Global Foreign Exchange (FX) Trading

London is the everlasting world capital of high finance, and Geneva is a hotbed of internationalization. With partners in both cities, a few of us who had lost patience with the lack of progress at Interactive Images left to form International Treasury Systems, based in Cambridge, Mass. ITS was initially conceived as a company to build Easel user interfaces for foreign exchange traders in the world's big banks; after about a year we switched to a rival touch-screen development environment, coding in plain C language.

In the first few months, we had to move all our software people to England for FX training, to rented space in a Rennies (British Tums) factory, with horribly cold restrooms, near Windsor. Soon we had a concept for a control panel to select currencies and counterparty banks, adjust prices and enter amounts. We took our prototype back to London for alpha testing in Morgan Grenfell, Australia-New Zealand (A-N-Zed) and other City banks.

Real-life FX traders turned out to be highly-strung overachievers accustomed to scrawling the details of billion-dollar deals on paper forms, and we saw back-office clerks coming up at the end of the day to ask what exactly did this or that trader mean to write. Fertile ground for our approach to an interface that captured all those details digitally in real time.

Developing code in the US to be tested in Europe caused some interesting communications problems. One night 3 of us had to sit up all night with an Ethernet/DataPhone hookup to shepherd less than a megabyte across the pond. Sometimes we'd go to the airport with a package of 5-1/4 inch floppy disks and find a London-bound passenger who was willing to carry them over and hand them to one of our UK people on arrival -- can't do that any more!

The scattering continued, with one development branch in Geneva and another in Hong Kong, the latter to serve our biggest customer, Hongkong and Shanghai Bank, who finally got tired of waiting for what our undersized staff could produce and bought our source code. HSBC put about ten times the personnel on it, upgraded the source from C to C++, and in a year announced a product.

Attempting to create a product which is nevertheless heavily customized for each client was not a winning strategy, and eventually it drank the money tank dry. Several of us MIT wise guys found ourselves unemployed for the first time in our lives.