Real World Trading Costs Include Errors

A trading magazine recently published an article about a university that runs a FOREX fund as part of its undergrad business program. The students manage $2,500 in real money and learn how to trade. School officials noted that more than 80% of the participants have landed real world internships and they think that’s due at least in part to the experience they gained managing the fund.

The trading system is based on a moving average crossover and seems to be a very profitable idea. It is more fully described on the University of Dayton’s web site. From October 2010 through May 2011, the magazine said that the system showed a profit of more than 62.5%. The account that the students managed delivered a return of only 7.2%.  The difference is the real world cost of trading in this example.

The first, and probably the most important, lesson we can draw from this article is that the system’s timeframe needs to match the trader’s schedule. This strategy uses four hour charts to generate buy and sell signals but only executes the orders at six predetermined times each day. In the real world, part-time traders have other jobs and can’t spend every minute of the day monitoring the markets. But the results that part-time traders see will most likely be extremely disappointing when compared with what the system achieved. Systems take all of the signals and part-time traders need to be in a position to take all of the signals. This is easy to do, if the system is well designed.

In the article, the authors note that there were other problems that impacted returns. Testing under the user-defined time restrictions for trading, the strategy would have returned 18.6%. This is still significantly better than the actual results.

There are several very important lessons here so let’s step back and put the three sets of results in a single place to make this easier to keep track of:

  • 62.5% total return assuming perfect execution from 24-hour trading
  • 18.6% total return with perfect execution trading at preset times
  • 7.2% total return in actual trading

These dramatic numbers illustrate a simple point. Part time traders need to design a system that fits their schedule. To make the trading possible, the students only placed trades every four hours, at 1:00, 5:00, and 9:00, AM and PM. This allowed them to complete other activities students do and they didn’t have to follow the markets between the designated times. This decision almost guarantees significant slippage, and we have a quantifiable measure of what that costs since the returns were decreased so dramatically.

One way to avoid this would have been to use end of day data in the system design and enter trades in the evening for the next day’s open. Their strategy uses several variables in addition to the moving average crossover signal and those parameters may help improve system performance using the four hour trading timeframe they chose. However, the most important part of their strategy seems to be the 10-period/20-period exponential moving average crossover. Long signals are taken when the short EMA crosses above the longer EMA and short trades are initiated when the 10-period EMA falls below the 20-period EMA.

The students refined the system so that it is not always in the markets based upon how far apart the moving averages are. These additional parameters may have a positive impact on returns, but one could argue that they aren’t trading the system they designed given their time restrictions and therefore would be better served by a less complex system that fits their schedule.

Testing the time period they operated the fund in, a simple crossover system using only end of data would have returned 48.2%, after allowing for trading costs of $5.00 per round turn. Perfect execution of their four hour strategy would have beaten this performance, but that seems to be theoretical for traders who honestly can not commit 100% of their time and undivided attention to the markets.

Actual results obtained by the students differed from the theoretical results because of additional errors. Only one error was detailed in the article, and that involved a student missing the scheduled trading time and placing an order late. This is another cost related to the fact that their system did not match their schedules.

Trading takes time, and missed trades are errors. Although not detailed, it is also possible that results suffered because errors were made when placing trades. In the real world, that definitely happens and although in theory, errors should be just as likely to deliver profits as losses, it just doesn’t seem to work that way. Experience shows that errors usually end in losses.

Errors can include failing to enter an order even when you are monitoring the markets. Distractions can take your attention away from the markets and lead to a missed opportunity. Most trading software and broker platforms allow you to set some kind of reminder alert, and it’s a good idea to take advantage of that.

Entering the trade can also be a source of error. Clicking buy instead of sell or keying in the wrong numbers on the trade entry screen can lead to expensive lessons. All brokers require you to confirm the trade before submitting it, although some will give you the option to submit orders directly without confirming the data you just entered. If you aren’t a full-time trader, it is best to spend the few seconds reviewing each line on the confirmation screen. By using end of day data, or another trading time frame that allows you time to think, there should be no reason to rush.

Finally, traders can never forget that computers crash and internet connections sometimes fail. Brokers have extensive backup plans, and every trader needs a plan in place for how they’ll enter orders if this happens to them. It could be as simple as a phone call to a trading desk, if you already know the number and don’t need to look it up on the broker’s web site after the internet goes down. Plans should also include having a backup of your software and data that’s readily accessible. You also need to know where you will go to place trades if the power goes out.

The importance of that backup plan can’t be overstated. As I write this, southern California is in the midst of a power outage and many parts of the East Coast have suffered extended outages because of severe weather. If you have a trade open and a hurricane is approaching, be sure you place limit orders that can get you out of a position automatically. This may not be the optimal way to execute trades (and it isn’t), but it will ensure you get out if the market turns sharply against you while you’re sitting in the dark.

We can learn a lot from the actual performance of that student account. First we should design systems that we have time to execute. Then we should ensure that we have safeguards built in to be certain that we can execute the system even in the face of a natural disaster. Finally, we need to take the time to double check each step in the trade entry process.

Colleges provide education, but many still say that experience is the best teacher. In this case we can learn with real numbers from the experience of others and see that trading errors and scheduling limitations cost the college students about 88% of their trading profits. A more realistic execution schedule using end of day data could have delivered higher profits with less effort. There are definite benefits to keeping it simple in terms of the system and implantation.

By Michael J. Carr, CMT