Bayesian Investor Blog

How well did the subsidies that I provided for some Intrade contracts promote price efficiency?
(more…)

A number of people have compared the final forecasts for the election (e.g. this), but I’m more interested in longer term forecasting, so I’m comparing the state-by-state predictions of Intrade and FiveThirtyEight on the dates for which I saved FiveThirtyEight data a month or more before the election.

Here is a table of states where Intrade disagreed with FiveThirtyEight on one of the first four dates for which I saved FiveThirtyEight data or where they were both wrong on July 24. The numbers are probability of a Democrat winning the state’s electoral votes, with the Intrade forecast first and the FiveThirtyEight forecast second.

On July 24, both sites predicted Florida, Indiana, and North Carolina wrong. FiveThirtyEight got Indiana right on Oct 1 when Intrade was still wrong, but Intrade got North Carolina right on that date (just barely) while FiveThirtyEight rated it a toss-up.
Intrade got Nevada right on July 24 (just barely) while FiveThirtyEight got it wrong (just barely).
For Virginia, Intrade was right in July and August while FiveThirtyEight was undecided and then wrong.
FiveThirtyEight got Colorado wrong on September 14, but Intrade didn’t.
FiveThirtyEight got Ohio wrong on August 22, while Intrade got it right.
Intrade rated Missouri a toss-up on July 24, while FiveThirtyEight got it right.

On September 14, FiveThirtyEight was fooled by McCain’s post convention bounce by a larger margin than Intrade, but by Oct 1 FiveThirtyEight was more confident about correcting those errors.
For states that were not closely contested, there were numerous examples where Intrade prices where closer to 50 than FiveThirtyEight. It’s likely that this represents long-shot bias on Intrade.

In sum, Intrade made slightly better forecasts for the closely contested states through at least mid September, but after that FiveThirtyEight was at least as good and more decisive. Except for Intrade’s Missouri forecast on July 24, the errors seem largely due to underestimating the effects of economic problems - errors which were also widespread in most forecasts for other things affected by the recession.

In the senate races, I didn’t save FiveThirtyEight forecasts from before November 1. It looks like both Intrade and FiveThirtyEight made similar errors on the Alaska and Minnesota races.

On the other hand, Intrade had been fairly consistently (but not confidently) saying since early July that California’s Proposition 8 (banning same-sex marriage) would be defeated. Pollsters as a group did a somewhat better job there by issuing conflicting reports.

I’ve made a change to the software which should fix the bug uncovered last weekend.
I’ve restored about half of the liquidity I was providing before last weekend. I believe I can continue to provide the current level of liquidity for at least a few more months unless prices change more than I currently anticipate. I may readjust the amount of liquidity provided in a month or two to increase the chances that I can continue to provide a moderate amount of liquidity until all contracts expire without adding more money to the account.
I’m not making new software public now. I anticipate doing so before the end of November.

Many people seem to be reacting to the recent stock market crash the way they wish they had to the 1987 crash, and a smaller number are comparing it to 1929.
The unusual resemblance to the crash of 1937 makes me expect something in between those two scenarios.

• The 1937 crash was caused in part by a sudden increase in caution by banks after the Fed significantly increased their reserve requirement. Banks played no interesting role in the 1929 or 1987 crashes.
• The 1929 and 1987 crashes followed stock market peaks in August, versus March and the prior October for the 1937 and 2008 crashes.
• The 1937 and 2008 crashes both came eight years after one of history’s largest stock market bubbles.
• The 1929 and 1987 crashes followed an increase in the discount rate to 6 percent. The 1937 and 2008 crashes followed decreases in the discount rate to 1 and 2.25 percent.

All four crashes happened mainly in October and their behavior in that month provides little reason for distinguishing them.
If the 1937 crash is a good model for what to expect in our near future, many investors who are currently following the lesson they learned from the 1987 crash will discover in early 2009 that the unexpectedly severe recession casts doubt on the belief that crashes create good buying opportunities. How many of them will stick to their buy and hold commitment then (when I expect it will be a good idea)?
When the extent of the recession becomes disturbing, remember Brad DeLong’s perspective:

Is 2008 Our 1929? No. It is not. The most important reason it is not is that Bernanke and Paulson are both focused like laser beams on not making the same mistakes as were made in 1929….
They want to make their own, original, mistakes..

(HT James Hamilton).

Last night an Intrade trader found and exploited a bug in my Automated Market Maker, manipulating DEM.PRES-TROOPS.IRAQ until Intrade rejected one of the market maker’s orders for lack of credit and the software shut down.
The bug involves handling of partial executions of orders, and doesn’t appear to be easily fixable (what happened looks nearly identical to the scenarios I had analyzed and thought I had guarded against).
For the moment, I’ve reduced the market maker’s order size to one contract, which will prevent further exploitation but provide much less liquidity.
I will try to fix the bug sometime in November and increase the order size (on the contracts that don’t get expired at election time) by as much as I can without adding more money to the market maker’s account. I will also analyze the information provided by the markets shortly after the election.

Book review: The Misbehavior of Markets: A Fractal View of Risk, Ruin & Reward by Benoit Mandelbrot.
Mandelbrot describes some problems with financial models that are designed to provide approximations of things that can’t be perfectly modeled. He pretends that pointing out the dangers of relying too much on imperfect approximations shows some brilliant insight. But mostly he’s just translating ideas that are understood by many experts into language that can be understood by laymen who are unlikely to get much value out of studying those ideas.
His list of “ten heresies” is arrogantly misnamed. Sure, there are some prestigious people whose overconfidence in financial models leads them to beliefs that are different from his “heresies”, but those “heresies” are closer to orthodoxies than they are to heresies.
His denial of the equity premium puzzle is fairly heretical, but his argument there is fairly cryptic, and relies on suspicious and poorly specified claims about risk.
He says market timing works, but the strategy he vaguely hints at requires faster reaction times than are likely to be achieved by the kind of investor this book seems aimed at.
His use of fractals doesn’t have any apparent value.
Mandelbrot is primarily a mathematician with limited interest in understanding how markets work. One clear example is his mention of a time when Magellan “was still a small fund, too small for any detractors to argue that its size alone gave it a competitive edge”. Any informed person should know that’s completely backward - larger funds have a clear disadvantage because they are limited to trading the most liquid investments.
Another example of a careless mistake is when he claims the evidence suggests basketball players have hot streaks, seemingly unaware that Tversky and others have largely debunked that idea.

The stock market reacted to today’s defeat of the bank bailout bill with an unusually big decline. Yet the news wasn’t much of a surprise to people watching Intrade, whose contract BAILOUT.APPROVE.SEP08 was trading around 20% all morning. Why did the stock market act as if it was a big surprise?
Did Intrade traders make a lucky guess not based on adequate evidence? Did they have evidence that the stock market ignored? Could the stock market have priced in an 80% chance of the bill being defeated (if so, that would seem to imply that passage would have caused the biggest one-day rise in history)? Could the stock market have been reacting to other news which just happened to coincide with the House vote? (It looks like the market had a short-lived jump coinciding with news that House leaders hoped to twist enough arms to reverse the vote, but I wasn’t able to watch the timing carefully because I was at the dentist).

It seems like one of these must be true, but each once seems improbable.

Arnold Kling, whose comments on the bailout have been better than most, was surprised that the bill failed.

I covered a few of my S&P 500 futures short positions at near the end of trading, but I’m still positioned quite cautiously (I made a small profit today).

Charlie Munger in the August 31, 2008 issue of Outstanding Investor Digest:

Let’s say you’re insuring against the outcome that people will lose money on a $100 million bond issue, and the credit default swaps, instead of amounting to $100 million, amount to $3 billion. Now you’ve got people with $3 billion worth of contracts that really have a big incentive in having somebody fail. And they may manipulate in some fraudulent or extreme way to cause a default in order to make the big collection.

There doesn’t seem to be enough transparency in financial systems to figure out whether this concern is relevant to this week’s panic.

To deter any suspicion that the comparisons I plan to make between Intrade’s predictions and polls are comparisons I selected to make Intrade look good, I’m announcing now that I intend to use FiveThirtyEight.com as the primary poll aggregator. I intend to pay attention to predictions that are more long-term than I focused in 2004, so the comparison I’ll attach the most importance to will be based on the first snapshot I took of FiveThirtyEight.com’s state by state projections, which was on July 24.

Also, as of last week, one of the Presidential Decision Markets that I’m subsidizing, DEM.PRES-OIL.FUTURES, has attracted enough trading (I suspect from one large trader) to make me reasonably confident that it’s showing the effects of trader opinion rather than the effects of my automated market maker (saying that oil futures will drop if the Democratic candidate wins, and rise if he loses).

News reports plus the pattern of crude oil fluctuations indicate that the large price increases around May and June were due mainly to Chinese desperation to guarantee a larger than normal margin of safety during the Olympics, not manipulation (although the results bear a good deal of resemblance to the results of manipulation).