Trading in Sample Sizes. The typical trader practically lives or dies (emotionally) on the results of the most recent trade. If it was a winner, he'll gladly go to the next trade; if it wasn't, he'll start questioning the viability of his edge. To find out what variables work, how well they work, and what doesn't work, we need a systematic approach, one that doesn't take any random variables into consideration. This means that we have to expand our definition of success or failure from the limited trade-by-trade perspective of the typical trader to a sample size of 20 trades or more. Any edge you decide on will be based on some limited number of market variables or relationships between those variables that measure the market's potential to move either up or down. From the market's perspective, each trader who has the potential to put on or take off a trade can act as a force on price movement and is,therefore, a market variable. No edge or technical system can take into consideration every trader and his reasons for putting on or taking off a trade. As a result, any set of market variables that defines an edge is like a snapshot of something very fluid, capturing only a limited portion of all the possibilities. When you apply any set of variables to the market, they may work very well over an extended period of time, but after a while you may find that their effectiveness diminishes. That's because the underlying dynamics of the interaction between all the participants (the market) is changing. New traders come into the market with their own unique ideas of what is high and what is low, and other traders leave.Little by little, these changes affect the underlying dynamics of how the market moves. No snapshot (rigid set of variables) can take these subtle changes into consideration. You can compensate for these subtle changes in the underlying dynamics of market movement and still maintain a consistent approach by trading in sample sizes. Your sample size has to be large enough to give your variables a fair and adequate test, but at the same time small enough so that if their effectiveness diminishes, you can detect it before you lose an inordinate amount of money. I have found that a sample size of at least 20 trades fulfills both of these requirements.Testing. Once you decide on a set of variables that conform to these specifications, you need to test them to see how well they work. If you have the appropriate software to do this, you are probably already familiar with the procedures. If you don't have testing software, you can either forward test your variables or hire a testing service to do it for you. In any case, keep in mind that the object of the exercise is to use trading as a vehicle to learn how to think objectively (in the market's perspective), as if you were a casino operator. Right now, the bottom-line performance of your system isn't very important, but it is important that you have a good idea of what you can expect in the way of a win-to-loss ratio (the number of winning trades relative to the number of losing trades for your sample size).Accepting the Risk. A requirement of this exercise is that you know in advance exactly what your risk is on each trade in your 20- trade sample size. As you now know, knowing the risk and accepting the risk are two different things. I want you to be as comfortable as possible with the dollar value of the risk you are taking in this exercise. Becuse the exercise requires that you use a 20-trade sample size, the potential risk is that you will lose on all 20 trades. This is obviously the worst-case scenario. It is as likely an occurrence as that you willwin on all 20 trades, which means it isn't very likely. Nevertheless, it is a possibility. Therefore, you should set up the exercise in such a way that you can accept the risk (in dollar value) of losing on all 20 trades. For example, if you're trading S&P futures, your edge might require that you risk three full points per contract to find out if the trade is going to work. Since the exercise requires that you trade a minimum of three contracts per trade, the total dollar value of the risk per trade is $2,250, if you use big contracts. The accumulated dollar value of risk if you lose on all 20 trades is $45,000, You may not be comfortable risking $45,000 on this exercise. If you're not comfortable, you can reduce the dollar value of the risk by trading S&P mini contracts (EMini). They are one-fifth the value of the big contracts, so the total dollar value of the risk per trade goes down to $450 and the accumulated risk for all 20 trades is $9,000. You can do the same thing if you are trading stocks: Just keep on reducing the number of shares per trade until you get to a point where you are comfortable with the total accumulated risk for all 20 trades. What I don't want you to do is change your established risk parameters to satisfy your comfort levels. If, based on your research, you have determined that a three-point risk in the S&Ps is the optimum distance you must let the market trade against your edge to tell you it isn't worth staying in the position, then leave it at three points. Change this variable only if it is warranted from a technical analysis perspective. If you've done everything possible to reduce your position size and find that you still aren't comfortable with the accumulated dollar value of losing on all 20 trades, then I suggest you do the exercise with a simulated brokerage service. With a simulated brokerage service, everything about the process of putting on and taking off trades, including fills and brokerage statements, is exactly the same as with an actual brokerage firm, except that the trades are not actually entered into the market. As a result, you don't actually have any money at risk. A simulated brokerage service is an excellent tool to practice with in real time, under real market conditions; it is also an excellent tool for forward testing a trading system. There may be others, but the only service of this nature that I know of is
Source:"Trading in the zone" by Mark Douglas

Trade well.

Translate to English Show original