Saturday, January 31, 2009

Boyd and the Trading Software Application

From "Boyd: The Fighter Pilot Who Changed the Art of War" by robert Coram:

"Colonel Boyd's development of the theory of maneuver warfare began, not with ground battles, but with a study of some mock air-to-air combat exercises {conducted at Nellis Air Force Base in 1974}that led him back to the study of air-to-air combat during the Korean war. American aviators were very successful that conflict. They achieved a 10:1 kill ratio over their North Korean and Chinese opponents. Colonel Boyd begain his study the the question: "How and why did they do so well?"

He noted that in several traditional measures of airrcraft performance the principal Communist fighter, the MIG-15, was fster, and it had a better sustained turn rate. But in two less obvious measures of aircraft performance, the F-86 was much superior to the MIG. First, the pilot could see out much better, while the MIG's flaired canopy made it difficult to see out. Second, the F-86 had high-powered and highly effective hydraulic controls and the MIG did not. This meant that while the MIG could do many individual actions--including turn, climb, and accelerate-better than the F-86, the F-86 could transition from one action to another much more quickly than the MIG.

Using these two superiorities, the American pilots developed a tactical approach that forced the MIG into a series of actions. Each time the action changed, the F-86 gained a time advantage, because the f-86 pilot could see more quickly how the situation had changed and he could also make his aircraft shift more quickly to a new action. With each change, the MIG's actions became more inappropriate, until they were so inappropriate that the MIG gave the F-86 a good firing opportunity. Often, it appeared the MIG pilot realised what was happening to him and panicked, which made the American pilot's job all the easier. "

If speed of executtion of the OODA loop how can these lessons be applied to developing a true trading application? First start with the network and the connection to the exchange. Then look at the software application itself. eSignal, Tradestation and other representations of 'charting applications' are more like the MIG aircraft in the above descriptions than a modern F-16 with its clear canopy. If you have used any of these legacy applications you can see they suffer from a software architecture that dates back to the late 1980's without improvement or innovation. Completely bloated database model, slow rendering of screen views that do not take advantage of graphic card technology that is now common to every trading computer, outdated SDI (single document interface) and MDI (multiple document interface) designed for word processors and spreadsheests, and the need to be 'generalized' for all markets rendering them a disadvantage for all but the slowest trading methods.

What then would a next generation trading application look like?

No comments: