Visualization Software in trading | October 2, 2006
by Jeb Boniakowski
I've seen a lot of visualization apps and toolkits popping up lately that are targeted at trading floors. However, I haven't been seeing a lot of visualization tools actually deployed on trading floors. It seems that for most people, the most sophisticated visualization they regularly look at is a historical price chart. At the same time, it's hard to think of an audience better suited to visualization products: many traders deal with massive amounts of data, have to comprehend the data quickly, and make decisions based on that data without sufficient time for a lot of in-depth analysis. It sounds like a natural application for viz software. So why are so many of your average trader's wall of screens dedicated entirely to flashing Bloomberg messages?
I came across a blog post that brings up one good point that I hadn't thought about before: the mapping between datasets and visualizations is non-obvious, the data has no intrinsic shape the way medical, geological, or engineering data does. At a glance, there doesn't appear to be nearly as many successful applications of viz technology to areas without natural, obvious mappings.
I think the larger issue though is that there's little use in financial markets for one-size-fits all visualization tools. Where's the edge in that? Most viz companies and projects are focused on the sophistication of the viz part, leaving the user to figure out how to get the data together to put into the viz. Also, customizing the viz is usually extremely difficult, generally requiring pretty advanced coding skills.
At Proto, complex visualization is not our focus: you won't see any screengrabs on our website of super-complex 3D visualizations like you do on a VTK's website. We are very interested in it though. Our mission is to give you the platform to build the tool you need, and charting and visualization are a often a part of that. With that in mind, we've focused on implementing things like Adobe Flash compatibility so that our users can flexibly extend the types of visualizations we can provide.
More fundamentally though, we suspect that in most cases, the best visualization is a pretty simple one, just very carefully and specifically used. With that in mind, we've started building Flash components to display data and wrapping them in Proto apps that let you easily manipulate and process your data to prepare it for display. Here's one that was uploaded a few days ago. It also takes advantage of new features in last week's update release. The first time Aron showed me this app I dropped in some sample portfolio data I had sitting on my computer. I'd seen heatmap-type market data visualizations before, but using Aron's app was totally different. One of the things we talk about at Proto is that often the interesting questions aren't obvious; you start with some set of data and a hunch, and only through asking exploratory questions do you find which questions are interesting. This is why a stack of static reports in an enterprise BI system will always be a letdown. Using Aron's app, you can look at something in two seconds just to see if it's interesting (like "does my equity portfolio beta vary with the PM who picked the stock?") and often, looking at the results of those inquiries inspires you to think of actually interesting questions. Basically, we think most people have been focusing on the wrong end of the problem: the sophistication isn't needed in crazy display styles, but in getting the data ready to display: calculating, summarizing, merging, cleaning, etc.
As always, we love feedback so if you've got an idea for a cool visualization component please shoot me an email (jeb at proto).
