|Trading systems||Empirical decision making|
A trading system requires the designer to make some empirical decisions that directly affect the system's performance - if there was no need for this decision making, everyone would be rich. Here are some basic factors that system designers must decide on and some guidelines:
• What time period should I use? All equities can be analyzed from multiple perspectives of time periods, ranging from one minute to one
decade (or more). Deciding which time period to test can drastically affect the performance of the system. More reliable results generally come from longer time periods, while short periods can be misleading when judging real market conditions. However, this does not mean that only extremely long price periods should be used. It is important to keep in mind that the longer the time period, the longer it may take for profit to be realized. Observe the following example of Microsoft's long term, a period of more
than 20 years, compared to its short term, a period of a few weeks:
We can clearly see that the short term is not an accurate representation of the
long term, and vice versa. As a general rule of thumb, five to 10 years is a good target for medium- to long-term system traders, and six months to five years is a reasonable range for short-term traders. Again, it depends on when you plan to liquidate.
• What price series should I use? Most equities are charted on an unbroken price series - that is, the charts are continuous. When trading futures and some other equities, however, there is an option to use actual contract data instead of continuity. Futures contracts themselves only last a few months, and system backtesting often requires a year or more of data; therefore, system traders often utilize continuous futures, which are a series of contracts combined to create a continuous stream of data. As a general rule of thumb, long-term traders should stick to continuous futures, while short-term traders should use actual contract data.
• What parameters and settings should I use? We explore this further in subsequent sections that address the construction of a trading system. Basically, parameters are selected by "guessing-and-checking," or producing "blind" simulations, or presetting a group of parameters, and then using the average to determine performance.
Again, many of these factors can be influenced by desired liquidity, time until liquidation, risk and a multitude of other factors, so it is important to take the time to decide which works best for you.
Contact for more learning: webmaster@freehost7com