When state-of-the-art doesn't cut it, you have to roll your own. This financial services client wanted infinite flexibility in finding good trade ideas in the marketplace, but hit the wall on a number of fronts: all existing search tools they had at their disposal had serious limitations, could not deal with multiple and simultaneous data feeds well, and additional steps had to taken to reconcile search returns (even with powerful booleans) with variable trader risk limits and risk violation settings that change on a daily basis.
The solution to these many thorny site-specific problems was to invent a new query language tailored specifically to the needs of high-volume derivatives traders. However, traders are busy people, and are of varying levels of technological sophistication, and few have time to learn the ins and out of a LISP-based query language, from syntax, grammars, logic, and industry and firm-specific variables and constants. A visual drag and drop graphical user interface had to be developed on top of this new language, and that's where I came in. We looked at MIT's Scratch as a starting point, and then developed it into a new trading framework that (from what I hear) has increased a trader's "vetted" trade volume nearly 1000%. Trading workflow has changed dramatically. Traders are no longer monitoring incremental price ticks or bid-ask spreads, nor are they manually working their 1000's of orders during the day. This new programming language essentially allows them to build human-computer parametric trade algorithms on the fly, constructing smart queries that are essentially based on their pre-/intra- market intuitions and insights. It's incredibly trader-specific.
Drag, drop, generate, auto-execute, auto-hedge, get a cup of coffee, watch a little Bloomberg.