By Steven Bertoni and Antoine Gara
Daniel Nadler, the CEO and founder of artificial intelligence start-up
The scene, like the name Kensho (to see nature) is intentionally Zen: Nadler wears prayer beads on each wrist, drinks green tea from a cast iron Japanese pot and eats oranges, precisely sliced, from black earthenware bowl. There is a Buddha on one wall. Yet, all this meditative calm betrays the frantic pace in which Nadler has grown and evolved the tech start-up.
Nadler launched Kensho five years ago out of the kitchenette of venture firm, General Catalyst, in Cambridge while simultaneously finishing a PhD in economics at Harvard. Fast forward to today and Nadler is selling the company (now 120 plus employees strong and profitable) to
That the biggest A.I. deal comes out of Wall Street (from a 158-year institution no less) and not Silicon Valley, shows just how much the future of finance might rely on bots over brokers. S&P is the world's biggest ratings agency, tracing back to the analysis of railroad bonds prior to the Civil War. Its obsession with statistics in finance led to the S&P 500 Index decades ago. Some $12 trillion in assets are now benchmarked to S&P's indices; its data covers 99% of company values globally. Now, A.I. is the new frontier in financial information and to own Kensho, a first mover in plying machine learning techniques to financial questions, S&P is paying a big pricetag. At $550 million, the deal is larger than Google's reported purchase of DeepMind Technologies, Intel's acquisition of Nervana Systems, in addition to A.I. buys from Apple and Twitter.
Nadler’s initial plan for Kensho: Use machine learning to make complex financial analysis as easy as a search on Google. With his cofounder Peter Kruskall, he built an algorithm dubbed Warren (after Warren Buffett) that could pour through millions of market data points to find correlations and arbitrage opportunities. The tech was complicated, but the interface was simple--a text box that let users ask complex questions in plain English. As FORBES wrote in May 2014, “Out of the box Warren can find answers to more than 65 million question combinations in an instant by scanning more than 90,000 actions such as drug approvals, economic reports, monetary policy changes, and political events and their impact on nearly every financial asset on the planet.”
With the aim of selling the powerful analytical tool to Wall Street traders, Kensho raised $10 million from General Catalyst, Accel Partners, Breyer Capital, NEA, Google Ventures and others in 2013. A partnership with
Wall Street was tough to break into (no highly paid CTO would opt to have his or her expensive proprietary tech systems ripped and replaced by a start-up). And Kensho's mission to democratize quantitative analysis went against a culture that valued secrecy and exclusivity. Says Nadler, "To be totally blunt, I thought the company was going to fail." As he pitched Kensho's innovative technology, "There were probably a hundred 'no's' in the first 18-months."
A big break came in the summer of 2014 when Nadler found an advocate in Martin Chavez, then the chief information officer of Goldman Sachs. Chavez led a change in mindset at the blue chip investment bank. In order to maximize efficiency and value for clients, Goldman didn't need to build or own all of its technology. It could partner with nimble upstarts like Kensho to accomplish a big breakthrough for traders and clients alike - and with less risk and cost.
According to Nadler, Chavez's attitude was, "I don't care if it's built here, I don't care if it's bought, or I don't care if it's rented. The primary objective is leveraging new technology towards increased efficiency and increased operating leverage." Over a number of months Chavez and his team worked to refine Kensho's technology for Goldman's traders. By Thanksgiving, Goldman became its largest investor by participating in a $15 million funding round. Other megabanks soon followed: JPMorgan, BofA and Morgan Stanley. A year ago, S&P Global led a final funding round at a valuation over $500 million.
"Daniel Nadler was prescient in seeing that A.I. -- specifically the application of machine learning to large, newly available financial data sets -- could yield amazing insights for investors", says Larry Bohn, managing director, General Catalyst, which hatched Kensho in its Cambridge, MA offices. "Prior to this, only analysts at quant funds could keep pace with the velocity of information and the custom programming required to achieve significant market insights. Basically, he helped democratize the approach through a hugely innovative software platform and made it available to banks and the broader market."
Forbes has been following Kensho's A.I. for investors for years. Initially, we dubbed Kensho as Google style search for stockpicking in our May 2014 magazine issue. Since the inception of our annual Fintech 50 list, Kensho has been a stalwart at the forefront of using machine learning techniques to supplement investment analysis. And when big market gyrations have occurred, Kensho's instant analysis has proven illuminating.
Take Brexit, Kensho's data archive produced actionable context to the surprise populist vote in seconds. It predicted any short-term recovery in the British Pound would be illusory because history shows populist votes lead to extended local currency selloffs. That's exactly what happened in the days and months after Brexit. The pound plunged to three-decade lows in July, sinking to $1.28 versus the dollar, before rallying slightly to $1.33. The currency then continued its plunge to as low as $1.22.
When populism crossed the Atlantic to the U.S., Kensho again offered compelling insight. After 3% post-election rally in the U.S. dollar history showed a selloff was in store. In early 2017 Kensho predicted that companies hit early by dollar rally - tech stocks -- would soon outperform. And that's what happened, it turns out tech was the big winner of 2017 and no surprise the U.S. dollar was a big loser.
What is Kensho now saying about President Trump's decision to implement tariffs to steel and aluminum imports?
Don't expect outperformance from the Russell 2000 Index despite some commentators' belief that domestic small caps should win in a protectionist environment. In fact, it is the Nasdaq, filled with global tech conglomerates, which traditionally outperforms in the three-months after a major protectionist move. Kensho found this in seconds by studying 14 tariff announcements since 1988.
Over time, Kensho's relationship with Goldman proved to be more than just a door opener on Wall Street. Chavez and the firm's Principal Strategic Investments team helped Nadler impart a big pivot at the company -- instead of focusing solely on trading (a high-profile yet relatively small piece of big banks) he turned his natural language algorithms to the entire firm.
Kensho's technology is now used in Goldman's $1.5 trillion asset management division, where a custom built "cross correlation engine" tracks investors' correlations between various assets. Kensho now also spits out a Sunday night report called the "Kensho Weekly," which offers A.I. driven contextual analysis of news flow, earning reports and analyst ratings changes. The open architecture of Kensho allows users to tweak variables as they like.
"What made Kensho very well received by the Goldman Sachs population was it made the user more powerful; it never disrupted the underlying business model,” says Rana Yared, a Goldman managing director who oversaw the Kensho investment. “Goldman Sachs professionals still have to come up with our own analysis, it just makes us more powerful.” Adds Nadler, "Artificial intelligence is a misnomer. It's accelerated intelligence... It's about doing very human things, which historically you thought were impossible for a human being to do, at a blinding speed."
Chavez, now Goldman's chief financial officer, further recommended Nadler expand the platform well beyond investing and trading. After all, Goldman's bankers could use Kensho's big historical database of information and machine learning capabilities to research companies under their coverage far quicker. So too could its corporate clients. That's the key to Tuesday's deal with S&P.
With the sale to S&P, Kensho's technology will integrate with one the biggest data cogs on Wall Street, used by all of America's biggest companies. Soon its A.I. may be ubiquitous for bond analysts and corporate finance departments alike, not to mention a far broader swath of the investing public.
"In just a short amount of time, Kensho's intuitive platforms, sophisticated algorithms, and machine learning capabilities have established a wide following throughout Wall Street and the technology world," Douglas Peterson, CEO of S&P Global said in a startement. "Via this acquisition, S&P Global is demonstrating a strong commitment to not just participating in the fintech evolution, but leading it."
For Kensho, the deal with S&P is structured in a way that Nadler believes will help accelerate growth. Kensho will remain a distinct entity, its brainy headquarters will remain in Cambridge, and it is retaining its Buddhist-inspired name. But the scale of S&P's relationships and its sale force can now push Kensho's technology to a whole new group of users without big cost.
And by teaming with S&P --one of the big financial data organizations-- Kensho's A.I. gets a whole new repository of information to parse over. Says Nadler, "We now have an incredible moat because overnight we have the global scale with every financial user in the world, and our system becomes exponentially smarter as a result."
For more:
