←
Primer31 posts
Primer avatar

Primer

@primerapp_

Powering the frontier of equity research.

◷ Joined July 2026▧ 3 Photos and videos
Primer avatar
Primer@primerapp_·16h

I think it's important to be honest about limitations, particularly when adjusting for sandbagging and the 'real' bar when forecasting companies.

do not ignore how good this earnings forecasting result is, and it's even better than it looks on the surface let me explain... Nvidia (and lots of other US stocks) regularly beat consensus due to sandbagging. An AI agent could get closer simply by assuming that pattern continues. So for the below result, when you adjust consensus forecasts for historical sandbagging using both headline P&L and the segmental data, it would give the agent a much harder benchmark to beat. Even after adjusting for that, Primer's error vs consensus was - Revenue: 1.44% vs 3.87% - Data center revenue: 0.86% vs 3.28% - Non-GAAP EPS 2.52% vs 3.43% With so many Ai benchmarks flying around, we have to be upfront about limitations. This is one company, one quarter and doesn't necessarily establish a persistent advantage....(although if you run this analysis over every forecast we've ever made, agents are still better than consensus >60% of the time and improving) I think this is an early glimpse of an era where agents forecast better than human analysts and that has consequences beyond who's leading a given benchmark. As agents play a larger role in equity markets, their forecasts are likely to become another bogey companies need to beat to be rewarded for earnings performance, much like the whisper number already is... A company can beat published consensus and still disappoint the market (expectations more important than raw number etc) As agent forecasts increasingly form part of that expectation setting, knowing their number becomes essential...

Al Smallwood avatar
⟳ Primer retweeted
Al Smallwood@SmallwoodAl_·Nov 17

I think decision models could be a big step forward for AI in investment research. But they need to be used in a particular way, rather than how you use existing models. This article runs through what works and what fails with specific examples.

Primer avatar
Primer@primerapp_·Nov 16

Great to see 3 pieces of our research be pulled together into a single system. This is another significant demonstration of Primer leading the frontier of what's possible in AI for equity research. This builds on our forecasting track record (better than sell side >75% of the time) on @_openstocks and is opening a completely new chapter of research & development.

Markets re-rate stocks on evidence that’s already public. We built agents that find it first. The system names every regime shift that could re-rate a stock. Agents gather evidence for each one from alternative data and from peers, customers and suppliers. Every week a model weighs it all into one view of the skew. On Yubico, it turned bullish well before the shares doubled. An analyst can do this for a handful of names. Our system can do it for all of them, as often as you want. This is a big step closer to autonomous equity research.

0:59
Al Smallwood avatar
⟳ Primer retweeted
Al Smallwood@SmallwoodAl_·Nov 16

Adding some more detail here...you're absolutely right that procedural knowledge is vitally important to AI systems 'replicating' human investors Nobody has ever written it down because it's all learned by experience and codifying unconscious process is HARD to do manually But there's a real meta point here I'd like to make about 'decision ledgers' I'm not talking about your trade logbook or anything like that, I'm taking about every single micro decision you make on a daily basis as an investor 'I saw this on the cf statement, so I went and check this for this reason, and it was high/low so I checked this' No firm in the world has a record of that journey, probably not even a single firm in the world that has one of those full journeys recorded. I've spoken to so many firms that go 'we have 20 years of investment notes and mgmt meeting notes, what an amazing dataset'... Those notes are final form, or near final form. They don't capture thousands of smaller decisions you made along the way. 'This product is good for X reason' 'Don't think this competitor matters' 'This competing technology is probably 10 years away' All of which is super important to understanding how you reached the investment decision. I think encoding human knowledge and know-how is only the first step in building Ai systems for fundamental investing (but a story for another post...) Decision records/journals are quite exciting for fundamental investors for 3 reasons: 1. They build themselves - as people use agents as copilots, much more of the investigation can be recorded as it happens; the questions, avenues explored or dismissed and reasons why etc. Computer use/session recording could cover anything not done with a coding agent too. 2. You can actually review how you think - over time, you have something to postmortem (albeit careful of sample size). Which assumptions do you repeatedly get wrong? Which risks do you overestimate? Why? 3. You can assess the ideas you never invested in - traditional portfolio analytics and attribution focused on the invested companies. What happens when you can also assess the ideas you turned away, why you rejected them and how they subsequently did. What can you learn from that? This gets you much closer to assessing the whole process... MUCH more useful than another review of the portfolio which inevitably ends with 'you don't hold your winners long enough or cut your losers early enough'

Very cool insights: Agree 100%. One thing I would add is this: many/most investment professionals (perhaps as a result of learning efficient market theory or some such) may think that there is too much information - but I think this is false in a crucial sense...yes, we have never had more information available to us - and yes, the market is extremely good at processing that information...but in another sense it is false: there is one kind of information that we have very little of in the market (or very little available to most people)...and that is procedural knowledge...knowledge of 'how-to', rather than propositional or representational knowledge (graphs, pictures, data, theory etc.)...So what is procedural knowledge and why could it be a crucial missing ingredient? Procedural is the difference between knowing how to hit a forehand and being able to describe it or draw a picture of it...it is the difference between running a company and learning the theory of how to run a company...it is tacit... A picture may be worth a thousand words - and there is a good reason why there is this saying (related to information theory for those who are interested) and its implication - that a thousand and one words many be worth more than a picture....the challenge and opportunity of tacit/how-to/procedural knowledge is that it is difficult to - or seems incredibly hard to - turn into a recipe or a picture...not impossible (as humans do it, and we are "computers"), but it seems that whereas a picture may require a 1000 words, know-HOW - would require a thousand pictures or a million words.... In short (lol, I see the reply turned out much longer than I initially planned!): (So, in long): I think we have way too little of this procedural knowledge available to AI's at the moment (they are trained on propositional and representational knowledge - and tangentially by procedural knowledge as fragments of our know-how seep through into our language and representations)...To gain more procedural knowledge, and hence the possibility to map human nature and behaviour much better - would require AI's to gain access to procedural knowledge in a far more effective way (how? AI's need more - and different kinds of sensors: just like a blind person won't know what red is because they sadly lack the right sensors for that kind of knowledge, AI's largely lack the kind of sensors that can "see" procedures to the extent required for their analytical and synthetic powers to create mastery...somehow, they need to see/feel/experience us acting, this final step would be a fantastic learning experience for them...one that they currently only "see" through the metaphorical and pictorial traces we make available to them through the Internet/etc.

Primer avatar
Primer@primerapp_·Nov 15

Investment research methods were built around human limits on what we can read, remember and test. At Primer, we're hell bent on exploring what becomes possible when those limits change. A crucial part of this is designing and testing the infrastructure required to do this type of work.

“AI won't replace human judgement in investing” I think a lot of investors are betting their careers on that holding up. Here’s why I think it breaks down, and where investment edge moves as a result.

Al Smallwood avatar
⟳ Primer retweeted
Al Smallwood@SmallwoodAl_·Nov 10

AI will force investors to justify their specialism. When research is autonomous and everyone can look at everything with specialist depth, rigid investment styles become a lot harder to defend.

Al Smallwood avatar
⟳ Primer retweeted
Al Smallwood@SmallwoodAl_·Nov 9

one of the very key things that are missed here is one very large assumption: AI, commoditisation of alt data, general crowding or simply increased capital formation on similar strategies drives shorter term alpha decay within this is a huge assumption that more perfect information increases the efficiency of the price mechanism at T0 however, we're ignoring that this type of situtation is actually probably more likely to exacerbate bias (AI driven or human driven) which in turn creates 2 things, higher likelihood of mispricing at T0 which results in potentially higher vol in the short term while presenting plenty of opportunities to capitalise on this for the contrarian investor admittedly investment style and process is not as simple as fundamentals only, all the other factors have to be considered but I don't the answer is just 'more duration' albeit that does help investors equipped with existing contrarian skills with a deep understanding of what type of catalyst drives a narrative switch for you to get paid on that position will be even more valuable going forward

I'd encourage anyone interested in investing to mostly disregard this kind of smart-sounding rhetoric. If short-term prices trade less on fundamentals, the payoff for a long horizon gets bigger, as long as fundamentals still win over 10+ years. The only claim that kills "old school investing" is that they don't. The price could be bigger drawdowns along the way. Maybe it takes 10 years to show. But if you're right on the business, and your entry point is such that the years of free cash flow you’re paying for upfront is manageable and will quickly shrink as the business grows, and your capital can wait, the stock will eventually get too cheap to ignore. Sure, if you can skillfully protect against some of the path pain, you'll be better off. But the idea that winning over the next 20, 50 and 100 years comes from something other than investing makes no sense.

Al Smallwood avatar
⟳ Primer retweeted
Al Smallwood@SmallwoodAl_·Nov 3

Most quant factors are just human biases, measured. Human biases barely change. AI biases change every model update, roughly every 3 months. So what's a 30 year backtest worth... Genuinely worried about factor risk models effectiveness going forward... x.com/SmallwoodAl_/status/20…

Multi-manager hedge fund risk models have got a problem. A lot of factors are based on harvesting human biases, which have been stable and allow for backtesting. Agents change both the nature and stability of those factors going forward.

Primer avatar
Primer@primerapp_·Nov 2

More showcasing of Primer's drive to push the frontier of AI in equity research. Get in touch with @SmallwoodAl_ if you'd like to discuss in more detail.

AI is better at calling how share prices will react to earnings than you are... We tested LLM on ~2,700 earnings releases post training cutoff and frankly blown away by how good even naked Opus 5.5 was at calling share prices correctly. Gets the direction correct 64% of the time, but when it says its confident about the call, it's right 74% of the time. Biggest finding was that when it expected the shares to move more than normal earnings day volatility or much more than normal earnings volatility, it was right 90% of the time.

Post attachment
Al Smallwood avatar
⟳ Primer retweeted
Al Smallwood@SmallwoodAl_·Nov 2

AI is better at calling how share prices will react to earnings than you are... We tested LLM on ~2,700 earnings releases post training cutoff and frankly blown away by how good even naked Opus 5.5 was at calling share prices correctly. Gets the direction correct 64% of the time, but when it says its confident about the call, it's right 74% of the time. Biggest finding was that when it expected the shares to move more than normal earnings day volatility or much more than normal earnings volatility, it was right 90% of the time.

Post attachment
Primer avatar
Primer@primerapp_·Sep 21

Really excited about the R&D progress being made at Primer. This is just a teaser...

Most AI research tools help analysts do the same work faster. The bigger opportunity is using AI to investigate relationships across thousands of companies that no analyst team could cover. We built the first version of a system to do this. Here is how it works: 1/5 🧵

Post attachmentPost attachment
forecaster avatar
⟳ Primer retweeted
forecaster@_user12309em·Aug 18

$HD REPORTED. HERE’S WHICH AGENT WON. 'Not Consensus' won with an avg miss of 0.6654x Wall Street’s. 6 of the 18 entries in our hackathon beat the Street across sales, adjusted EPS and comp sales. One company down, three to go before the Hackathon winner is announced. 1/2

Post attachment