A Data Analysis View on Cryptocurrencies — Optimizing the Trade Moment
I ended up helping on mining cryptocurrency data and calculated how many consequent trades follow while the rate is dropping. This doesn’t predict the future rate, but when doing high frequency trading (or something approximating it), placing the trades according to expected lengths of rate drop streaks might give an advantage that is minor but worth using. Please note that this is not advice on cryptocurrency trading, just interesting observations on the data.
Here’s an example of a few currency pairs from data between 2021–01–01 and 2021–04–17 that’s all publicly available from the Poloniex API.
I have calculated how many drops there are on average in the rate before it goes up again.
For example, consequent trades with rates 0.40, 0.38, 0.37, 0.39 mean a drop streak of 2 trades followed by rise.
It is easy to see that different currencies don’t follow the same pattern (I filtered to only 5 or less drops for relevance):
So, based on this, it might make sense to buy Dogecoin with Bitcoin after the first drop, but for example when buying Ethereum with Bitcoin, it might make sense to wait till there are two price drops to improve the odds of getting a price lift immediately after the trade. Or perhaps it might even be worth while waiting for a couple of more drops, because while it would make the trades more rare, it might make them even more profitable.
This analysis only contains only consequent trades, and it doesn’t consider the time between them (it could be a minute or a second or even a fraction of a second). So, by looking into averages over different periods of time, it might be even easier to recognize trade opportunities.
Data mined using Microsoft SQL Server Management Studio. The SQL code used: