When building or evaluating trading systems the many benefits of systems that trade very frequently are often overlooked. A system that trades frequently has many advantages over less active systems that appear to be more desirable because they have better performance ratios.

If a strategy is profitable the more it trades the more money we should make. I apologize for stating what should be obvious but you would be surprised at how often I hear discussions about selecting systems with the highest level of "expectancy" or highest "profit factor" without relating these measurements to the system's trading frequency. Simply stated, our goal should be to show the most profit with the least amount of risk and trading frequency plays a critical role in maximizing profitability and controlling our risk.

Trading frequency represents opportunity for profit. The more opportunities we can find the more profit we should expect. For example, a strategy that has a very high profit factor of 4 (profit factor is total profits divided by total losses) may not produce as much profit as a more active system that has a profit factor of only 2. Since the strategy with the lower profit factor is profitable and it has many more opportunities it may easily produce more total profit than the system with the much higher profit factor.

Active systems should give us a higher confidence level when analyzing our test data. In addition to increasing total profits, a very active system gives us much more data to analyze when doing our preliminary research. If we have a long-term trend-following strategy that produces only 50 trades over five years of data our positive results may not be nearly as reliable as the results from analyzing a more active strategy that produced 1000 trades over the same data sample. I would be willing to bet that the system with the larger sample of trades is more likely to produce profitable results in the future because our level of confidence must relate to the number of samples in our testing.

Active systems should produce more reliable results and a smoother equity curve. If we flip a coin only ten times our odds of having 50% heads and 50% tails are not very good. However if we flip the coin one thousand times we are likely to come much closer to obtaining 50% heads and 50% tails. The same logic applies to our real-time trading. If we have a large sample of real trades then our results should come closer to our expectations than if we only have a one or two trades. The active system will approach our expectations much quicker than the system that trades infrequently. If we have 50 or more trades per month with a good system we might reasonably expect to be profitable every month. However if we have a system that is only producing two or three trades per month then our monthly results will less predictable and inconsistent. The infrequent trading system might be expected to produce a profit every year but it would not be realistic to expect it to show a profit every month because the sample size in a month will be very small.

Here is a quick summary:

1. Active systems give bigger samples in testing which make the test results more reliable.

2. Active systems have more opportunity for profits and should produce more total profit over time.

3. Active systems should produce a smoother equity curve.

When setting your goals for a trading system you should give trading frequency a high priority even if it means that some of the usual performance measurements may suffer.

How to increase trading frequency:

1. Use quicker parameters for entries. Trading frequency can usually be increased in a system by using more sensitive (usually shorter) parameters for the indicators that determine the entries and exits.

2. Be aggressive when taking profits. Quicker exits that lock in profits will tend to increase the number of trades but may reduce the average profit per trade. That trade-off may prove to be worthwhile and may substantially increase the total profit over the long run.

3. Trade multiple markets. Diversification among markets should produce more trades and more consistent results.

4. Trade multiple systems. Adding more trading systems will increase the activity level. Use as many systems as you believe to be practical in terms of available capital and as many systems as you can accurately monitor.

5. Trade multiple time frames. A system that works well on daily charts may also produce positive results on hourly charts or weekly charts. Don't make the mistake of assuming that only one time frame must be used.

Here are a few final thoughts.

Drawdowns: I have found that adjusting a system to be more active will almost always increase the size of the drawdowns. However in many cases the larger drawdowns are simply the result of having a much larger sample size over the period being tested. If you perform a Monte Carlo simulation on a system that trades infrequently you will observe that larger drawdowns become more likely as the number of trades in the simulation increases. This means that your test data on the inactive system is showing an unrealistic drawdown that is probably lower than what you might expect to actually experience when trading that system over the long run. If you don't have a program to perform Monte Carlo simulations and you are analyzing a system that trades infrequently you should double the size of the historical drawdown to give you a more realistic idea of what to expect in real trading with a larger sample size.

Increased costs: Trading more often will increase your trading costs in terms of commissions and slippage. Be sure to factor in realistic costs when doing your testing. The idea of trading frequently is to make more money. At some point increased trading will begin to reduce your total profitability. You should know where that point of diminishing returns kicks in.

I'm not a mathematician but I think that system traders need a formula for comparing systems that includes the frequency of trading. (For example: Expectancy times frequency or Profit Factor times frequency.) Any other suggestions?

by Chuck LeBeau

## Wednesday, February 28, 2007

Subscribe to:
Post Comments (Atom)

## No comments:

Post a Comment