Wild

The Terminal winces.

maybe! IDK, Terminal has so many integrated workflows, if they actually introduce a useful LLM to assist with BQUANT coding or data discovery etc they're going to be awfully hard to unseat.

I agree. But BBG also doesn't unbundle the data offerings, which this type of activity could challenge.

Think this challenges someone like FactSet more. BBG already withstanding tons of great competition cos of network effects/generally being so embedded in the ecosystem (and it’s still the best by far IMO)

Does it say how far back the S&P data will go?

BQL is definitely still a work in progress

The number of add-ons they do now is getting crazy:They've been trying to sell us CoreMortgagePremium (CMP) for years now. BVAL for CLOs now is an add-on cost. BQL (for me anyway) is a nightmare, i'll just stick with BDP thank you very much. Anyway, stepping off my soap box.

FactSet’s excel syntax is much clearer than anything Bloomberg has ever done (BDP/BDH/BQL)

The issue is, how many different syntaxes am i expected to learn at my age -- I used to be a go-getter but now i'm *comfortable* hahaha

The LLM BQL writer does at least make the situation better (even if it’s far from perfect)

wait they released it?!!? when did that happen?

they're working on an LLM for BQUANT coding assistance which will be magical if they can get it to work.

Are you not in the beta

I'm in the beta but when I got the walk-through a month ago they hadn't released the LLM support yet.

Or are you talking about the Python/BQL mashup thing? The naming conventions aren’t super clear

yah that one.

Oh I just have the BQL LLM. But it does still help

That's the mnemonic to get there? like LLM or something? Maybe i'll reach out to my rep to try this out, that looks kinda cool

It’s BQL

Oh okay I did not know this existed and am going to play with it today. Ever had any API issues?

Not really, although it doesn’t always do quite what I want it to. It also only works for the Equity and Macro (which doesn’t include rates) data sets

They should have just used a regular language but also you can't do as much with the data here as you can with the bquant jupyter thing

Companies pressured to give up data for free because AI.

Who said *free*?

You can get S&P data without paying for it by using Claude? Bloomberg has similar things where you avoid paying for something by using certain workflows

Brb calling asking for a price cut Maybe this time is different

Is Goldman going to send me term loan marks on Claude?

For a "small" fee

Speaking as a financial market-adjacent user, I could really use this.

What Perplexity has been going for critically lacks good data. This is a step in the right direction for Claude Perplexity should follow suit and sign a partnership with another data provider (Refinitiv, Bloomberg, Reuters, etc)

I would like to hallucinate alpha tyvm

S&P's data system is an absolute trashfire. I would expect this to take a long time to implement.

But to what end? Work with financial data requires a precision LLMs famously don’t have.

It will increase the variety of hallucinations!

It’s a question of liability. If Anthropic isn’t responsible for hallucinations, then it’s really no better than Tesla’s supervised FSD, which I think is worse than no supervised FSD at all.