An analyst for every stock
London
Joined February 2026
- Tweets59
- Following81
- Followers375
- Likes50
Baseline retweeted
Guys at baseline are building something really useful and cool. Been using for my PA lately, definitely worth checking out
When I launched coverage on the sell side, the first thing I did was pull up the five-year chart, mark the big up and down days, and work out what happened on each.
We’ve now built that into Baseline. Every significant move comes with a researched explanation, alongside indexed moves for genuine subsector peers and valuation metrics, so you can see whether a move was the company, the sector or the market.
Free to look up any name, no sign-in. Link below.
$TW cut its outlook on Wednesday. $PSN's sales rate rose 4% on Thursday after the Bank of England raised rates to 4%.
The difference is price point and region. Persimmon sells the cheapest homes of the large builders. Taylor Wimpey has more of its business in the South, where deposits are harder to save.
It shows up fast because builders barely set their own price. Second-hand homes and mortgage lenders do. Seven of the eight UK builders we cover score 2 on Pricing Power.
That's also why the Budget's first-time buyer scheme hit the shares. A 4% developer fee is £12,000 on a £300,000 home, against about £23,000 of operating profit at Taylor Wimpey's current margin.
2
183
Baseline retweeted
While everyone is building a better chatbot for stocks, we built something different - a library for your chatbot.
Say you're researching a stock and need a quick view on management credibility. A "vanilla" Claude will take 20 minutes to research this. It will struggle to get full transcripts, earnings releases, shareholder letters, etc. and will answer based on fragmented info. When you try to close off the analysis with a final "Are you sure?", it will confess its shortcuts and why it "owes you an apology".
Now consider your Claude with access to Baseline MCP. It will now have access to off-the-shelf, pre-processed analysis on management credibility for every company. This analysis is generated via piped filings & transcripts, run through 5+ chained prompts for context management, and is tested on strict evals. For example, every disclosed metric is ensured to be on the same basis before it's compared.
The result? A Claude that's 10x faster and smarter.
Don't take our word for it. Just connect to the Baseline MCP and ask Claude himself. Try it @BaselineByA5
Baseline retweeted
Baseline boldly asks the question, “What if we compressed the margins of every AI for investing tool by 90%?”
$40/month is insane for something like this tbh
🤝 Paid partnership
While everyone is building a better chatbot for stocks, we built something different - a library for your chatbot.
Say you're researching a stock and need a quick view on management credibility. A "vanilla" Claude will take 20 minutes to research this. It will struggle to get full transcripts, earnings releases, shareholder letters, etc. and will answer based on fragmented info. When you try to close off the analysis with a final "Are you sure?", it will confess its shortcuts and why it "owes you an apology".
Now consider your Claude with access to Baseline MCP. It will now have access to off-the-shelf, pre-processed analysis on management credibility for every company. This analysis is generated via piped filings & transcripts, run through 5+ chained prompts for context management, and is tested on strict evals. For example, every disclosed metric is ensured to be on the same basis before it's compared.
The result? A Claude that's 10x faster and smarter.
Don't take our word for it. Just connect to the Baseline MCP and ask Claude himself. Try it @BaselineByA5
Baseline retweeted
Like this platform a lot because the biggest hurdle I have in using models like Opus 5.5 and GPT Sol is the speed of execution
Feels like you are always trading off better quality of responses for timing, particularly when you are looking for quick responses in the middle of a meeting or IC
A faster platform with the same capabilities is a huge change in the right direction, particularly if you can give it access to all the relevant data sources you want to pull from
🤝 Paid partnership
While everyone is building a better chatbot for stocks, we built something different - a library for your chatbot.
Say you're researching a stock and need a quick view on management credibility. A "vanilla" Claude will take 20 minutes to research this. It will struggle to get full transcripts, earnings releases, shareholder letters, etc. and will answer based on fragmented info. When you try to close off the analysis with a final "Are you sure?", it will confess its shortcuts and why it "owes you an apology".
Now consider your Claude with access to Baseline MCP. It will now have access to off-the-shelf, pre-processed analysis on management credibility for every company. This analysis is generated via piped filings & transcripts, run through 5+ chained prompts for context management, and is tested on strict evals. For example, every disclosed metric is ensured to be on the same basis before it's compared.
The result? A Claude that's 10x faster and smarter.
Don't take our word for it. Just connect to the Baseline MCP and ask Claude himself. Try it @BaselineByA5
Ask your claude how it feels with and without Baseline MCP
Every AI startup for stocks is building an inference engine, and that’s the wrong approach. The hard part of investing was never researching a stock you already picked. It is finding the stock in the first place.
How do you prompt Claude to find the best investment in a universe of 5k+ names? Even for a single stock, the answer only gets good around prompt 20, after you have fed it the context yourself and asked “have you checked X?” ten times.
What if all the questions were already answered and all the answers were already there? These guys went ahead and did that for 4k stocks and put that into a database for everyone to use.
My hunch is that this is what the future of AI is going to look like. Not another chatbot, but pre-computed inference that surfaces what’s most important.
Everyone has the same model. The difference is what the models have read.
Connect your AI to the Baseline MCP library and watch it get 10x faster and smarter on equity research.
While everyone is building a better chatbot for stocks, we built something different - a library for your chatbot.
Say you're researching a stock and need a quick view on management credibility. A "vanilla" Claude will take 20 minutes to research this. It will struggle to get full transcripts, earnings releases, shareholder letters, etc. and will answer based on fragmented info. When you try to close off the analysis with a final "Are you sure?", it will confess its shortcuts and why it "owes you an apology".
Now consider your Claude with access to Baseline MCP. It will now have access to off-the-shelf, pre-processed analysis on management credibility for every company. This analysis is generated via piped filings & transcripts, run through 5+ chained prompts for context management, and is tested on strict evals. For example, every disclosed metric is ensured to be on the same basis before it's compared.
The result? A Claude that's 10x faster and smarter.
Don't take our word for it. Just connect to the Baseline MCP and ask Claude himself. Try it @BaselineByA5
Assessing the accounting quality of companies coming to market can be overwhelming. Anthropic's S-1 is out and the price range is still blank. Before it prices, I went back through the 20 big US listings since January 2025 that we cover to see what the filings said once the roadshow was over. 8 of the 20 score 2 on accounting. Across 3,744 companies it is 6%.
These flags however have not affected prices or valuations. The median company scoring 2 is up 5% on its IPO price and the median scoring 4 is down 19%.
1
32
Read accounting analysis on 3,700+ stocks here: baseline.a5labs.com/app/scre…
12
Baseline retweeted
Pretty cool. Feels like 2027 is going to be the year AI really impacts the investment industry. This basically feels the equivalent of a sell-side research library of primers and earnings notes available for the average retail investor for free. I find it hard to be bullish jobs!
Stockpicking will never be the same again.
No more fumbling 20 min with prompts, waiting for the AI to finish research. Baseline asks every possible question on 4k stocks and aggregates the answers in a single database. Every answer is tagged and scored so you know where to look.
Baseline's database is accessible via MCP, so your Claude can start research with 100% context on every company, rather than "noodling" 20 minutes to get the latest earnings call transcript.
Try it @BaselineByA5
Baseline retweeted
This is a really cool product. Basically it's doing LLM analysis on stonks. There are a lot of services doing that, but this one is pretty refined/polished as it has involved a lot of real work from a real hedge fund analyst(s) and real software developers. You can see some of the output for $TAP in the screenshot below: I think the ratings are fair but maybe a bit harsh given that it was refreshed before the recent earnings/guidance miss.
I've been testing and giving a bit of feedback over the last few months. In particular I wanted street estimates integrated into the process, since my prompt process ("Computer") doesn't have access to that information off the shelf and scraping it is a pain. Now, the real cool part imo is that once you link this in with your Chat/Claude via MCP connection you can use it as a starting point/reference for further analysis/prompting.
Currently the platform is limited by quarterly scoring/review cadence and minimal street estimate integration, so I think it's most suitable for screening ideas and/or integrating into other processes via the MCP. I will be doing some more testing/experimentation now that they've run a quarterly refresh. I understand that distributing/licensing street estimates/trends is a sensitive matter, but I see that as the most likely value-add for Computer. Similarly, synthesizing industry reports could be valuable. My main concern is that Computer is already a really good analyst so there's a clear tradeoff between giving it a head-start/reference on analysis (without new information) versus eating into its context. Basically my feedback at this point is that a combination of synthesis + new information/data is probably the most promising angle.
Stockpicking will never be the same again.
No more fumbling 20 min with prompts, waiting for the AI to finish research. Baseline asks every possible question on 4k stocks and aggregates the answers in a single database. Every answer is tagged and scored so you know where to look.
Baseline's database is accessible via MCP, so your Claude can start research with 100% context on every company, rather than "noodling" 20 minutes to get the latest earnings call transcript.
Try it @BaselineByA5