How to Make an Algo Trading Crypto Bot with Python (Part 1)

We can see that only six trades occurred. These trades generated a profit of 5.09%, which started with 1 BTC and ended with 1.05086506 BTC.

This result is not impressive, considering the risk involved. However, this strategy is as simple as it gets and has vast room for improvement:

Comparing to buy and hold Just holding ETH, i.e., converting our entire stack of BTC to ETH at the beginning of the testing period, we would gain 24.93% (market change indicator), but this is not something we can generally expect. We had far less exposure staking 10% of our stack per trade and not the whole of it. It is important to test our strategy in different conditions – that is not only when the market is growing, but also when it is shrinking.

Trading more coin-pairs We only considered Ethereum, which is one of the hundreds of coins we can trade. This limit only allows for one trade to happen at a time, which is clearly suboptimal.

Using more advanced strategies We used arguably one of the simplest strategies out there, which used only simple moving averages as indicators. Adding complexity doesn’t necessarily mean better performance, but there’s a massive number of indicator combinations we can backtest against eachother to find the best strategy.

Optimizing parameters Currently, we haven’t attempted to optimized any hyperparameters, such as moving average period, return of investment, and stop-loss.

Smaller time periods We only considered daily candlesticks, which is one of the reasons why the bot finds only about 0.02 trades per day, making far fewer trades than a human trader. A bot can potentially make more profit by making more frequent trades and looking at more fine-detailed candlesticks.