It would be worth a million bucks to a trader to be able to price securities in minutes while the competition takes hours.
Now International Business Machines Corp. says it is offering just such a pricing system, and sure enough, it will cost users almost a million.
Big Blue says its mathematicians have developed a way to do pricing calculations that makes Monte Carlo simulations look like they are standing still.
Vast improvements in the speed of pricing would presumably reduce both risk and price spreads in the market. But for now, traders do not appear to be in a rush to get in on IBM's good thing.
The Monte Carlo method is the present standard for pricing techniques. It's so called because it is the computer equivalent of rolling the dice or spinning a roulette wheel repeatedly to generate a more or less random series of outcomes within a defined range - like the 36 possible combinations on a pair of dice or the 37 or 38 numbers on a roulette wheel.
But even using the computer, rolling and spinning are time-consuming when large numbers of variables are involved, as in mortgage securities. So they must either be done on very large computers or take long hours of running time on smaller ones.
Enter IBM with the product it calls the Deterministic Simulation Blaster. The Blaster uses a simulation technique that runs far faster than the Monte Carlo simulation and with more accuracy, IBM says. The simulation technique is not new but had previously been cumbersome and now has been much refined, it added.
Robert Howe, general manager of IBM's banking, finance, and securities business unit, says the product is "one of the most significant breakthroughs in trading analytics in many years."
He said it would allow users to make more and better trades.
Some Wall Street people appeared skeptical about the IBM offering. Anand Bhattacharya, managing director in charge of mortgage and asset-backed finance at Prudential Securities, New York, said his company was already doing calculations of option-adjusted spreads using Monte Carlo simulations on a large computer and could do pricing very quickly.
"The long and short of it is that we have never had (speed of calculations) as an issue. If the technology were offered to me, I would say, it's wonderful, thanks but no thanks."