Empirical Mode Decomposition Based Quantitative Trading Strategy

Author: ChaoZhang, Date: 2023-12-22 14:41:34
Tags:

Overview

This strategy is based on the Empirical Mode Decomposition (EMD) method to decompose the price series and extract features from different frequency bands, combined with the mean to generate trading signals. It is mainly applicable for medium and long term holdings.

Strategy Logic

Use the EMD method to bandpass filter the price and extract price fluctuations
Calculate the moving average of peak and trough sequences
Generate trading signals when the mean line exceeds a certain percentage of peak and trough lines
Long or short based on trading signals

Advantage Analysis

The EMD method can effectively decompose the price series and extract useful features
The peak and trough lines control the strategy to trade only when the price fluctuation is greater than a certain amplitude
Combined with the mean line, it can effectively filter out false breakouts

Risk Analysis

Improper selection of EMD method parameters may lead to overfitting
It takes a long cycle to form a transaction signal and cannot adapt to high frequency trading
Unable to cope with market conditions with dramatic price fluctuations

Optimization Directions

Optimize the parameters of the EMD model to improve adaptability to the market
Combine other indicators as stop loss and take profit signals
Try different price series as strategy input

Summary

This strategy uses the empirical mode decomposition method to extract features from the price series and generates trading signals based on the extracted features, realizing a stable medium and long term trading strategy. The advantage of this strategy is that it can effectively identify periodic features in prices and issue trading orders during large fluctuations. But there are also certain risks, and further optimization is needed to adapt to more complex market environments.

/*backtest
start: 2022-12-15 00:00:00
end: 2023-12-21 00:00:00
period: 1d
basePeriod: 1h
exchanges: [{"eid":"Futures_Binance","currency":"BTC_USDT"}]
*/

//@version=2
////////////////////////////////////////////////////////////
//  Copyright by HPotter v1.0 12/04/2017
// The related article is copyrighted material from Stocks & Commodities Mar 2010
// You can use in the xPrice any series: Open, High, Low, Close, HL2, HLC3, OHLC4 and ect...
//
// You can change long to short in the Input Settings
// Please, use it only for learning or paper trading. Do not for real trading.
////////////////////////////////////////////////////////////
strategy(title="Empirical Mode Decomposition")
Length = input(20, minval=1)
Delta = input(0.5)
Fraction = input(0.1)
reverse = input(false, title="Trade reverse")
xPrice = hl2
beta = cos(3.1415 * (360 / Length) / 180)
gamma = 1 / cos(3.1415 * (720 * Delta / Length) / 180)
alpha = gamma - sqrt(gamma * gamma - 1)
xBandpassFilter = 0.5 * (1 - alpha) * (xPrice - xPrice[2]) + beta * (1 + alpha) * nz(xBandpassFilter[1]) - alpha * nz(xBandpassFilter[2])
xMean = sma(xBandpassFilter, 2 * Length)
xPeak =  iff (xBandpassFilter[1] > xBandpassFilter and xBandpassFilter[1] > xBandpassFilter[2], xBandpassFilter[1], nz(xPeak[1])) 
xValley =  iff (xBandpassFilter[1] < xBandpassFilter and xBandpassFilter[1] < xBandpassFilter[2], xBandpassFilter[1], nz(xValley[1])) 
xAvrPeak = sma(xPeak, 50)
xAvrValley = sma(xValley, 50)
nAvrPeak = Fraction * xAvrPeak
nAvrValley = Fraction * xAvrValley
pos = iff(xMean > nAvrPeak and xMean > nAvrValley, 1,
	     iff(xMean < nAvrPeak and xMean < nAvrValley, -1, nz(pos[1], 0))) 
possig = iff(reverse and pos == 1, -1,
          iff(reverse and pos == -1, 1, pos))	   
if (possig == 1) 
    strategy.entry("Long", strategy.long)
if (possig == -1)
    strategy.entry("Short", strategy.short)	   	    
barcolor(possig == -1 ? red: possig == 1 ? green : blue )
plot(xMean, color=red, title="Mean")
plot(nAvrPeak, color=blue, title="Peak")
plot(nAvrValley, color=blue, title="Valley")

More

Forums
|
API
|
Github
|
About
|
SiteMap