Databento
A simpler, faster way to get market data
About
Curious about net revenue retention on the quant desks vs the retail-adjacent buyers. One of those cohorts churns like crazy and I have my guesses.
The engagement curve on this tweet is doing something weird around hour 3. Either a big account quote-tweeted or the algo just likes the word Series.
The announcement thread is refreshingly boring in a good way. No cinematic drone shot of a Bloomberg terminal, no orchestral swell, just numbers and logos. Respect.
Two people and a Postgres could probably ship an MVP of this in a weekend. Scaling it to tick-level historical without going bankrupt on egress fees is the actual moat.
Market data pricing has been a hostage situation since 2004, so watching a startup walk in with a normal API and a menu is basically financial thriller content.
Waiting for the inevitable on-chain oracle spinoff where equities tick data gets posted to some L2 nobody uses. You know it's coming.
If the infra team is hiring, I've got a low-latency C++ engineer sitting on an offer deadline this Friday. DMs open.
Nobody talks about how painful it is to actually fulfill historical data orders. Curious how the ops side scales when a client asks for ten years of options tick data at 2am.
I still remember when getting a CSV of daily bars felt like a heist. If this works as advertised then my 2019 self would be extremely jealous.
One more thing: a natural language query layer over the historical archives. I want to type give me every SPY print during a Fed minute release and just get a parquet back.