The basis of any trading strategy is having a good backtesting solution, and you can’t backtest unless you have data.
In this post I’m going to step through how to download and save historical data from the Binance API over a given timeframe.
This example does not require an account on Binance as it utilises public API calls.
Working with dates
It’s convenient to be able to pass human readable dates, unfortunately the Binance server only understands millisecond timestamps so we need to convert our date.
This can be done by using helper websites like current millis but we’ve got the power of python so let’s use it.
First install the
Now we can create a function to take a readable string into milliseconds.
With this function we can convert all sorts of handy date formats to milliseconds like the examples below.
Binance Kline endpoint
Now we have that out of the way we can start to work with the Binance API. For our purposes we are interested in the get_klines endpoint to fetch the actual data.
This takes parameters
symbol- e.g ETHBTC
interval- one of (1m, 3m, 5m, 15m, 30m, 1h, 2h, 4h, 6h, 8h, 12h, 1d, 3d, 1w, 1M)
limit- max 500
limit are straightforward,
endTime we now have covered by the above
We note from this that we can get a maximum of 500 results each request, so we will need to build a loop if we are fetching over a long time period.
interval we have a list of so these are passed easily.
The get_klines endpoint returns an array of klines in this order
We are mostly interested in the open time, open, high, low, close and volume values. But why not save all the info while we have it.
interval is a string which the API understands, but for our purposes we will need to convert it to a millisecond representation to help with our loop.
We will ignore “M” or month in this case as it’s tricky to determine how long a month is and given that we will fetch 500 results and Binance has only been running for a year we
can fetch all “1M” results in one request.
With this function we can pass any of the Binance interval values and receive the corresponding value in milliseconds.
Fetch the Klines already
Ok enough messing around, we’re ready to build our function to fetch historical data.
With this function we have a really simple way of fetching a list of klines using simple to use dates and intervals.
Save to file
Once we have fetched the list of klines, it makes sense to save them to a file for later use.
Well after all that I thought I may as well add these functions to the python-binance library for everyone to use.
interval_to_milliseconds have been added to the new binance.helpers file.
get_historical_klines has been added to the binance.client, so now all you need to do is run
With these cached klines we can open them at a later date and run backtesting on them.
See the related [Kucoin post][kucoin-post] to download historical klines on Kucoin.