Many individual traders will choose the trend tracking strategy when designing their own trading system.
There are two methods to build trend tracking system: one is price breakthrough, the other is linear crossover.
The friends who use price breakouts to track the trend usually seek some typical prices, such as the high and low points in the period, or the highest and lowest points in the recent trading days (such as the 20 day high and low points of turtle rule). Once the price breaks through this kind of typical price, open a position to follow up the transaction.
The use of linear cross for trend tracking mainly uses the average price cross of different time periods. Such as the double moving average cross after the closing confirmed open follow-up. The crossing of price to single moving average is the crossing of MA (1) and single moving average MA (n), which is also a deformation of double moving average crossing. Or three moving average cross filtering. Or use MACD as a trend signal. In essence, MACD is also a deformation of the intersection of the two moving averages. If the zero axis is used to filter the cross signal, it is actually the deformation of the three moving average cross filtering.
Many friends design their own trend trading system for back testing, the results are often unexpected. It is clear that there is a trend in the market, but why does the back test data not show a considerable return?
The usual solution is system optimization. The first question is the parameters. Is this parameter used by everyone, so it fails? Try to try to find that the results of various parameter combinations are similar. Parameters that perform well in one period perform poorly in another. Parameters that perform well in one variety have little profit in another.
So the system is optimized again. Is it better to add some restrictive conditions? Such as the number of floating loss after the mandatory stop loss, or the number of floating win after the stop profit or partial stop profit. But the effect is still not satisfactory. Some profitable trades have been stopped. Some transactions that can obtain excess returns are also limited by profit stopping.
Once again, the optimization will basically involve the position problem. For example, after continuous profit or net worth growth, open position will be reduced. After a series of losses, trading is increased.
In this way, a trend tracking system slowly embarked on the routine of data fitting.
If you are now facing such a distress, I would like to remind you to pay attention to whether you have entered a trend tracking error.
The mistake is to use trend signals as trading signals.
We can think that in a certain period of time, if the price breaks through a typical high, it is a rising trend signal.
Similarly, we can also think that in a certain period of time, if a combination of moving average parameters has a golden fork, then it is a rising trend signal.
So, can the upward trend signal be used as a long trading signal?
When we believe in the power of trend and track trend trading, in essence, what we trade is the process of trend operation, not the point where the trend forms and ends.
In other words, the golden fork of the two moving average is the time point of the formation of the upward trend, and the dead fork of the two moving average is the time point of the end of the upward trend. In these two time points, no matter how prices run, the upward trend is certain.
Therefore, if the trend trading, we can draw a conclusion: in these two time points to do a long single. And we can only make long orders. Therefore, trend signal is the prerequisite or necessary condition of trading signal.
However, is it possible to make a long order at any time in these two time points?
The answer is obviously No. This is related to the loss to profit ratio. We can all understand that it is obviously much more risky to be long in the late stage of the trend than in the early stage of the trend.
So the question is, the position where the golden fork is formed is obviously the initial stage of the trend formation, so can we use the trend signal as the trading signal directly?
Before answering this question, let’s deconstruct the premise that trend tracking can be profitable.
If you want to make a profit in a trend tracking transaction, you need two prerequisites: a large trend running space, and a long trend running time.
We all have the experience that if a trend runs for a long time, but there is not much space, it is likely to be a larger cycle of adjustment, and the final profit of this trend transaction will be quite limited.
Another situation is that after the trend is formed, there is a large range of space to run, but the trend duration is extremely limited, the fall is faster, and the market is determined quickly. In this case, tracking trend trading is likely to lose out. It’s even possible to hit back and forth in the face.
Ideally, slow bull and negative market. The formation of the trend of low-key, after a long time, out of wide amplitude, and finally can end in a low-key. However, such opportunities are very rare.
Unfortunately, the trend signal can provide us with information about the direction of the trend, but it can not tell us the time and space of the trend in this direction. Even if we describe it more carefully, the information that trend signals can provide us is only the possibility of trend direction.
In the case of sufficient samples, about 1 / 3 of the trend signals are invalid false signals, which are caused by narrow fluctuation or over adjustment of prices. There is no profit margin for such signals, some of which can be identified and filtered by trading experience.
There are about 1 / 3 signals. Although they correctly point out the direction of the trend, either the space is limited or the duration is short, and they can barely maintain the break even on the whole.
The remaining 1 / 3 signals are really valid trend signals. This is the main profit source of trend tracking system. However, it is extremely rare to be able to give 3200 point amplitude as shown in the figure above and eat up 80% of the whole amplitude.
It is precisely because of this 1 / 3 effective signal that some trend traders believe that as long as the profit / loss ratio of 3:1 is guaranteed, it is OK. Compared with paying a higher position cost, they are more worried about missing every trend signal that may trigger a huge market. As a result, they prefer to use trend signals as trading signals.
If every trend signal is tracked, the winning rate will be locked at 30%. It is difficult to control whether we can get an average profit loss ratio of 3:1 in these 30% transactions. Moreover, in the long process of trend tracking, we will definitely meet the test of black swan. The damage of such risk events to account funds can not be controlled by historical back testing.
Even though the profit and loss ratio of 3:1 is maintained, and fortunately it has not been hurt by the black swan, the problem is that the profit and loss ratio is only enough to protect the capital. Where does the profit come from?
Some friends will give examples of turtle trading. The turtle rule is great, but if you read it carefully, the author makes it clear that if money is limited, it will be difficult to replicate this success. If you only focus on one market, you are likely to be unprofitable in the face of 70% of the time, or even hang up before the trend comes. If we focus on multiple unrelated markets at the same time, the huge demand for principal will become an obstacle.
That’s why the trend system of some stocks is very good, and the actual trading is blind. Back testing assumes that there are enough funds to track every profit point. In the case of a firm offer, a few bullets were fired out, leaving only a dry stare.
Therefore, in the case of limited funds, when designing the trend tracking system, try not to simply and rudely use the trend signal as a trading signal. Generally speaking, in order to facilitate the execution, the mechanized trend tracking system will directly use the trend signal as the transaction signal, and will enforce each transaction signal. In this way, we can avoid missing the effective market, but in the long run, we can only obtain the effective market β profit.
If trend traders want to get α Income, it is necessary to use some small strategies to improve the odds or break even ratio. The problem, however, is that these little strategies are hard to mechanize. That’s why I prefer manual trading.