Test Trading Performance Based on MACD (Python)

November 3, 2020

algo-trading

Note

Use MACD signal only: Buy when Signal line cross above MACD line (Postive Histogram), Sell when Signal line cross below MACD line (Negative Histogram)
Plot candle chart and MACD chart for better visualization
Plot buy and sell points on the chart
Print buy/sell transactions and profit/loss
Meassure performance

NOTE: The Best Times to Use the MACD Indicator

Code

Install libraries

pip install yfinance
pip install mplfinance

Refer Install Python TA-Lib (Ubuntu)

import yfinance as yf
import mplfinance as mpf
import talib as ta
import numpy as np
import pandas as pd
import math

ticker_name = 'MSFT'
yticker = yf.Ticker(ticker_name)
data = yticker.history(period="1y") # max, 1y, 3mo

# macd
data["macd"], data["macd_signal"], data["macd_hist"] = ta.MACD(data['Close'])

# plot macd
macd_plot = mpf.make_addplot(data["macd"], panel=1, color='fuchsia', title="MACD")
colors = ['g' if v >= 0 else 'r' for v in data["macd_hist"]]
macd_hist_plot = mpf.make_addplot(data["macd_hist"], type='bar', panel=1, color=colors) # color='dimgray'
macd_signal_plot = mpf.make_addplot(data["macd_signal"], panel=1, color='b')

# buy/sell
buy_signals, sell_signals, signals = detect_macd_signals(data)

# plot buy/sell
buy_plot = mpf.make_addplot(buy_signals, type='scatter', marker='v', markersize=100, panel=0)
sell_plot = mpf.make_addplot(sell_signals, type='scatter', marker='^', markersize=100, panel=0)

# print buy/sell transaction and stats
print_performance_summary(signals)

# plot candle chart and all
plots = [macd_plot, macd_signal_plot, macd_hist_plot, buy_plot, sell_plot]
mpf.plot(data, type='candle', style='yahoo', addplot=plots, title=f"\n{ticker_name}", ylabel='')

def detect_macd_signals(data):
    """Use MACD cross-over to decide buy/sell

    Args:
      data: panda DataFrame with OHLC with MACD data
    
    Return:
      buy_signals, sell_signals: for chart plot
      signals: buy/sell transaction for summary printing
    """

    buy_signals = [np.nan]
    sell_signals = [np.nan]
    signals = []
    last_signal = None

    for i in range(1, len(data)):
        if data['macd_hist'][i-1] < 0 and data['macd_hist'][i] > 0:
            price = (data['Open'][i] + data['Close'][i]) / 2
            buy_signals.append(price)
            last_signal = 'buy'
            signals.append({
                'date': data.index[i],
                'action': 'buy',
                'price': price
                })            
            sell_signals.append(np.nan)
        elif data['macd_hist'][i-1] > 0 and data['macd_hist'][i] < 0 and last_signal == 'buy':
            price = (data['Open'][i] + data['Close'][i]) / 2
            sell_signals.append(price)
            last_signal = 'sell'
            signals.append({
                'date': data.index[i],
                'action': 'sell',
                'price': price
                })            
            buy_signals.append(np.nan)
        else:
            buy_signals.append(np.nan)
            sell_signals.append(np.nan)

    return buy_signals, sell_signals, signals

def print_performance_summary(signals):
    """Print buy/sell transactions and statistics
    
    Args:
      signals: recorded buy/sell transactions
    """
    pairs = zip(*[iter(signals)]*2)
    rows = []

    profit_count = 0
    profit_pct_avg = 0

    for (buy, sell) in pairs:
        profit = sell['price'] - buy['price']
        profit_pct = profit / buy['price']

        if profit > 0:
            profit_count += 1
        profit_pct_avg += profit_pct

        row = {
            'buy_date': buy['date'],
            'duration':  (sell['date'] - buy['date']).days,
            'profit': profit,
            'profit_pct': "{0:.2%}".format(profit_pct)
        }
        rows.append(row)

    df =  pd.DataFrame(rows, columns=['buy_date', 'duration', 'profit', 'profit_pct'])
    with pd.option_context('display.max_rows', None, 'display.max_columns', None):
        print(df)

    total_transaction = math.floor(len(signals) / 2)
    stats = {
        'total_transaction': total_transaction,
        'profit_rate': "{0:.2%}".format(profit_count / total_transaction),
        'avg_profit_per_transaction': "{0:.2%}".format(profit_pct_avg / total_transaction)
    }
    for key, value in stats.items():
        print('{0:30}  {1}'.format(key, value))

Result

Output of print_performance_summary for MSFT

We manage to do 8 pairs of trade (buy/sell) in 1 year
5/8 (62.50%) trades are profitable
Average profit per trade is 3.89%, with most profitable trade making 18.98%, while the worst lost -3.96%.

    buy_date  duration     profit profit_pct
0 2020-01-09        18  -0.248023     -0.15%
1 2020-01-30        22   7.896236      4.59%
2 2020-03-26        48  28.745260     18.98%
3 2020-06-09         6  -2.134837     -1.13%
4 2020-06-16        29  15.532458      8.06%
5 2020-08-04         7  -8.439593     -3.96%
6 2020-08-20        15   2.615005      1.23%
7 2020-09-29        24   7.324997      3.52%

total_transaction               8
profit_rate                     62.50%
avg_profit_per_transaction      3.89%

mplfinance MACD buy/sell

What’s Next

We assume the cross-over of MACD and Signal line is the buy/sell trigger. We could run more tests to determine whether delaying N day after cross-over would yield better result.
Maybe we want to consider the weightage if the signal line is above 0 or below 0.
We should test with more stocks, either S&P 500 or NASDAQ 100.
We should experiment of using MACD together with other signals, such as SMA or Stochastic/RSI.

❤️ Is this article helpful?

Buy me a coffee☕ or support my work to keep this space 🖖 and ad-free.
If you can't, do send some 💖 to @d_luaz or help to share this article.

👶 Apps I built

Pixtory App (Alpha) - easily organize photos on your phone into a blog.
COVID-19 - data, chart, information & news.
暖心芽 (WIP) 🌞❤️🌱 - reminder of hope, warmth, thoughts and feelings.
Travelopy - travel discovery and journal
LuaPass - offline password manager
WhatIDoNow - a public log of things I am working on now

By Desmond Lua

A dream boy who enjoys making apps, travelling and gardening. Follow me on @d_luaz.

Tags:

algo-trading

Python Get Historical Stock Prices With Yfinance