Premium Market Analysis

Trader education, Trading Strategies

Fully Automated Trading Systems: A Dubious Idea

The idea that a retail trader, investor or even fund can employ automated trading systems and profit consistently without continuing adjustments or interventions based on fundamental conditions is dubious. The reason is simple: if math could provide the solution to consistent profits from trading, then mathematical geniuses would have already figured it out and would own the market. But that is not the case.

I got my first job in Wall Street in 1988 and the most successful people I have met since were the ones that effectively combined fundamental and technical trading. I have never met a consistently successful mechanical trader employing fully automated systems, although according to probability and statistics there should be some. The fact is that all systems fail when adverse conditions arise and successful traders are those who know when to stop them from executing trades. For example:

  • Trend following systems must be stopped during sideways markets
  • Mean-reverting systems should be halted during momentum surges
  • Breakout systems should be switched off during periods of mean-reversion
  • All systems should be stopped when there are adverse fundamental developments

Yet, many, especially retail traders, think it is possible to have a winning systems without a meta-system, i.e., a system that tells you how to use systems. In most cases, the meta-system is based on experience and the ability to understand price action. In my opinion and based on over 25 years of experience with the equity, fixed income, forex and commodity markets, it is hard or even impossible for anyone to profit longer-term using automated trading systems and without having experience with the markets. The experience provides the foundations for using the systems effectively. Systems are valuable because they automate signal generation and order placement process, among other things, often in multiple markets, but they can be a disaster when applied at the wrong time. I am tempted here to offer the example of those EUR/CHF trading systems that were based on a persisting peg of the CHF to EUR. When the peg was removed last month, the system lost multiples of what they have made in some cases.

In the stock market, for example, a simple automated system is based on the 50-200 crossover. Currently, there is a persisting golden cross, i.e., the 50-dma has been moving above the 200-dma for 766 day for a gain of nearly 67%, as shown on the SPY chart below.

SPX_20150206_50_200

Can anyone assume that the gains will not be significantly reduced or even wiped out in a month or even in a week due to adverse fundamental news and before the golden cross reverses to offer an exit signals? The idea that one can rely on this type of a mechanical system to profit in the longer-term is dubious because many things can go wrong. The system can be useful but without proper analysis of fundamentals it can end up being a disaster, as in the case of those EURCHF mechanical systems. Experience is necessary to know when to pull the plug, go flat and get out. Therefore,

Systems augment experience by providing an unbiased timing method for entering the markets and experience augments systems in determining when they are out of whack. Any scheme relying solely on systems has little chance of succeeding.

Note that the above discussion does not apply to mechanical systems that act as market making agents, such as HFT or scalping systems, because of the fast executions but this is not to say that there are no problems of fundamental nature with such automated algorithms. Also, note that on the sell side, the terms automated trading system and trading algo often refer to liquidity seeking and order management algorithms and are not related to the same term used by the buy side and especially by retail traders and fund managers.

You can subscribe here to notifications of new posts by email.

Disclosure: no relevant positions.
Charting program: Amibroker
Disclaimer