FastBreak

There are three steps in developing a FastBreak trading system: (1) determining a method of ranking, (2) determining the list of candidates - the fund or stock family, and (3) adjusting the parameters to optimal values by backtesting.

Determining a Method of Ranking

Here are the methods of ranking the FastBreak allows:

(1) Simple ranking over a period of X1 days – which fund increased the most in price?

(2) Curve fitting over a period of X2 days, then rank by slope at the end point. You can fit a line, a quadratic, or an exponential

(3) General Anchored Momentum (GAM), which I will not describe, because

(4) Most Anchored Momentum is a sub-case of General Anchored Momentum, and always seems to work better. Here you rank by taking a ratio of a more recent Exponential Moving Average (EMA) and a longer Simple Moving Average (SMA).

I originally thought that MAM was the best method of ranking, and it certainly has advantages. In August, 1999, however, I ran a collection of tests which convinced me to use Slope, calculated exponentially. So at that point I switched from MAM to slope. I still believe that for individual stocks, MAM works better, however.

Consider the chart below with three funds plotted, normalized to 1 at the first of the 35 day period. They are imaginatively labeled X, Y, and Z. If you did a simple ranking (which fund went up the most) over the 35 day period, clearly Z wins out. On the other hand, if you do a slope ranking over the last 5 days, then any method of curve fitting (linear, quadratic, or exponential) will prefer fund X, since the slope at the end point will be the greatest. Longer periods of ranking will give different results, depending on the length of the ranking period, and the end point.

Now consider funds X and Z over a longer period, but starting at the same time. The 10-day simple moving averages are also drawn. Remember that MAM ranks a fund on how far above its simple moving average it is. So after Z has made its big move around day 30, it is clearly going to have a higher MAM rank. But it levels off shortly thereafter. It is probably the case that any method of ranking that is not very long term will be fooled by the sudden increase of Z, and prefer it around day 30 over X. Yet buying at Z at that point yields a smaller gain than buying X. This is just one of the difficulties with this kind of ranking; yet it still usually gives excellent results.

Determining the List of Candidates

The next question is how to determine a list of funds or stocks to trade.

First, it is important to trade no-load funds that do not have a penalty for selling the fund before holding it a certain number of days (a “back-end” charge). The more funds in your list the better, even though it will take your computer a longer time to develop a system. Be sure to include foreign funds.

The best source I know of for information on fund characteristics, loads, fees, etc., is Microsoft Investor. In addition, fund information can be found at Yahoo Financial.

Be sure that all the funds in your list are open to new investors.

Two caveats, however.

The Rydex Sector funds do not seem to mix well with other funds of any kind. Develop a separate system for them. The same may be said for the ProFunds Sector funds.

I have never developed a system to trade the ProFunds index funds that works. Adding the ProFunds in with your other funds, however, seems to work well.

Adjusting tha Parameters to Optimal Values by Backtesting

The great thing about developing a system using a tool like FastBreak is that you can try various parameters over various periods of time, and then see which ones give you the best results. So when you are using a system you developed, you know that the odds are on your side. That is critical. I hear a lot of people predicting the market. How often have they been right? What are the chances that a given trade will be successful? What are the key parameters used to make the decision? What happens if you change those parameters? You can answer those questions. Can the market gurus?

Another thing you can do with FastBreak is to create an FNU file which tracks your strategy just like it was a fund in the Fasttrack database. Then you can display it and compare it to other strategies, funds, or indices.