The Market and Us

Because I love programming since my first experiences at the age of six I wanted to explain you how you can code your own backtests.

Did you find a signal you would like to know how it worked in the past 20 years?
We do that know together!

First the candlesticks …

At first we need the candlestick data with date, open, high, low, close and volume of each day.
You can get the data from your broker, for example through the API of Interactive Brokers or the csv-file export of TC2000, or you use one of the free or cheap APIs (IEX, Quandl, …).
For our example you can download the needed data here.

… and the import in your program

Now you’re thinking: Hey, I didn’t code anything yet! No problem, it’s not that difficult.
I know many coding languages like JavaScript or PHP but after programming my tools for stocks and trading around 5 years I can recommend to you:

The best language for programming with financial data is Python!

Let’s start our example with the SPY, daily candles and a simple moving average crossing strategy.

To read the data of our source file (SPY.csv, located in the same folder), we just need write this line into a text file named “backtest.py”:

data = pd.read_csv("SPY.csv")

The “pd” stands for pandas, a library for data calculations and comparison which we have to integrate in our code before the line above:

import pandas as pd

To check the data by printing the last 5 rows/days on the screen we can add

print(data.tail())

The strategy

The simple moving average over 200 days (200 SMA) for all rows can be calculated by

data['SMA200'] = data.Close.rolling(200).mean()

and to check if the close is above the SMA 200 just type

data['above']  = data.Close > data.SMA200

Now we have 2 new columns with ‘SMA200‘ and ‘above‘ in our dataset. We will add three more for the “not above” and for the buy and sell signal when crossing (- stands for not and here we have to use & instead of and):

data['notabove'] = -data.above
data['buy']  = data.above & data.notabove.shift(1)
data['sell'] = data.notabove & data.above.shift(1)

Checking each day with one loop

The ‘shift‘ moves a column one row up (shift(-1)) or down (shift(1)), so here we can compare with the day before.
But how to code the backtest for every row of data? While iterating through the rows, we invest at the buy signal and close the position at the sell signal while setting the variable invest true or false:

total_return = 0.0
invested = False
for index, row in data.iterrows():
    if row.buy and not invested:
        entry = row.Close
        invested = True
    if row.sell and invested:
        exit = row.Close
        trade_return = exit/entry - 1
        total_return += trade_return
        invested = False

Show our result

And finally we just have to output the total return:

print("Total return from", data.index[0], "to", data.index[-1], "is", round(total_return*100, 2), "%.")

Now you start our program with this command – and hopefully the same output if you used the source file of the above link:

>> python3 backtest.py
Total return from 1993-01-29 to 2019-10-08 is 172.33 %.

Therefore you had profits of 172.33% if you invested the same amount with our buy signal since 1993. It’s not difficult to change it to the cumulative return or to compare with a buy & hold strategy.

If there are any questions, experiences or wishes about how to code a backtest I would like to read your comment below. I wish you good results with your code, fun with many combinations and a strategy that fits best to you!