This article contains an introductory description of trading systems, including many related aspects such as margin, account size, system-assist brokers, back-testing, hypothetical versus actual-trade results, drawdowns, evaluating the quality of a trading system, and portfolio design. If you are new to trading and trading systems, you should find this article helpful to get started. If you understand the material in this article, then you will be in a better position to understand the rest of the material on this website. Any opinions presented here are those of the author.
First, what is trading? I think of "trading" as buying and selling stocks, futures contracts, or other liquid items with the goal of making a capital gain (profit) by taking advantage of anticipated price movements over a relatively short period of time. Trading often involves high anticipated profits, and high risk. The perceived price patterns are often based on "technical analysis" - a mathematical, logical, and/or pattern-based analysis of price (and sometimes other market parameters such as volume) versus time. Trading is somewhat like gambling because a given trade will either be a winner or a loser, but traders who have found repeatable patterns may believe that the odds are in their favour to win more than they lose, over a possibly large number of trades.
This is different from "investing", which is usually a longer-term approach to making a profit from a combination of capital gain, interest, and dividends by buying and selling stocks, mututal funds, or other investment vehicles. Investing is often based on "fundamental analysis" - which looks at corporate data such as market area, sales, and profits; plus economics, weather, politics, demographics etc.
Who should trade? And how much? Some studies have suggested that placing 10% or perhaps a maximum of 20% of your capital in futures investments can benefit your overall portfolio. But trading is a high-risk venture, and most brokers and advisors will warn you: do not trade with money that you cannot afford to lose! This is good advice! If you want greater profits, then there is usually greater risk. Risk means that you can lose money too; you could lose all the money in your trading account, or even more(!). This does happen.
A trader can usually go either long or short. Going long means to buy an equity or commodity, with the expectation that the price will go up. Going short means to sell an equity (that you don't own), with the expectation that the price will go down. For futures contracts, it is just as easy to go short as it is to go long. For stocks, however, the shorting mechanism is more involved.
Next, what is System Trading? This term would probably apply to any systematic form of trading where there is a defined set of rules. For example, the rules for one simple system are to (1) buy when the weekly closing price of a composite index rises 4% or more above the lowest weekly close since the previous sell signal, and (2) sell when the weekly closing price falls 4% or more from the highest weekly close since the previous buy signal. Some systems are based on standard indicators such as moving averages, relative strength index, and Bollinger bands. A system may be based on pivot points, candlestick patterns, or Elliott wave theory. Some systems are based on complex patterns that may or may not use indicators in their rules.
The term mechanical system is often used to emphasize that a system is based on a fixed set of rules. This term may also imply that the rules have been implemented in a computer program, which outputs buy and sell signals based on the rules and a stream of input market data. The trading systems that we report largely fall into this category.
Margin is the amount of money that a trader must put up (i.e. have in their trading account) in order to buy or trade an equity or futures contract. The margin requirement is typically around 10% for commodity futures and 20% for single-stock futures. For day-trading systems, where positions are not usually held overnight, the margin requirement is one-half. This means that you could purchase or sell a contract valued at 5 to 20 times the amount of money in your account! This high "leverage" is one reason that there is such a high potential for gain, and correspondingly such a high risk of loss.
Account size. One of the curious things about trading is that there is usually no fixed amount of money required to trade. Of course, the margin requirement is the absolute minimum, but very few people would be willing to trade with that small amount of money, because a single losing trade would take their account below the minimum and then they would not be able to trade until they added more money. Most system vendors recommend a minimum account size of 3 times margin. We prefer to use a more conservative account size of 5 times margin for our percentage calculations. Some advisors recommend larger account sizes, perhaps as high as 10 times margin. When deciding on account size, investors often look at the maximum drawdown (loss) that a system has had in the past, and fund their account so that the maximum drawdown would represent a tolerable percentage of their account, e.g. 25%. A balanced portfolio can be essential here, because the expected maximum drawdown for a well-balanced portfolio should be much less than the maximum drawdown for some of the systems in the portfolio. (One of the main features of Futures Examiner is that you can experiment with various mixes and weightings of components and see the effect on the overall portfolio.)
Money management comes into play as your account size increases and decreases. The basic idea is that as your account size grows, you will likely want to trade additional contracts, and when your account is in drawdown, you may want to trade fewer contracts. There are a number of mathematical approaches to determine the optimum number of contracts, but this topic is not within the scope of this website. Some trading systems have built-in money management rules, and suggest the number of contracts to trade as your account grows or diminishes.
When results are reported on Futures Examiner, they are usually not compounded. In other words, results are calculated assuming that the trading level is always one contract. This method is commonly used when results are reported for trading systems where the trader usually determines the number of contracts to be traded. For example, this method is used in Futures Truth magazine, on many system-assist broker's websites, and in many TradeStation reports provided by system developers.
By comparison, compounded results are usually reported by CTAs (for Managed Futures products) and CPOs (for Commodity Pools). In other words, results are calculated assuming that the trading level changes in direct proportion to the cumulative equity. Another way of thinking about this is that any profits are automatically re-invested (and any losses reduce the level of investment). However, for many CTA Managed Futures products, the level of trading is actually determined by the investor, and changes only when the investor so instructs. In cases like this, the compounded results reported by the CTA would not track the equity in an investor's account, which might be misleading. For a CPO's Commodity Pool, however, the level of trading would change as the cumulative equity changes, so compounded reporting would produce numbers that should track those in an investor's account.
Reference and link to Basie Features and Conventions.
Trading systems may use a short, medium, or long time-frame. Day-trading systems use a short time frame, and always exit the market at the end of the trading day. Some day-trading systems may trade a maximum of once per day, while others may trade as many as two or even three (or more) times per day. Swing-trading systems may hold a position overnight, and often for several days. Occasionally they will hold a position for many days. Typically these systems might trade anywhere from 5 to 10 times a month. Position-trading systems use a longer time frame, and will often hold a position for weeks or even months at a time.
The average number of trades /year that a system makes is usually reported. A day-trading system might trade anywhere from 30 to 300 times /year. A swing system might trade between 50 and 150 times per year. And a position-trading system might trade anywhere from 3 to 50 times /year.
Typically a system developer will formulate a system, implement the rules in a computer program, and offer it to the public on either a purchase or a monthly lease basis. Many of these systems run on the "TradeStation" (tm) computer "platform", which is a computer program designed specifically to support mechanical trading systems. TradeStation runs on a PC computer under Windows, and includes the "EasyLanguage" (tm) programming language, which is used by the system developer to implement the trading rules for their system. If you purchase a trading system, then you usually receive an executable version of the software, and may receive the source code as well. In order to trade the system, you need to run the software on a computer that has the necessary supporting software (e.g. TradeStation) and a suitable data feed. Data feeds are available from a number of sources, who charge a monthly fee for the service.
Some developers, typically when they lease a system for a monthly fee, provide the trading signals rather than the software. The trading signals may be sent to subscribers or their brokers by email. In this case, there is no requirement to have a computer to run the system software on.
System-assist brokers are stock, futures, and/or Forex brokers who will execute a trading system for you. They provide the necessary computers, TradeStation (if required), and data feeds. They have staff who monitor the system signals and market activity throughout the trading day, and will make the trades as signalled by the system on your behalf. You pay for all this in the commissions that they charge.
The developer will usually "back-test" their system, which means to feed the computer program a stream of historical data, often going back five or ten years or more, to see what the results would have been if the system had been traded during that period. The results from this kind of testing are called "hypothetical", and they are obtained using a particular system version and parameter settings.
After a system is released to the public, the initial version may be used for many years without change. However, a new version may occasionally be released by the developer. This might occur, for example, after a difficult set of market conditions has resulted in poor performance, and the developer has found a way to improve the performance under those conditions. Or the developer might simply have found a way to improve the performance generally, that justifies release of a new version.
Many trading systems use some parameters in their calculations. Parameters might be used to specify such things as data source(s) and rate(s), stop size, trigger points, time offsets, number of bars, maximum trades per day, various ratios, or risk level. By varying the parameters, the developer can determine the best value(s) to use for certain periods of historical data, and observe how sensitive the profit/loss is to various parameter settings. For some systems, the parameter settings can have a profound impact on the profit/loss.
After a system has been released, the market conditions may change, and the system's performance may be quite different from what the results of the pre-release backtest would suggest. Hence the old adage: "past performance does not necessarily indicate future results". On Futures Examiner we publish both actual-trade and hypothetical results, so it is important to realize that backtest results preceding the release date may not be representative of future results. Some developers use the term "out-of-sample data" to refer to post-release conditions.
The parameter settings used in a backtest may be different from those used in the field. For example, a system could be traded in the field with one set of parameters for a period of months or years, whereas a subsequent backtest using the same system version but with different settings might produce much superior hypothetical results. Again, users are cautioned to view hypothetical results with caution, particularly when the system version and/or parameter settings are different from what is being used in the field. Results in Futures Truth magazine are good in this respect, because although they are hypothetical, they are based on forward testing.
When a developer runs a back-test and/or publishes hypothetical results on their website, they may or may not include an allowance for slippage and commission.
Slippage is the difference between the price when the system issued its "buy" or "sell" signal, and the actual price (fill) that the trader got. Slippage can make the difference between a system being profitable or the system losing money, particularly for day-trading systems which often average only a small profit per trade. For example, for the popular S&P 500 futures contract, slippage values from 0.45 to 0.65 points per round turn are often seen. For the corresponding mini, slippage from 0.2 to 0.4 points per round turn are often seen.
Commission is the price that a broker charges to make a trade for you. Commission is usually quoted per round-turn, which means both entering and exiting a position. For day-trading, round-term commissions are usually charged because you enter and exit a position the same day. For swing and position trading, your daily statement will probably show only half the commission (one "side"), because a position may be held for a number of days. A system-assist broker might charge from $30 to $50 per round-turn to trade a full-sized contract for you. For trading a mini contract, the commission might be the same, or could be as little as half the full-size commission. Note that the commission setting for a TradeStation report is per side, so the commission per round-turn will be double the setting.
Fees are usually charged for each trade. They may be included in the commission, or charged separately. Ask your broker. Fees (e.g. clearing, brokerage, exchange, transaction, regulatory, NFA) may total around $3.50 per trade.
A system developer will usually publish the hypothetical back-test results for their systems. Often this data is made available on their website. Another source of hypothetical data is Futures Truth magazine, which is published every two months and contains post-release hypothetical results, on a monthly basis, for about 200 systems. As noted two sections above, pre-release hypothetical results should be used with caution.
Some system-assist brokers publish actual-trade results on their websites. Actual-trade results may differ from hypothetical results for many reasons. Actual-trade results include actual slippage, rather than an estimated value as used in hypothetical results. Some brokers may provide better "execution" than others (i.e. lower slippage). In actual trading, some trades may be missed, e.g. because the computer running the system was down, or because a new version of the trading system was being installed and was not yet ready. Trades may be missed because the exchange is down - for example the Globex exchange is computer-based and has suffered a number of outages. An exchange outage can be particularly troublesome if you are "in" a losing trade when the outage occurs, and cannot get out. In this case, some brokers will attempt to minimize your loss by taking an offsetting position in another vehicle; for example the mini-Dow might be used to offset a position in the mini-S&P when Globex is down. Actual-trade results may also differ from hypothetical because the trader decided to not take some trades. This may be more likely to happen with position-trading systems, when the trader may be actively following the market and the trading signals, and feel that a particular trade is likely to be a loser, and instruct their broker to skip that trade.
Human emotion is a major consideration to keep in mind when looking at previous results, whether hypothetical or actual-trade. When you are looking at the equity curve for a system that has done well, please bear in mind that these results are for uninvolved trading of a fixed number of contracts throughout a particular period of time. In the real world, however, human factors and emotions are in play, and there are decisions to be made:
When you are looking at trading system results, it may be important to understand the difference between closed equity and open equity. Closed equity is a measure or profit-or-loss after a trade is closed (i.e. there is no longer a long or short position in the equity). Open equity is the value of a position, relative to the entry price. Open equity varies from minute to minute through the trading day, but is often reported only at the end of the day or at the end of the month. Position-trading results on Futures Examiner usually include the open equity at the end of each month. For day-trading results, there is no open equity at the end of the day or month. For swing-trading systems, there is often an open equity at the end of a day or a month; for reporting purposes, however, we usually report a swing trade that straddles a month in one of the two months, and ignore any open equity at the end of the month.
Trading systems are great when they're making money. But it can be quite uncomfortable for the trader, and even painful, when they are losing money. When a system loses money this is called drawdown. Drawdown is measured from an equity high to an equity low (or to the present when a system is currently in drawdown). For example, if the cumulative profit for a system peaked at $35,000, and then it lost $7,000, the system would be in a $7,000 drawdown (and the cumulative profit would drop to $28,000). If the system then lost another $4,000, the drawdown would increase to $11,000 (and the cumulative equity would decrease to $24,000). Drawdown can be measured in a number of ways. The two that we report are as follows.
The more data that is available, the larger the maximum observed drawdown is likely to be. Some observers caution that the worst drawdown is likely yet to come. So one should be careful using the maximum observed drawdown when there is only a year or two of data.
We also report (on the Statistics page) the maximum end-of-month drawdown for each 12-month period for which we have data. The highest such number should be equal to the maximum end-of-month drawdown which was described above. For systems with two or more years of data, these numbers can provide a feel for how much pain (i.e. drawdown) a trader might have experienced each year.
Mini contracts are available for some commodities, and are one-half to one-fifth the size of the big contracts. For example, the mini S&P 500 (ES) is one-fifth the size of the big S&P (SP), and has become a popular trading vehicle for traders with small accounts. Most trading systems that are designed for big contracts can also be traded with minis, but the results are usually not as good. Usually the data signals for the big contract are fed into the trading system software, and its "buy" and "sell" signals are applied to both big and mini contracts. Sometimes, however, the data signal for the mini contract is fed into the trading system software to provide "buy" and "sell" signals that apply specifically to minis. Also, there could be a "mini" version (or parameter settings) for a system.
Minis are usually traded electronically, whereas the big contracts are usually pit traded. For example the mini S&P is traded on the Globex exchange. The price for a mini usually tracks that for a big contract fairly closely, but in times of sudden change they can vary significantly in price. And the Globex exchange, being computer-based, has sometimes gone "down" such that minis could not be traded. This can be bad if you are in a losing mini trade at the time!
The slippage with a mini S&P is often less than with a big contract, perhaps by a quarter point or so, but the larger commission usually more than offsets this. For some systems, we provide a "computed mini" component, whose results are based on results data for the big contract, but are adjusted for the different slippage (estimated) and commission. This computed mini provides a rough idea of how a mini might have performed, but is only an approximation, and should be used with caution.
How can a trading system be evaluated? Ideally, everyone would like an investment that would make a consistently high percentage gain, month after month, with no losses. This would produce a straight-line equity curve - one that went up at a steady rate, e.g. 10% per year or 20% per year. Unfortunately, most straight-line investments, such as Certificates of Deposit, or Guaranteed Investment Certificates, have a rather low yield, perhaps from 3% to 5% per year currently.
When we talk about a "trading system", we must bear in mind that some systems are designed to trade only one commodity, while others are designed to trade several or many commodities. For example R-Mesa is a day-trader that trades only the S&P 500 index, while Trendchannel is a position trader that trades many commodities.
One simple measure that can be used as a quality factor is the ratio of the average annual profit to the average maximum annual drawdown. When this ratio is computed for the most recent 3 years, it is called the Sterling ratio. This is a sort of "pleasure/pain" ratio, and the higher the number the better. For some of the better day-traders, this ratio is typically greater than one. For example for R-mesa the ratio is around 1.5 based on 24 months of data. For many of the better position-traders, this ratio is typically less than one for many individual commodities, but greater than one for a basket of three or six or more commodities. For example for Trendchannel, the ratios (2004-Jun) for the five commodities that we report, based on 72 months of data, range from 0.5 to 3.4 with an average of 1.2. Three of the ratios are less than one. By comparison, the ratio is 2.9 for a basket that contains one contract for each of these commodities, which is excellent (note that the EuroDx component is based on three contracts). This measure should be used with caution, because the drawdown numbers are based on sliding 12-month periods, and the numbers may go either up or down as time progresses.
Some other measures that can be used to evaluate trading systems (or any other type of investment) are as follows.
There are many other metrics that are published for trading systems, such as profit/loss ratio, maximum losing and winning trades, percent winners, points /trade, etc. These metrics are usually included in the standardized system backtest reports that are often available from system developers.
There is likely no single best system. Each system has its own pros and cons. One may have a greater percentage profit but greater drawdowns. For two systems that have about the same percentage profit, one may have a higher Sharpe ratio, and the other may have a greater Sterling ratio. This is where portfolio design comes in - choosing a combination of systems that complement each other in a way that produces both a good percentage profit and a smooth equity curve.
How does one develop a Balanced Portfolio? You may often hear the advice that a well-balanced portfolio should be diversified over commodities, methods, and time-frames. This is a simple guideline intended to facilitate designing a portfolio with a smooth equity curve. Another guideline is to combine components whose correlation with one another is low. (We may report correlations in the future). One of the key features of FuturesExaminer is that you can experiment with portfolio design by composing baskets that contain different combinations and weightings of components. But please keep in mind the often-seen caution: past performance is no guarantee of future results.
Is the system broken? This is a question that traders are likely to ask when a system goes into drawdown. Systems with great past performance can go into very painful drawdowns. (This often happens just after I begin trading it!) One way to attempt to answer this question is by comparing the current drawdown with the maximum historical drawdown. If the current drawdown is less, then the system may not be broken. This comparison is more meaningful if the maximum historical drawdown is based on many years of data and a large number of trades. When you begin trading a system, you should be anticipating that the worst drawdown is yet to come. Some observers recommend setting, at the outset, a maximum drawdown beyond which they would stop trading a system.
What is your pain tolerance? Perhaps this question should be kept in mind along with the previous question about a system being broken. It is very tempting to "load up" your trading (i.e. trade more contracts) when a system is doing well, but sometimes this is just the time that a system is likely to go into a major drawdown (ouch!!). This is where adequate account funding is important. We use a "conservative" account size of 5 times margin as a consistent measure for all systems. But more realistically, the determination of account size should take into account how diversified the portfolio is, what level of drawdown the trader is prepared to endure, and a projected future maximum drawdown based on the maximum historical drawdown.
I hope that this article has been informative and helpful, and has better equipped you to understand the rest of the material on Futures Examiner. If you have any suggestions on how to improve this article (or anything else on this website), please let me know!
Ken Morin
Last updated 2006-Feb-25