This post is the first in a series of expert profiles that will give readers a better understanding of Automated–or Algorithmic–trading, as well as other trading technologies.
Our series starts with Ben Gimpert. Ben is a professional software developer and an expert in algorithmic trading. He offers keen insight to the current state of trading technology–and where it’s heading.
You can learn more about Ben and his technology at his site, Something Modern Logic.
What are some of the biggest challenges for Automated Traders and developers of Automated Trading systems?
The distinction between automated trading system developer and automated trader can be a blurry one. Where we draw the line hints at what I believe is the most difficult problem. Most trading systems, automated or otherwise, output time-sensitive signals that should be realized in the market. (i.e. “Go long S&P futures, and short volatility on gold right now.”) If a human trader manages the position once the entry is signaled, then the strategy will struggle with your typical human mental biases.
Managing the entire lifecycle of a trade with software is the primary difficulty and opportunity in automated trading. This means coding up more than a signal to enter a trade. Your automated system should specify precise position sizes, as well as stop-loss and take-profit levels. The risk and money management logic behind these decisions should look at a market’s volatility, the level of account equity, and important minutiae like contract multipliers and broker fees.
Ironically an automated trading system need not actually use a broker’s API! For example, a system that puts on a lot of risk might adversely signal the market with an exchange limit order. Instead a good automated trading system would specify a precise stop-loss and take-profit level that the human trader calls in as market swings. The real work in automated trading is in specifying the exact position size, stop-loss level, take-profit level, and maximum holding period — in addition to the entry signal. Next to those calculations, talking to the API is easy!
Is there room for small operators in the Automated Trading space or do the big banks and hedge funds have things wrapped up?
Absolutely. This is probably the most common myth in trading. “How can the little guy win, when a thousand overpaid MBAs are already working on this?” The assumption is wrong because investment banks and hedge funds are simply not that hip! As someone who spent years working in the trenches on trading floors in London, the average level of software engineering and artificial intelligence expertise is shockingly low. Bankers and hedge funds pay well because of the barrier to entry (jargon), because of the environment (hyper-competitive), and because of the exhaustion (long hours). The bonuses ain’t for elegant and efficient systems! The credit crisis has made this point more loudly than I ever might. Like any large organization, investment banks and hedge funds are bureaucratic and resistant to change.
There is also an argument that small can be an advantage, because some markets are too illiquid for the big players to care. If Goldman puts on $100m of risk in an obscure penny stock, the market will instantly move against them. Maybe the Goldman’s of the world have such good software systems that the marginal cost of trading across every potential market is zero. But I doubt it.
Automated Trading, Investing, Online Investing AI
AI, algorithmic trading, Automated Trading, experts, hft, machine learning, profiles, something modern, trading
by Nassim Taleb, and I must tell you it is a pretty interesting book. Although his wandering and complicated style is over my head, he brings up some very interesting issues that are ignored in just about every other book I have read about trading.
is one of my favorite financial books of all time. It's an allegory about a guy who has a terrible job, and how he overcomes challenges to achieve his dreams.
