The Supreme Court recently sided with the Consumer Finance Protection Bureau (CFPB) in a case that many presumed would upend a major agency if it had gone the other way. I wrote an article a few months ago about how the Supreme Court is paying a lot of attention to 5^th circuit decisions. The outcome of CFPB case was one where my intuition was validated. The CFPB case was not the only 5^th Circuit decision where aspects of an administrative agency are under review. Another pivotal case where a an agency and its decisions are under a magnifying lens is in SEC v Jarkesy. Supreme Court oral arguments in the case concluded on 11/29/2023.

In a prior post, I examined the oral argument in Grants Pass with an analytical platform I developed with a collaborator called Optimized Legal Audio (OLA) that looks at predictive indicators of oral arguments to identify a judge’s leanings. We have been testing this technology on several dozen cases in both state and federal judiciaries from the Supreme Court on down where there is accessible audio. We have learned where the limitations are on our platform as well as where the strengths lie. We recently tried to read the tea leaves on the audio footprint of the SEC v. Jarkesy case. We had the opportunity to test it and I wanted to share the results.

This platform is not omniscient as its predictive capacity depends on a spirited debate between the judges and each attorney, barring which, we miss the ball on our predictions. For example, I wrote about how Justice Thomas trails all other justices in words spoken in this article, which for our purposes makes him a difficult justice to predict based on his oral argument behavior alone. Chief Justice Roberts’ votes are also sometimes hard to predict. In Warner Chappelle Music v Nealy he was almost silent – he only spoke twice, which is well below his usual speaking averages. The transcript of those two occurrences when Justice Roberts spoke to the attorneys show him saying: “Anything further? Thank you, Counsel, rebuttal Mister Shanmugam.” and “Thank you, Counsel Justice Thomas Justice Sotomayor, Kagan Justice Gorsuch. Anything further? Justice Kavanaugh? Thank you, Counsel. Thank you, Mister Earnhardt.” To add to all of that, Justice Kavanaugh says nothing at all during oral arguments. No platform that analyzes audio can make meaningful conclusions in such an environment and based on this, we decided that we could not make a meaningful prediction about the outcome in Warner Chappelle Music based on oral arguments.

Coming back to Jarkesy, there all Justices spoke a fair amount to each arguing attorney. Based on these features, we looked for signals of potential votes from the oral arguments. The audio, the transcript, and the delivery altogether had very interesting leading indicators of decision-making. Below is our prediction which is presented with the caveat  that the accuracy depends on how well we can measure the justices’ speech, so that if a justice speaks a bit to each attorney we should be able to generate a more accurate prediction.

You can see a sample of the inner workings of our platform here. For example, it spikes at the precise turning point of sorts for Justice Kagan when questioning Respondent’s attorney S. Michael McColloch.

The spike correlates with the point in the argument where Justice Kagan says: “we’re pretty close because I think that just resolves the case…That’s the issue… I mean, that’s the issue. That’s the results…The seventh amendment is no bar.”

Justice Barrett’s turning point, appears in the instance below:

That instance relates to Justice Barrett’s comment: “So I, I guess I just don’t understand the logic here but for a different reason than Justice Kagan saying at least.”

These are just two of thousands of interwoven data points that our platform analyzes to create a leaning of the vote. After all that was said and done, the platform does offer a potential conclusion.

Our platform predicts a 6-3 decision for Petitioner Jarkesy and against the SEC. The level of intensity from each of the justices was high. The involvement and debate seemed very fitting of the magnitude of the decision.  Since our predictions are not only binary but come with a relative degree of certainty based on the justices’ speech, our platform even provided that Justice Jackson may have been on the fence to make the vote a 7-2 this case (although this is less likely than a 6-3 outcome). This should be taken with a grain of salt as well since one should not assume every federal agency case would be decided like this – after all, rewinding time by just one week, we see how the CFPB prevailed. Now we get to sit and await the official decision in SEC v Jarkesy, but now we can measure our individual conceptions of how we thought the arguments went with a statistical prediction of a possible outcome.