Here's how 4 litigation financiers are thinking about data and analytics — and why some players think it's an industry prime for disruption

Business Insider interviews Legalist CEO Eva Shang about the use of analytics and artificial intelligence in litigation finance underwriting.

Jack Newsham
12/4/2020
Original Article

Artificial intelligence has been making inroads into law for years, from contract-review software that can help a lawyer quickly figure out which of her company's agreements lack an important provision to chatbots that can help normal people fight a parking ticket.

But one area where AI has not penetrated deeply has been litigation finance, a growing field where third-party funders invest in lawsuits.

"If you're going to watch Jeopardy, you should bet on IBM's Watson computer," said Will Marra, an investment manager at Validity Finance. "The AI is there. But it's not there yet for commercial litigation finance."

Commercial litigation finance is a global industry with dozens of players in the US alone who invested $2.3 billion in US-connected lawsuits in 2018, according to an estimate by Westfleet Advisors, a brokerage firm. At least two publicly traded funders, Burford Capital and Omni Bridgeway, have a US presence, and dozens of private companies and multistrategy investors including D.E. Shaw & Co and Fortress Capital Management also invest in litigation.

Turning disputes into data

From one perspective, there's no good reason why litigation finance should be impervious to digital disruption.

Federal courts and many state courts allow lawsuits to be filed online. Even cases that aren't immediately machine-readable are often tagged with metadata that a funder might be interested in, like the case type — maybe as simple as breach of contract, but also more bespoke, like confirmation of an arbitral award — and the dollar value of the claims at issue.

Some vendors have structured the data that courts and tribunals make public. Lex Machina, a legal analytics company owned by LexisNexis, is popular among litigation funders for getting data on a particular judge, law firm, or case type. Thomson Reuters advertises its Monitor Suite as a business-development tool that catalogs litigation outcomes, among other data points, and startups like ArbiLex also tout their ability to provide helpful data for niche categories of disputes.

A few funders, like Legalist and LexShares, are open about their use of datasets to find funding opportunities.

Legalist's use of analytics is an important part of its process for evaluating a large volume of middle-market funding opportunities, said Eva Sheng, its CEO. She said Legalist's 133 or so investments average about $500,000; by comparison, an average commitment in one of Burford's main portfolios is almost $11 million, based on numbers given in its 2019 annual report.

Jay Greenberg, the CEO of LexShares, said his company's "Diamond Mine" scans federal- and state-court databases for new complaints — it's ingested upwards of 1 million over the years — and automatically screens the text of the complaint and other lawsuit data for 17 parameters, like the ascertainability of the damages and the financial resources of the defendants. The system scores a case from zero to 6.25 diamonds and employees pitch the good ones, Greenberg said.

But commercial litigation finance firms are human-centric when it comes to the fundamental decision of whether to invest in, or underwrite, a dispute, and on what terms. Greenberg said it will be at least a decade before software can make that call. Shang avoided using the term "AI" in a recent interview, and said key parts of Legalist's decision to fund a case, like calculating potential damages and the ways they could be collected, are done by people.

"I think a lot of 'artificial intelligence' companies are using that term for marketing purposes," she said.

The missing piece: settlement values

One of the biggest challenges for any kind of automatic case evaluation, according to several people in the industry, is a lack of data on how cases are resolved. Most business disputes — even those rare ones that go to trial — are settled confidentially. Established litigation funders and law firms may have their own confidential settlement databases, and use that information to model opportunities, but trying to train software to make good underwriting recommendations using only public data is a challenge.

Maurice Power, the CEO of Apex Litigation Finance Ltd. in the UK, said his company has invested in over 30 cases so far and hopes to create artificial intelligence products that will enable it to scale up its underwriting abilities. He said Apex is talking to "various law firms and holders of settlement information" to try to obtain training data.

"The most valuable database to us is our own database of all the cases we are involved with," said David Perla, a co-chief operating officer at Burford Capital. "Even when we don't finance the case but we know what happened ...we put that into our model and we can model out likely damages, how a case is likely to resolve."

Databases can help people answer questions — how long do appeals take in such-and-such a court? How does a particular defendant tend to litigate cases? Does it tend to make lowball settlement offers or high ones? — but "there isn't enough data, nor is it normalized enough, to ever make an investment decision," Perla said.

Some companies developing case-evaluation tools have closed, like UK-based CourtQuant, and others do say that the technology can be hard to monetize.

Toby Unwin, an executive at Premonition.AI, said his company has case datasets that are superior to those offered by legal information giants like Thomson Reuters and LexisNexis. But he said there is a tendency for litigation funders to trust lawyers with "steely gravitas and just the right amount of gray hair" instead of looking for opportunities in data.

"I'm not saying they pick crap cases, just that there are better cases out there that they could have access to that they're just not seeing," Unwin said.