Algo trading – too much of a good thing?

Back in the Olden Times when I started watching the markets, 30-40 million shares on the NYSE were considered to be big volume days. Institutional execution was done in the upstairs market by brokerage block traders calling around their accounts: “we have 100,000 XYZ for sale, do you have any interest?”

That has all changed.

Today, there are automated trading bots, or algos (short for algorithm), everywhere. It began with algos trading VWAP (volume weighted average price) orders. With the advent of day trading automation inevitably followed and today you can find specialized trading bots on the internet and even floor brokers are now using algos (see report). There are even some blogs, such as Skill Analytics and Zen Trader to name a few, devoted to this topic.


Flying on autopilot is not always a good thing
With so much automation around, some “reality check” questions come to mind:

• Is anyone doing a reality check on these algorithms?
• Can someone game these algos?

For example, there are brokers offering VWAP algos and guaranteed VWAP trading for individual investors for very low cost. Given the widespread adoption of these algos, don’t you think that someone could write a pattern recognition program to watch for a VWAP algo buying or selling a stock? This information is worth something to someone. As a trader once told me, portfolio insurance program trading (that exacerbated the Crash of 1987) was a huge boon to him because you knew that once the portfolio insurer had started the trading program he had more orders behind it.

If you must use algos and go on autopilot, do it intelligently. As I wrote in a previous post:

Mrs. Humble Student of the Markets, who is a pilot, calls it “raising your head up from the instrument panel and looking out the cockpit window once in a while”.