Lm10 cosine wave homeopathic grid strategy

  Quantitative strategy development , High quality community , Trading ideas sharing and other related contents

『 Text 』

ˇ

Hello everyone , Today we share heterogeneity CTA The first 10 piece , The first of the grid inversion oscillation series 3 Strategy content .

In this article, we examine the strategy construction from the perspective of risk control , In the first two articles, we constructed pure price grid shocks and reversals , And the oscillation and inversion of Fisher inverse transform index . But through the near 2 Months of observation found , The hidden risk of this strategy is not small , Although the overall back test is upward , But it is not difficult to see from the details ,“ Scalp extraction ” The risk of the strategy breaking out at a certain time point is very confusing to Xiaobai . As shown in the figure below ：

Today, what we want to solve is the time point risk outbreak of this scalp pulling strategy , We adopt the idea of trend mode and cycle mode , Combined with the phase theory in trigonometric function , Integrate into the transformation cycle 、 Trend oscillation algorithm .

One 、 Strategy background

There is a cycle in the trend , The cycle deducts one trend by one . The market in the ideal world is probably like this . All we have to do is find the trend in the cycle to do . Here are the trends in the real world + Cycle period , As shown in the figure below ：

secondly , Is the phase in the trigonometric function , The source is the Fourier series , Because Fourier series uses trigonometric functions , Trigonometric function is to generate signals , This signal is not a signal for us to open and close positions , It's a wave in physics , Or electromagnetic waves , Because electromagnetic waves use trigonometric functions . And the core of what we have to do —— Fuse the signal into the opening and closing position signal . therefore , These are used behind the scenes .

Phase occurs in periodic motion , The most direct understanding of phase is angle , This angle exists in circular motion .（ I won't talk about Fourier series , These are all Fourier series ）

As shown in the figure above , Suppose a moment T, Simple harmonic motion in large P spot , The corresponding circular motion is the small one in the figure p spot , Connecting with the dot is OP, The included angle is θ, That is, phase .

      

Phase formula ：θ = wt+φ;

      

When T=0 That is, the initial moment , that θ = φ. Big P Do simple harmonic motion , Small p The displacement is above the circumference . The coordinates can also be calculated according to the above formula . Simply speaking ,w It's the frequency ,t Is time , the last one φ Is the initial phase angle （ namely t=0 When , Only φ）. Phase is (ωt+φ).w be equal to 2π Divided by the period T.

If we tile the trajectory of the circle on the time axis , Now what you get is a sine function , As shown in the figure below ：

All the above are high school students 、 College knowledge , Those who don't understand Baidu first , Ask me if you still don't understand .

After the above picture comes out , You know what I'm going to say , Look at the sine function and the phase evolution , Plus the picture I mentioned above .

Two 、 Policy code

A brief explanation , This beta In fact, it can be seen as w,delta Represents the bandwidth of the filter , In direct proportion to . The advantage is that the filter price will be more responsive , Relatively in the rapid change of the cycle .

And then , We filter the processed signal with a low-order filter , As shown in the figure below ：

The above logic has been quantitatively studied in heterogeneous communities 14 Share in . Members of the group can browse the original translation documents , Look up and understand the logic .

The logic of the appearance is SF37 Of Krange Adaptive appearance , Part of the code is shown in the figure below ：

3、 ... and 、 visualization

national debt +LPG

The above only tests the national debt （T、TF） and LPG Three varieties , Because of the time problem, we can take it to the work area and test it by ourselves .

Let's take a look at the visualization of several transaction records , As shown in the figure below ：

T short

T long

Lpg long

summary ：

1、 This strategy is routine CTA Exit logic , There is no heterogeneity , It can be modified here

2、 The policy algorithm has a certain period 、 The case that the series of a group of parameters is not convergent , This kind of optimization must be observed visually . for example ：VNPY Running directly is fruitless , To optimize , Because the default parameter probability is divergent .

3、 This strategy, especially treasury bonds, is used to follow LM09 For combination , The aim is to avoid unilateral risks , So the strategy heterogeneity lies in two points ： The first is to enter the site , But the entry heterogeneity . The second is to follow LM09 Fisher inverse transform oscillation algorithm inversion combination used together .

4、 Good reversal 、 It's hard to have a good variety of trends , National debt is one of them . National debt is used by everyone 20 The cycle moving average is profitable in each cycle , It shows that it has periodic momentum , But the national debt “ Mean reversion ” It's also very strong , Late entry of trend strategy will be beaten , Enter the market early, stop the loss on the left and come out . Therefore, in this strategy, we suggest that we mainly anchor the national debt as a portfolio .

Due to the differences of various platforms , Back test performance to TBQ The version shall prevail ！！！

This strategy is only used for learning and communication , The investor is personally responsible for the profit and loss of the firm offer .