Dry mode & backtesting

Cassandre provides a dry mode allowing you to simulate exchange interactions. You can enable it by setting the parameter cassandre.trading.bot.exchange.modes.dry to true in src/main/resources/application.properties.

With this mode, you can simulate accounts, orders, trades, and positions. Cassandre will emulate valid exchange replies. This way, you can test your strategy, see the gains you will make, and validate you have the results you expect.

The first step is to configure your(s) account(s) balances; in Dry mode, Cassandre will search and import all files starting with user and ending with .tsv or .csv in src/main/resources or src/test/resources.

In those files, for each account, you set the balances of each cryptocurrency. For example, this is the content of user-trade.csv :

BTC,0.99962937
USDT,1000
ETH,10 

When you start Cassandre, you will see this:

22:53:38 - Adding account 'trade'
22:53:38 - - Adding balance 0.99962937 BTC
22:53:38 - - Adding balance 1000 USDT
22:53:38 - - Adding balance 10 ETH

Now you can create orders and positions, and this will increase/decrease your account. Of course, Cassandre checks that you have enough assets to accept your orders.

You can use the dry mode to test your trading bot with real data, but you can also use it for backtesting.

In simple words, backtesting a strategy is the process of testing a trading hypothesis/strategy on prior time periods. Cassandre trading bot allows you to simulate your bots' reaction to historical data during tests.

The first step is to add cassandre-trading-bot-spring-boot-starter-test to your project dependency.

Edit your pom.xml file and add this dependency :

	<dependencies>
		...
		<dependency>
			<groupId>tech.cassandre.trading.bot</groupId>
			<artifactId>cassandre-trading-bot-spring-boot-starter-test</artifactId>
			<version>4.0.0</version>
			<scope>test</scope>
		</dependency>
		...
	</dependencies>

​​​​

Now, we need to generate the data we want to use during the JUnit tests. We can use the Kucoin API; to do so, run this on the command line :

startDate=`date --date="3 months ago" +"%s"`
endDate=`date +"%s"`
curl -s "https://api.kucoin.com/api/v1/market/candles?type=1day&symbol=BTC-USDT&startAt=${startDate}&endAt=${endDate}" \
| jq -r -c ".data[] | @tsv" \
| tac $1 > tickers-btc-usdt.tsv

It will create a file named tickers-btc-usdt.tsv that contains the historical rate of BTC-USDT from startDate (3 months ago) to endDate (now). Of course, you can choose your own dates and currency pairs.

Place this file in the src/test/resources folder of your project and add this line to your JUnit test class:

@Import(TickerFluxMock.class)

Now, during the tests, instead of receiving tickers from the exchange, you will receive tickers imported from the tsv/csv files you put in src/test/resources.

You can see an example of dry mode and backtesting in the Technical analysis chapter.