Bitcoin as a prominent cryptocurrency has experienced unprecedented growth in the past decade. It has become an accepted asset class to diversify investor returns (along with some other major cryptocurrencies). The question is whether using simple trading rules can improve the risk-adjusted return of such a portfolio.

Log returns are a very useful concept when working with timeseries. It turns out given statement is even more true when working with time-series database and materialized views in clickhouse.

You are now able to filter screener by non-numeric fields, such as when the crypto-currency started from, or if it is still active.

We added a new functionality, where users can upvote or add new feature requests.

I will describe here how in CryptoQuant I architected parsing user-input trading rules into a machine runnable AST.

This parsing system forms the critical bridge between human-readable trading strategies and executable code that can efficiently process market data - ultimately allowing our users to automate sophisticated trading strategies without requiring deep programming expertise.

CryptoQuant is basically one person startup, hence, the architectural choices made need to be pragmatic.