We’ve discussed a lot of issues needed to create a sound algorithmic trading strategy, but it’s necessary to put it all together to solidify those ideas. If you’ve decided to base the strategy on a long-term trend, there are still a number of specific decisions to make with regard to the strategy, including:
- The trending technique.
- The rules for buying and selling.
- Stop-loss or other individual trade risk controls.
- Profit-taking and reentry.
- Single or multiple entries and exits.
There are also other decisions that apply to any strategy:
- Position sizing.
- Volatility filters.
- The test plan, including the markets to test, the date range over which the strategy should work, and the criteria you’ll apply to decide if it’s successful.
- Creating and testing a portfolio of stocks, ETFs, or futures markets.
You could get much more complicated, but if you can work through these points successfully, you will have created a strategy that’s likely to succeed.
The concept is that a long-term trend captures the underlying direction of the market, that is, the way prices react to economic policy, and tries to ignore the noise caused by day-to-day news releases. We know that the long-term (“macro”) trend has a history of success, and we want to participate. The three most likely candidates for identifying the trend are the moving average, the breakout, and the linear regression.
As we’ve stated earlier, any one of these can be doing better ...