather than a “bias”, hence there’s no way to take advantage of it.

  • Louise7

    Good news for you Melinda Doolittle fans! She’s set to release her debut album titled, Coming Back to You on Hi-Fi Recordings in January 09. Produced by Grammy-nominated producer Mike Mangini (Joss Stone, Run DMC, The Jonas Brothers), Coming Back to You was recorded with a team of live musicians in Nashville and New York City.

    The first single, “It’s Your Love” will be released to radio in October.

    “Keep yourself updated with Melinda’s career at melindasbackups. is a web site owned by Melinda Doolittle. Melinda’s Backups™ is a non-profit supportive association inspired by Melinda and authorized by her to operate this web site on behalf of her fans.