Committed to quantitative strategy development, high-quality community, trading ideas sharing and other related content
Hi, my name is Le Chiffre.
What I bring to you today is the 6th super trend line series. The entry and exit of this issue is different from the previous one. The previous super trend line consists of a moving average with different algorithms and an interval constructed by ATR volatility.
The following shortcomings were found after observation and summary:
1. Due to algorithmic reasons, the ATR component in the strategy causes the volatility to amplify sharply during high-level AV fluctuations, which leads to a lag in the reversal of the market entry. Stop loss is too large, resulting in losses and retracements.
2. The main body of moving averages with different algorithms in the past showed the situation of synchronization and lag, such as: Kaufman adaptation of SF29, variable exponential dynamic average of SF35, variable moving average of SF36, etc.
In order to correct the above two flaws and shortcomings, I thought and studied SF37 - Super Trend Line 6. The underlying main moving average of this strategy was obtained by John Ehlers and Ric Way, two predecessors who did not know how to study it, and the name of the moving average algorithm was Zero Lag Exponential Moving Average, which is a zero-lag exponential moving average. The indicator visualization is shown in the following figure:
Compared with the previous moving average algorithms, the characteristic of this indicator is that at some market inflection points or inflection point time periods seen after the fact, with the interpretation of the market, the moving average appears the fastest initial speed compared to other moving average algorithms.
Second, this strategy does not use the ATR as the fluctuation range of the algorithm, but I share the volatility characterization mentioned in "Heterogeneous Communities Quantitative Research 3" in "Heterogeneous Communities". algorithm. Why do you do this?
First of all, according to my observation, the ATR algorithm includes the data of the highest price and the lowest price, and the highest price and the lowest price themselves belong to the extreme values of emotions or random walks, and the reference is not meaningful, but the market is objective, and the futures market Unlike other markets, you can erase the K line at will. Therefore, after thinking and comparing again and again, the algorithm is replaced, and only the size of the K-line entity is obtained to describe the volatility, but there is a missing place in this logic, which is not a shortcoming, that is, the gap is ignored in the volatility structure. proportion.
Secondly, through visual qualitative analysis, the correlation coefficient between the overall fluctuation structure of the new method and the ATR fluctuation is about 0.9, but the overall size is generally smaller than the ATR, and the magnitude of the fluctuation varies between 0.3 and 0.7. The visualization screenshot is as follows:
The above three pictures are from different time periods and with different structural characteristics, and the resulting picture ATR and the fluctuation visualization of the research shared by our community. While the overall volatility is the same, the details do show different results.
Core essence:
1. Abandon the high and low opening, as well as the highest and lowest prices. The characterization of volatility does not mean how big the absolute number is, but the relative volatility of volatility.
2. In the process of volatility amplification, especially in the skyrocketing market, high-level shocks and reversals will cause volatility to amplify, and this amplification of volatility will lead to a lag in the reversal of the super trend line.
The structural characteristics of entering the market are finished. Let’s talk about the appearance. On the basis of the SF35 adaptive Krange, I added the acceleration to the appearance. To put it bluntly, it is to describe the fluency of the trend. ”) to accelerate the trailing take profit and stop loss line to ensure that the exit AV market leads to profit taking.
Let's take a look at each performance, the visualization is as follows:
combination
J
EB
Ni
JD
Mandarin-coke-long
Mandarin-Coke-short
Vnpy thread performance, no selection parameters, everyone privately tests the selection parameters
Due to the differences between platforms, the backtest performance is subject to the TBQ version
This strategy is only used for learning and communication, and investors are personally responsible for the profit and loss of real trading.