”But what if you found out that key data underlying that breakthrough were actually wrong?”

Academic research is fun-ish. A problem with making money from studies is that by the time it has been released, the likelihood of the “anomaly” still existing in a form and size that can be captured approaches net zero. Simple world – there are a lot of smart people with a lot of money to put to work and they do an excellent job of neatly ruining a quiet pocket of opportunity. If you know something, get your position first!

But this is juicy because it highlights a lot of social science which references the above paragraph. It is often hard to duplicate social science research as a result. And then there is simply fraud for personal achievement.

One further note. Anomalies can exist for longer than you can stay liquid, like those that don’t have a catalyst to close within one bonus year. Those are interesting if you have the right money and the right time horizon. Longer-dated warrants and options. Value investing.

Credit Quants Reeling After Dodgy Data Found in Seminal Research

Professors retract seminal finance paper after detective work uncovers flaws
By Justina Lee

What if researchers discovered a scientifically sound formula for predicting returns in the bond market? If you were a number-crunching investor known on Wall Street as a quant, it could validate your approach and offer clues to new trading strategies. But what if you later found out that key data underlying that breakthrough were actually wrong?

That’s the unsettling prospect facing the quantitative investing world now, thanks to the detective work of a then-28-year-old doctoral student named Alex Dickerson. While working on his degree at Warwick Business School in England in 2021, he tried to understand established thinking about bonds by replicating the field’s seminal research. But he found he couldn’t do that with an influential paper by three Georgetown University professors.

Dickerson’s results were so different from those of the original paper, which has been cited more than 200 times, that he assumed he must be making a mistake. But when he began to ask around, he found a McGill University doctoral student, now working at Morgan Stanley, who’d run into similar issues. In April he posted a rebuttal paper with two co-authors that left the worlds of academia and high finance reeling. The prestigious Journal of Financial Economics (JFE) retracted the original paper—the only time it’s ever done so—at the request of the authors, who also asked Georgetown to initiate a formal review.

“When you read a paper that’s published in one of these journals, you assume everything in there is true,” says Dickerson, who finished his doctorate and is now a lecturer at the University of New South Wales in Sydney. “The point of academia is someone publishes a paper as a baseline of the literature and you as a student or a scholar take that work and build on it. Now if the foundation of that work is flawed in some way, there’s nothing to build on.”

The controversy centers on what are known in the investing world as factors: characteristics of securities that are believed to predict higher long-run returns. In equities, these include value (a low price relative to a company’s fundamentals), quality (profitability and low debt) and momentum (recent price trends). Factors have been documented by decades of academic research, including by Nobel laureates.

Although trillions of dollars follow these widely known-methods in the stock market, their use in fixed income has historically been frustrated by a mix of poor data, old-school trading methods and the outright size and complexity of the bond market. Now a boom in electronic trading of debt is creating more data and enhancing liquidity, and quants are racing to identify the factors that sway bond performance.

That’s a task many believed the 2019 Georgetown paper had achieved, giving the factor hunt an academic stamp of approval. It identified factors based on a bond’s downside risk, illiquidity risk and credit risk. Then Dickerson and co-authors Philippe Mueller and Cesare Robotti—two professors at Warwick—exposed some remarkable fundamental errors. They zeroed in on two major issues with the paper, which was dubbed BBW after its authors, Jennie Bai, Turan Bali and Quan Wen.

First, BBW got the calendar wrong. In the case of two factors, there was a lead error, meaning the authors gave the returns of, say, February as the returns of January. For the third factor, there was the opposite problem of a lag error. Once these mistakes—made for some periods of the data but not all—were corrected, the factors behaved a lot like one another and, worse still, similar to a simple bond benchmark. In other words, they provided no investment advantage at all.

Second, BBW appeared to have removed the most extreme losses for some bonds, making the factors look less volatile than they were. In their retraction notice, Bai, Bali and Wen acknowledged that an error of “temporal misalignment” voided the results. “In academia, we care about integrity more than anything else,” Bai wrote in an email to Bloomberg News on behalf of herself and her co-authors. She wrote that they’ve requested a formal review at Georgetown to clarify whether the errors resulted from “unintentional mistakes or misconduct.” Teresa Mannix, a university spokesperson, says the authors decline to comment on the trimming of extreme returns.

Before its repudiation, BBW had become the basis for multiple other papers on bonds. JFE editor-in-chief Toni Whited says the editors are trying to determine if more papers will have to be retracted as a result of BBW’s flaws. “That’s a big deal because it’s been cited a million times,” she says. “It’s not some tiny little coding error that fixes four numbers in Table 5.”

Read the full article at Bloomberglaw.com.