Bettors at the prediction market Intrade are quite confident that the Supreme Court will declare the individual mandate unconstitutional when it rules on the government’s health care bill. (The decision is expected to come on Thursday.) At this writing, the consensus prediction of the market is that there is a 74 percent chance that the court will vote to overturn the mandate.
This represents a reversal from earlier, when the conventional wisdom held that the Supreme Court was more likely than not to uphold the mandate.
That was before oral arguments were made in the case in March, however. These were evidently perceived to be a “game changer.” The government’s lawyer, Solicitor General Donald Verrilli Jr., sometimes struggled in offering a crisp defense of the law, and the questions of him from the court were often skeptical.

After these oral arguments, Intrade’s betting shifted dramatically. Quickly, the chance of the mandate’s being overturned climbed to about 65 percent from 35 percent at the market.
The odds of the mandate’s being overturned have fluctuated some since then, but have risen even further in the past few weeks, according to the market.
Is such a large shift in sentiment justified? In my view, probably not. This may be another case of traders being overconfident about the value of their information, a property which has also been observed in the stock market.
To be clear, the evidence is that oral arguments do have some predictive power in forecasting Supreme Court decisions, according to a variety of empirical analyses from academic and legal scholars. It’s just a question of how much power they have, and how unambiguous the evidence about oral arguments is in any particular case.
These studies found that the side being asked more questions from the bench during oral arguments is more likely to lose the case, since this may indicate skepticism from the court about its position.
Which side was that in the health care case? Technically, it was the lawyers representing the State of Florida, who were arguing that the law was unconstitutional. But really, it was more of a tie. Some 54 percent of the words from the bench were directed at the lawyers arguing against the mandate, while 46 percent were directed at Mr. Verrilli, who was defending its constitutionality. That close balance in questioning might seem to argue against a substantial change in the odds either way.
However, there were some troubling signs for the government when it came to the distribution of questions from individual justices. In particular, four of the eight justices asking questions during the hearing directed more questions to the government, including potential swing votes Chief Justice John G. Roberts Jr. and Justice Anthony M. Kennedy. If joined by Justice Clarence Thomas, who did not ask questions during oral arguments but is thought to be almost certain to vote to overturn the mandate, they would constitute a 5-to-4 majority to overturn the mandate.
The academic studies find that the number of questions posed by individual justices also has some predictive power in determining how they will vote, though this must be balanced against the number posed by the court as a whole. The justices’ behavior is in part strategic: they compete and coordinate with one another when asking questions (and to some extent also in issuing their rulings). So the tendencies of the individual justices and the behavior of the court in the aggregate should be considered.
But Justice Kennedy, a crucial swing vote in the health care case and in many others, has been found by these studies to be difficult to read. The number of questions he asks each side has historically not had that much predictive power in determining his vote. (Justice Kennedy makes matters harder because he tends not to ask many questions during oral argumentation and asked relatively few during the health care hearing.)
Although these question counts shift the odds, they do not do so by all that much. The study by the legal scholars Lee Epstein, William M. Landes and Richard A. Posner found that when the petitioner was asked more questions by the justices, his odds of winning the case dropped to 53 percent from 62 percent. When the petitioner was asked fewer questions, his odds of victory rose to 71 percent from 62 percent.
These are meaningful differences: a shift of about 10 percent in the odds in one direction or another after oral arguments. But they are not in line with the much larger shift, about 30 percent, that occurred on Intrade.
One need not look that far for cases in which the question counts proved to be an unreliable guide. In the Arizona immigration case, which the court ruled upon on Monday, the bulk of questioning during oral argument went against the government, but the court ruled in the government’s favor on most counts.
What about all the “intangible” factors that pundits and legal experts have cited as being important – everything from Mr. Verrilli’s coughing to Justice Ruth Bader Ginsburg’s comments about broccoli?
These indicators are probably of little value. Instead, studies have found that predictions made by “expert” commentators on the Supreme Court do barely any better than a coin flip and are beaten by the statistical methods (a finding that follows the poor overall track record of experts in making predictions under many other circumstances). These experts are irrationally confident about their ability to read the tea leaves, and their predictions suffer for it.
There are, however, some statistical indicators that might be of predictive value above and beyond the number of questions during oral arguments. For instance, as Mr. Epstein’s study noted, the petitioner wins about 60 percent of cases before the Supreme Court. In this case, the petitioner is Mr. Verrilli and the federal government, so that would favor the mandate being upheld.
In addition, Mr. Epstein’s study found, the petitioner does especially well when it is the federal government, as it is in this instance. Historically, the federal government has prevailed in about three-quarters of such cases.
Then again, statistical systems agree with the conventional wisdom about one thing: the Supreme Court has become extremely
conservative. These ideology scores have been found to have predictive power and would be a point in favor of the law being overturned.
There is also a small but nonzero chance that the court will decide not to rule on the mandate. That would count as a “win” for those betting against it being overturned on Intrade, although the court could revisit the case later.
In short, the conventional wisdom may have gotten ahead of itself. The more tangible factors, in my view, roughly balance one another out rather than favoring one or another side by a 3-to-1 margin.
Most of the rest is grasping at straws; the Supreme Court just isn’t so easy to read. As Justice Ginsburg advised: “Those who know don’t talk. And those who talk don’t know.”