First Person: Jeremy Mabbitt

©Toby Coulson

Jeremy Mabbitt’s software helped create the complex investments that fuelled the banking crisis

Whenever I watch those movies and documentaries about the credit crunch, and the endless debates about who or what caused it, I’m amazed that no one ever mentions Galapagos. Galapagos was the software used by most investment banks to create the complex investment products known as collateralised debt obligations. CDOs are what popped the credit bubble.

I developed Galapagos initially in my back room in Brighton. The name was chosen because my software was inspired by evolution in computer science. Galapagos used artificial intelligence algorithms, in particular algorithms based on evolution in nature, to structure credit portfolios by a process of intelligent search, to generate “better” CDOs.

I’ve always been a bit of a computer geek and studied evolutionary and adaptive systems as a postgraduate at the University of Sussex.

In 2003, I found a few like-minded people to join me and we visited businesses, from computer gaming companies to enterprises within the biological science space, to find out if they could use this type of software. But it was the banking industry that was most interested. At the time we didn’t know anything about CDOs, but, after talking to a Canadian investment bank, we started to design software to their specifications. Then we started visiting other banks, mentioning that we were working on solutions for their competitors. It was essentially a game of bluff.

Once the first few banks started using Galapagos, most of them followed. Some senior structuring staff would only move banks on condition they were bought a Galapagos licence. One team got so frustrated with the delay in getting the sign-off to buy the software that they offered to pay for it out of their own bonuses.

Galapagos exploited any weaknesses, or even on occasion outright errors in the rating agencies’ models, in order to make more money for the arranging banks. But the rating agencies were willing partners – they licensed us to use their models in this way, seeing it as a revenue-generating opportunity.

But in early 2007 the credit crunch started brewing. There was a strange atmosphere – almost like tumbleweed blowing by. People didn’t want to commit to deals or make any decisions. However in August 2008 we signed a deal with Lehman Brothers. It was our most lucrative project to date, but in September Lehman went bust and everything went into panic mode.

Eventually we had to sell up. I don’t regret not making money, although I’m frustrated that the rug was pulled from under us. The problem was that the banks were structuring products that were poised for downgrades. The banks liked to ramp up the complexity of their deals, so they could stay ahead of the investors and rating agencies, and maximise their margins. It was complexity for complexity’s sake.

Mostly I blame the firms that invested in these things – the fund managers, insurance companies and banks. They were keen to invest in rated credit products that paid as high a yield as possible for a given rating. And they have got off lightly. They used people’s money and never took a proper look at what they were investing in. If you’re dumb enough to invest in AAA-rated CDOs without knowing what you’re investing in, you’re bound to get into trouble. It’s like buying a car and demanding certain specifications as cheaply as possible. The car won’t last.

My experience with Galapagos proved to me that the underlying approach of evolutionary computing is a powerful one. I have realised that beyond synthetic CDOs, the ordinary investments that people hold are also highly complex. I believe there’s potentially a far bigger opportunity here, to provide a service to ordinary investors, enabling them to make the best possible decisions while avoiding hefty fees. So that’s what my new business does. It’s called Vidivici – from the Latin – I saw, I conquered. The name has nothing to do with my past experiences.