But this will make traders take all kinds of shortcuts, so that they may make all kinds of mistakes when making trading strategies. Although the short-term trading method is simple, the mistakes caused by shortcuts will eventually affect real-time trading.
Fortunately, most traders' mistakes in strategy formulation are basically similar, and once they are determined, they can be corrected and improved. However, it is not easy to eliminate these problems; Without shortcuts, it will become more difficult to formulate and implement strategies correctly.
So, what are the most common mistakes and shortcuts? There are roughly three kinds. Let's explain the negative effects of these three kinds of mistakes and how to find and correct them. Avoiding these common mistakes will help to build a better trading system.
Trap 1: The strategy is too complicated to understand.
In the process of trading, you will inevitably encounter some complicated trading strategies. For independent traders, there may be "analysis paralysis": it can be clearly seen that the following figure is full of various technical indicators, technical lines and support and resistance areas. For an algorithm trader, a complex method will contain thousands of lines of code, and dozens or even hundreds of variables need to be optimized and adjusted.
These two methods have one thing in common: they are extremely complicated and involve a lot of content. Many inexperienced traders will think that this is a way to develop systems; They believe that the more indicators, the better the fitting degree of the algorithm to the past data, and the better the trading strategy. But this is not the case.
There is a fallacy in measuring complex strategies, which leads us to the wrong conclusion that this strategy is superior to the general strategy: a strategy that can achieve better results in historical data does not mean that this strategy can also achieve the same success in actual transactions. In fact, the strategy is constantly adjusted by adding indicators or adding new rules to the algorithm, which makes the trading strategy more complicated; This practice usually only gives traders a false confidence. Improving and adding more rules in the strategy does not mean that the strategy will become better.
Many people may find it hard to believe that simple trading strategies are usually the best. For independent traders, a relatively simple chart with only one or two technical indicators and a deep understanding of price trends and market dynamics is often much better than a chart full of technical lines and technical indicators. For algorithmic traders, a simple entry order is usually better than a rule that requires 5 to 10 conditions to execute a transaction.
Trap 2: Trading strategy without considering market friction
Comparing the trading systems on the market, you will find that most trading systems will list a short exemption clause: commission and slip are not included. Similarly, many people who develop their own systems will also ignore commissions and sliding costs; Even if they consider these costs, they usually underestimate the actual amount.
You will hear various reasons that the system does not include commission and slippage. The most common excuse is that "different brokers usually charge different commissions"; Another common reason is that "my system only uses limit orders". But the real reason is that it can make the system look better. If the actual transaction cost is considered, it is difficult to find a profitable system.
Take the trading system of E-Mini Standard & Poor's 500 Index Futures (CME:ESM 14) as an example. This system uses a technology to capture very small market changes through small transactions and frequently run intraday trading hedging to run speculation. If the commission and slippage are not considered, the system runs 20 transactions every day, and the average profit per transaction is 15 USD. Traders see a profit of $300 a day and think these deals are not bad. However, if you add the commission of $5 per round of trading and the slip point of 65,438+0 basis points (which may be an optimistic estimate), the daily profit of $300 becomes a daily loss of $50.
For a trading strategy that does not consider commission and sliding cost, one potential impact is that the system may make traders run too many trades. Here is an example. Suppose system A is the small-scale speculation system mentioned above, and it can make a profit of $300 a day if commission and sliding cost are not considered. Relatively speaking, system B only runs one transaction every day, regardless of the transaction cost, and each transaction can earn an average of $50. Anyone who compares the two systems will choose system A, however, with commission and slippage, the result is just the opposite; System B is the only valuable strategy; The number of transactions is less, and the proportion of commission slip points to total profits is also much smaller.
Therefore, in the early stage of system development, it is very important to consider certain commission and sliding cost. For E-Mini Standard & Poor's 500 index futures, a relatively reasonable assumption is a commission of $5 per round and a slip point of one or two basis points.
Trap 3: System testing uses all historical data.
The third mistake many traders may make when developing a system is to use all available historical data when testing the system. Most inexperienced traders will optimize and analyze all historical data before today. This is because they want to ensure that the policy reflects the adjustment of the latest data.
Of course, if the first test fails, the trader will add some rules or filter conditions to the system (so it is more likely that the strategy mentioned above is too complicated) and then run all the data again.
Finally, traders will find a feasible trading strategy system and apply this system to actual trading. But when there is a problem with trading strategy, this method is almost always an inevitable solution.
A better but more difficult method is to verify the trading system by testing the data outside the sample. For example, a trader may use the data of the past 10 years to make a trading strategy, but keep the data of the latest year. When developing the system, run the test with unknown (i.e. out-of-sample data); If the system works well, then the system can be used for real-time trading.
In addition, you can also test the mobile phone pane. By using multiple time periods outside the sample, this method is more likely to succeed, because the generated stock curve has completely included the optimization results outside the sample.
One disadvantage of moving pane test or out-of-sample test is that once the out-of-sample test is run once, no further tests will be run based on the real out-of-sample data. Therefore, if the test is run many times, it is easy to inadvertently turn the out-of-sample test into the in-sample test. However, the method of out-of-sample testing is better than the method of running optimization on all data.
There is no shortcut to system development.
It is quite difficult to design a feasible trading strategy. In fact, many traders have never really done this; Many times, they will take shortcuts when developing the system or make oversimplified mistakes in the process of system development. Of course, it is much easier to add one rule after another and one filter condition after another in the system than to find a suitable and concise rule.
Similarly, it is quite easy to find a feasible strategy without considering the commission and the friction cost of slipping.
Finally, compared with moving pane test or out-of-sample test, running optimization on all available historical data is a relatively simple method, and the results will be better.
But the crux of the problem is that if we establish a trading strategy system by optimizing all data operations, it is equivalent to designing such a trading strategy: only by applying this trading strategy system in the test time sample can we make a profit. Of course, this requires a time machine, which is obviously much more complicated than building a profitable system.
The enlightenment we get from the common mistakes in the above-mentioned system development is that if a certain method makes the system design simpler, or the success rate of backtesting is obviously improved, then this is actually an early warning signal that errors may occur. A reasonable and suitable system development process is always full of difficulties.
However, in the long run, it is better to develop a trading system in the right way than to lose money in the market because of the mistakes in system development.