Title
#users-market-data
Newskooler

Newskooler

10/06/2022, 12:21 PM
Why is the most upvoted answer here flat-files?https://quant.stackexchange.com/questions/29572/building-financial-data-time-series-database-from-scratch This makes no sense to me. Any thoughts by people using relevant data?
Andrey Pechkurov

Andrey Pechkurov

10/06/2022, 12:31 PM
A basic structured file system of serialized data chunks works far better.
That sounds like a primitive database on its own to me.
Newskooler

Newskooler

10/06/2022, 12:32 PM
Indeed… I find it fascinatingly ridiculous that this is the most upvoted answer. At the same time, coming form this world, I am anything but surprised. I only misjudged the collective intelligence on the web. Nonetheless I wanted to share here and see if I am missing something.
Andrey Pechkurov

Andrey Pechkurov

10/06/2022, 12:32 PM
Any complex queries with such approach would require complex logic - the application would have to act as a query engine. If there is no query language, each query will be a unique snowflake - you'd have to write some new code for it.
Newskooler

Newskooler

10/06/2022, 12:32 PM
You’re correct Andrei and that’s exactly what ends up happening.
Andrey Pechkurov

Andrey Pechkurov

10/06/2022, 12:33 PM
Sometimes people love extravagant answers on SO
j

javier ramirez

10/06/2022, 2:33 PM
As the guy who replied acknowledged, they only do lookups by time and symbol. The moment you try to do something more sophisticated, good luck with that solution without building yourself a complex engine on top. I can only imagine how many engineering hours went to build something that sounded simple. And of course now the company is stuck with this guy for any future bug fixes or updates because no one else is using that tool.
Newskooler

Newskooler

10/06/2022, 3:00 PM
Spot on