Technology and Traders

Every day I find it absolutely striking how "old-school" trading remains and how much of an edge one can get simply by understanding technology. I think this holds especially true in fixed income. In equities, there is a fair bit of sophistication around electronic platforms and algorithms, but I would say what I am about to discuss is probably true. In structured products the key to those products comes in modeling them correctly (and not getting forced out of your position), but being able to build tools to react to market changes quickly still remains pivotal.

Fixed income still large revolves around the telephone. Everyone seems extremely relationship-centric and most products still trade over a telephone call. Many institutions get things as simple as modeling an interest rate swap wrong. An elemetary part of trading interest rate products, building a yield curve, is done in some surprisingly simplistic and ultimately incorrect ways. While most participants won't be off by more than a basis point, that basis point can be a rather significant edge over the long run (especially for a market-maker!). They won't be off for some simple ten year swap, but ask for specific dates on the forward curve and you can sometimes arb between dealers, albeit not in significant size, on something as basic as a swap.

In addition to the simplicity of their pricing tools, a lot of institutions have rather mediocre risk tools. Risk is absolutely paramount to a trader. If a trader knows his risk, he can trade around it and manage it. Any faulty risk will create PnL anomalies--one of the primary reasons for traders to get fired. A good risk system can mean the difference between a profitable desk and a non-existent desk. For a market-maker risk systems are especially paramount because they tend to make money on bid-offer, but not being able to hedge correct can cost millions.
It occurs to me that the reason these simple issues remain so unsophisticated is that most traders don't understand systems at all! The savvy young technologist can find many more efficient and more elegant solutions to many of the issues a trading desk runs into. Furthermore, a good couple of quants can usually come up with very good risk measures. The issue, however, is getting the technology people to understand the needs of the trading desk and getting the quants to understand what the trading desk needs to see.

One of my bosses used to say "visual display of quantitative information is the key to trading." I couldn't agree more. The trader who will survive and be able to pick-off other traders it he trader who can at once view everything in the market and everything in his risk in a consolidated, understandable form. I always see technology people going for the the most complicated and most cutting-edge solution when it doesn't help the trader's task at all (and often it takes so long that the trader's urgency with the problem makes them want to explode). With quants, the primary issue tends to be trying to find the "elegant" solution (I have to say I have been guilty of this) when the simple and quick solution will do just as well. Another issue with quants is that they often neglect the presentation of the information, which is really one of the most important parts to a trader who is constantly being bombarded with information and needs to be able to filter out the important from the unimportant. By providing a trader with proper systems to be able to quickly and concisely view market and risk information, he stands at the best advantage to make money in the markets--thus getting everyone paid.

How does an institution solve this dilemma? Of course it comes down to the people at the institution. Goldman often says "the key to the success is their people. Without their people, they would be like any other bank." I believe that statement to be 100% true. One way to solve the problem is to get the right quants and technology people who understand the trader's issues. This can happen either by having quants and technology guys who have been traders (nothing gets one to understand the traders sense of urgency and priorities more than actually trading), or finding ones who have a combination of quant/tech skills along with some significant experience with traders. I think the former approach works best. The other way to solve the problem is to get traders who understand technology and higher mathematics. Personally, I think this path is the better of the two.

In my opinion traders should all have the ability to re-create all their systems given sufficient time. No, they won't be the best, most efficient or fastest programmers. And maybe they won't be the best of mathematicians. They should, however, have a thorough understanding of how everything they use works. My experience shows me that the best way to do this is to actually have the primary technologists and quants hired by the desk. Specifically, that mean they don't work for the technology area of the company or risk or any other "branch." The desk themselves have to hire and pay the quants and techs. That dedicates the professionals to the well-being of the desk and helps them understand the needs of the desk. Just as important, the desk then needs to pick and choose the best of the techs and quants to actually become traders.

In this way, the people who best understand both the systems and the traders' needs become the next generation of traders. I have found this method of building a desk to be almost always successful. Another added benefit is the boost in morale the desk gets from this path. When traders can build the tools as well as the quants and techs, then the quants and techs respect the traders. In many places traders, techs and quants have a certain animosity for each other. If the traders know what the others are doing (because they've done it before) and can look over their shoulder to offer intelligent tips, that's what builds respect on the trading desk. The high morale that results from everybody being able to respect the breadth of knowledge thr traders have is priceless.

When I was first interning at a bank I was once told that most areas of the bank have a pretty good interview process, but the trading desks have the most issues because it's so hard to filter for the skills that make a good trader. I think my method of training new hires to think like traders and build the traders' tools makes the most sytematic and successful method of weeding out the best of the candidates to be traders. Too few desks on the street have a good systematic way of producing both good traders and good trading tools. In trading there is a never-ending supply of good ideas that need to be developed, tested and implimented. It seems no matter how long you've been developing tools, there are always new ideas to make the tools smarter and faster. The key to running a successful business in trading these days lies in the ability to continuously improve one's tools, and only the trader can drive that process.