Back to Portfolio for the Future™

The Superstition of Mean Reversion: JPMorgan Edition

January 23, 2013

JPMorgan Chase has recently released a task force report (132 pages long) about the losses it suffered in its Synthetic Credit Portfolio in 2012, a portfolio managed by the firm’s Chief Investment Office (CIO). The portfolio in question consisted of both short and long positions in credit default swap indices and related instruments.

Through the first quarter of 2012 and well into the second Ina Drew, as the Chief Investment Officer, was responsible for the team trading in this portfolio. Bruno Iksil came to be the most notorious team member, receiving the nickname “the London Whale” because his positions dominated this specialized market.

The report has much to recommend it. But what I would propose as the one sentence takeaway is this: you can’t count on mean reversion. This is an old lesson, but one that bears repeating. Think of betting on black or red at the roulette table. If red comes up a lot, some people will be tempted to start putting all their chips on black. After all, it should come up black close to half the time in any given day, and as the day wears on with red outperforming, black is due … right?

The nice name for a person who thinks like that is “contrarian.” One not-so-nice name is “sucker.” The ball can keep landing on red until you’re out of money, and then can keep landing on red again until your relatives won’t lend you anymore.

Inside the Story

In this report, we get from the inside a story that is in its external aspects already quite familiar.

For example, we learn that on April 5, Drew emailed the JPMorgan Operating Committee to give its members a head’s up that on the following day both the Wall Street Journal and Bloomberg were going to run stories on the London Whale.

In response, the committee asked her for information and analyses. She complied, passing along in the process a comment by one of her traders that the mark-to-market prices in the SCP would mean revert. It’ll start coming up black, real soon.

The SCP was at this moment in a net long (risk) position. In other words, the CIO was selling more risk protection than it was buying, although it continued to hold short risk positions and jump-to-default protection. Drew acknowledged that the position was “not sized or managed well,” and said the losses to date were approximately $500 million, netting out to a loss of $350 million as a result of gains in other positions.

[The eventual losses from the Whale’s positions in fact totaled to roughly $5.8 billion.]

Skipping forward: on April 11, Drew forwarded to Jamie Dimon, Braunstein, and others an e-mail written by one of her traders. The anonymous trader said that yes, the SCP had not performed as well as expected, but that was because the present performance of the market “goes against all economic sense.” He blamed this in part on the press coverage. The thing to do, this trader concluded, was to wait out the abnormal market performance. This was a variation on the “mean reversion” argument, but only a slight one.

All this was background to what became an infamous first-quarter earnings call, on April 13. Jamie Dimon, Chief Executive Officer of JPMorgan, said that the fuss over the SCP was just a “tempest in a teapot.” He still seems to have believed that mean reversion was going to make it all subside. In the same call, Braunstein said that the firm was “very comfortable” with that portfolio’s positions.

Not So Comfortable

Two and a half years ago, Jared Woodward on the “Trading Markets” website wrote, “[If] a mean reversion trade goes bad, it can go very, very bad.” This was one of those times.

After that April 13th earnings call, JPM’s CIO losses steadily increased. Finally, in late April, a team of people from outside CIO finally looked over the portfolio, and reported that exposure was much greater than anything that office had yet reported to the board.

By May 10, Dimon had forgotten about the teapot and was describing the SCP strategy as “flawed, complex, poorly reviewed, poorly executed, and poorly monitored.”

One of the conclusions of the task force is that the risk limits that applied to the CIO were too general, not “sufficiently granular.” Specifically, there were no limits specific to risk factor or asset type. The absence of such limits “played a role in allowing the flawed trading strategies to proceed in the first quarter, especially as the positions grew in size.”

Maybe so but, in the end, there were just too many people around who believed in and relied upon the superstition of mean reversion.