Negative Latency – Breaking the Zero Barrieron Apr 01, 2010 in Real-time web by Alex DeLarge
Every second counts - or in financial markets, we’re usually talking milliseconds or even microseconds. The faster you can get a price to a client the better.
We have been very busy recently on R&D and are now close to releasing the Agatha negative latency trading platform.
Using advanced predictive algorithms we are able to send price changes to clients in advance of all other trading platforms.
This technology effectively breaks the zero latency barrier by predicting future events. The Agatha platform allows users to “pretrade” leading to more profitable arbitrage strategies.
Agatha is self tuning. The more confident the price prediction the earlier it will be sent out. There is a danger that when prices are predicted too early that it can actually have a knock on effect on future events.
Ongoing testing shows predicted prices are accurate 99.8% of the time, and the minority of prices that are inaccurate are reported back into the system acting as a feedback loop which allows the self tuning to adjust how early prices are sent.
Over the years we have carried out benchmarks on Liberator that focus on latency, sending high update rates to large numbers of clients in the shortest possible time. We have always been proud of our ‘close to zero’ latency figures.
The problem is that the margins of improvements are getting smaller and more and more people are in the low latency game now. Customers are demanding more and we had to make a game changing move.
Vast sums of money are made, and lost, based on latency of information. In this article by InformationWeek it was estimated that a 1 millisecond advantage in trading applications can be worth $100 million a year to a major brokerage firm.
Think of the effect negative latency could have on profits!