QuestDB

Agents are becoming part of the dev toolbox. With #QuestDB, they can explore data and run advanced analytics via SQL + REST, even using extensions like HORIZON JOIN. Two simple prompts are enough to go from schema discovery to slippage and implementation shortfall.

More info at #per-column-parquet-encoding-compression-and-bloom-filters" target="_blank" rel="nofollow noopener">questdb.com/docs/query/sql…

We added per-column Parquet controls to #QuestDB. You can now configure compression, encoding, and bloom filters directly in SQL. More control over storage, performance, and interoperability with open formats.

Full details at questdb.com/blog/questdb-9…

We just released #QuestDB 9.3.5 and QuestDB Enterprise 3.2.5. This release adds LATERAL JOIN, SQL-standard UNNEST for arrays and JSON, new statistical window functions, and multi-table HORIZON JOIN. Details in thread 👇

We are extending HORIZON JOIN in #QuestDB. You can now join multiple right-hand-side tables in a single query, making multi-source markout analysis much simpler. No self-joins. No CTEs. Just write it directly. Query example in the image

Blog post at questdb.com/blog/zero-shot…

The hard part of ML is getting the data. With #QuestDB you can access time-series data via SQL, Parquet, or directly from partitions. This post shows how it fits into zero-shot forecasting workflows. Details in thread 👇

We are working on EARLIEST ON in #QuestDB. It returns the row with the earliest timestamp for each unique combination of columns. The inverse of LATEST ON, making “first event per key” queries simpler and faster.

(merged) Pull Request at github.com/questdb/questd…

We added LATERAL JOIN in #QuestDB. LATERAL lets subqueries reference columns from preceding tables, making per-row logic and correlated subqueries much easier to express in SQL. And it is optimized with decorrelation to avoid per-row execution. Details in thread 👇

Docs at questdb.com/docs/query/sql…

Did you know #QuestDB now supports UNNEST for arrays? It lets you expand arrays into rows directly in SQL, which is handy for order books and JSON arrays from external APIs. Once expanded, you can filter, join, aggregate, and analyze them like any other rows.

@IcoMarketer Thanks! Glad you liked it :)

@QuestDb That's a game-changer for execution analytics! Custom lookback windows per row mean cleaner SQL—no more Python hacks for mid/spread calcs. QuestDB keeps leveling up 🚀

Dynamic WINDOW JOIN ranges in #QuestDB let each row define its own time horizon. Very useful for execution analytics, where trades or order slices often use different lookback windows to compute mid, spread, or liquidity context directly in SQL. Query example 👇

@IcoMarketer Thanks!

@QuestDb Great example of real-time systems in action—time-series infra like QuestDB is exactly what enables low-latency, interactive analytics at scale. Always interesting to see how teams build on top of it.

We have a guest post on the blog from Arden Charts explaining how they use #QuestDB as the engine behind their real-time charting platform. A nice look at how time-series databases power interactive analytics and visualization products. Details in thread 👇

Release notes at github.com/questdb/questd…

We just released #QuestDB 9.3.4. Dynamic windows in WINDOW JOIN, Parquet row group pruning with bloom filters, new array functions, and performance improvements across ORDER BY, joins, GROUP BY, and Parquet I/O. Details in thread 👇

Coming soon to #QuestDB: storage lifecycle policies. Sample SQL statement ⬇️ CREATE TABLE abc (col1 LONG, ts TIMESTAMP) TIMESTAMP(ts) PARTITION BY DAY STORAGE POLICY(TO PARQUET 3d, DROP NATIVE 10d, DROP LOCAL 1M, DROP REMOTE 6M);

read the post at questdb.com/blog/arden-cha…