Aloha TimescaleDB community from Hawaii!Complete beginner to TimescaleDB here; developing a cryptocurrency analytics application.Context:
We are attempting to store open, high, low, close, volume data in addition to 190 technical indicators (RSI, MACD, Fibonacci pivots, etc.) for over 500 different assets across 7 timeframes.Use case:
Use ML to identify patterns in price action
Apply AI models leveraging RAG architecture to keep context of all timeframes in modelingQuestion:
Using Timescale’s architecture of Hypertables and Chunks what’s the best way to organize this? I’m at a bit of a loss as how to keep things organized, I would like to keep things in one place; especially since we would like our models to keep context of various time frames.
Should each timeframe be it’s own hypertable?
Or should each timeframe be it’s own separate database?
I hope this makes sense! Forgive me for my lack of understanding! Look forward to your responses, deeply appreciate the time you take to help me out in advanced.
It’s amazing that you’re doing it in the database! I was in financial markets for quite some time, and it’s possible to make it.
I started the finance templates and if you want to also continue from there and even bring modules with finance indicators, contributions are very welcome and building a community around finance can also be a great strategy for having solid algorithms validated by multiple users.
And see if you can also build your components like stats_aggs and other components that you can rollup using hierarchical continuous aggregates.
Building aggregations on SQL is not hard, and you can define a package of finance indicators built as aggregations.
Every continuous aggregate is also a hypertable. In the case of databases, I’d consider different databases only if you’re talking about different markets that will never be analyzed together.
The safest method of setting up a database as a service is generally setting a 1:1 database to a single service, so you needn’t worry about it.
If you want to organize the database like ‘folders’, you can use namespaces, AKA schemas, and put tables inside them. For example, one schema for crypto and another for stocks.
