Understanding Systems and Markets
Algorithmic trading systems bring the investor good returns and increased predictability. Compared to discretionary trading it is analogous to the turtle and the hare. You can’t rush a system or make it do anything other than what it was intended for. Each strategy has a particular profile: trend systems have many small losses and fewer large profits, and mean reverting systems have many small profits and a few large losses. While you can alter these numbers somewhat, you cannot change the big picture, nor should you, and you cannot force a market to produce a profit on demand. It’s a matter of accepting the way a system performs, and the way prices move, and working with them.
Of course, among the many trending and mean-reverting methods there are better ones. The best always have a sound premise. They are not created by scouring the computer for combinations of indicators and stop-losses. They are the results of observing the markets and understanding what makes them move.
Using trend-following as an example, we have seen that the most persistent trends are in the interest rates. That has been the result of Fed policy, effectively lowering rates over the past 25 years. Until recently many youngWall Street analysts have never seen a market where interest rates have risen. Those interest rate trends directly affect FX prices. Money flows to the countries with the highest returns net of inflation (and other political risks); therefore, lower rates create a trend towards lower currency value.
Then long-term trend following is really trying to be on the same side of the market as government policy. It is a sound premise. On the other hand, we know why there are short-term trends – changes in supply and demand, a natural disaster, seasonality – but in most cases these trends are erratic and of unknown length. They can be profitable, but they are far less consistent than long-term trends. Based on this reasoning, many hedge funds and Commodity Trading Advisors (CTAs) have adopted macrotrends as a large part of their portfolio with great success.
The Research and Development Process
Having decided on a method, the next step is to develop the rules for trading and controlling risk. Some of the important steps that we follow are: The more data the better More data contain more patterns and a chance to see how the strategy works in many different conditions. Although some would say that the old data is no longer representative of that market, we don’t believe that. The market is full of uncertainty, and a system is robust only if it can deal with bull and bear markets, price shocks, and doldrums.
Apply consistent rules across all markets
We know that markets have their own personality. Apple and Amazon are not the same as a utility or even Bank of America. Corn is not the same as crude oil. What makes these markets similar are the investors, the way they react to news, both macro and micro. A successful trading strategy must consider the differences, such as volatility, and the similarities, such as the trend or arbitrage, but account for them in a systematic way, using a common set of simple rules and formulas that adapt each market. The alternative is to have very specific rules for every situation and every market. Using the same rules is a robust solution. Using different rules tends to overfit the data and has little predictive value. We subscribe to the approach that “loose pants fit everyone.”
Control the risk
Risk management is equally as important as a sound premise and a good strategy. Traders that focus their resources on a single market may reap huge returns — or huge losses. Concentration of capital increases risk. One aspect of risk control is diversification.
Proper diversification should include:
- Markets that are unique from one another
- Multiple strategies that are unique in the way they see price movement
- Equalizing risk across markets, sectors, and strategies
There is also individual trade risk. Use a strategy that takes you out of the market rather than a stop. Some traders limit risk using a stop-loss order; however, it is better to have a “natural” stop that conforms to the rules of the strategy. It is also possible to control risk by varying leverage, most common for portfolios of futures markets.
Finally, there is portfolio risk. We rightfully expect that daily portfolio returns will be less volatile than individual stock or futures market risk; however, that does not mean that the risk won’t be large during periods of stress. For futures, a method called “volatility stabilization” alters the leverage to attempt to keep daily volatility near a target level, often about 14%. Because stocks are not normally leveraged, portfolio risk is a combination of: Trading equal value of each stock Hedging with a broader index when necessary Diversifying into unique strategies Using a stop-loss when there is no “natural” system exit Specific risk controls are discussed in more detail in the description of the individual strategies.
For more detail and an in-depth discussion of risk, see Chapters 23 and 24 of Trading Systems and Methods, Fifth Edition (Wiley, 2013).
[button link=”https://kaufmansignals.com/products” size=”xl” color=”silver”]Subscribe to KaufmanSignals[/button]
Find out how it all works – Register for a free membership with KaufmanSignals.com