Adrian

Everyone should read this... The High Shoulders Theory (a visual thread)

Cornell Professor and Former Head of Machine Learning at AQR Capital: "Most machine learning funds fail." Researchers search for alpha alone. In 40 minutes, Marcos López de Prado explains why alpha discovery should be a loop, not a one-shot attempt. Hypothesis → test → evaluation → new hypothesis. That's how real edge is created. Watch the lecture, bookmark it, then test the framework yourself for free: join.horizon.trade/sopersone-x

stop asking Claude one question and thinking you understand the topic. you don't. Stanford proved a better way. it's called STORM. peer reviewed. 25% more organized output. open source. the trick: don't ask one question. ask five. from five different experts. >the practitioner: what do they know that academics miss? >the skeptic: what's the strongest counterargument? >the economist: who profits from the current narrative? >the historian: what pattern has played out before? >the academic: what does the evidence actually say? 4 prompts. 5 minutes. no software. no GitHub. just paste into Claude. single prompts give you what everyone already knows. STORM gives you what nobody else found. this article has all 4 prompts ready to copy. pick your hardest topic. paste prompt 1. you'll know more in 5 minutes than people who spent days reading.

The reason hedge fund data teams archive a copy of every vendor file the moment it's published: Aakarsh Ramchandi (@2sidesofacoin) explains: "I will never trust what someone tells me their signal, prediction, or data set is." "I create a copy of it exactly when you published it." "Once I have enough copies over enough time, I know how many times you're actually overwriting your data." "Now I know: can I trust it? When should I trust it? Do you change your numbers all the time?" "Before we give you anything — is there discipline in how we collect this data first?"

Quant Primer, BofA, 318 pages I’m not a quant, but why not ..

MICHAEL SAYLOR SAID WORKING HARD IS THE WORST ADVICE YOU CAN GET. “YOU DON'T WANT TO MAKE MONEY BY BEING TALENTED AND WORKING HARD. THE ROBOTS ARE GOING TO WORK HARD. THE CARS ARE GOING TO DRIVE THEMSELVES.” “ONCE YOU TRAIN THE AI ON A SHAKESPEAREAN SONNET, IT WILL SPIT BACK SONNETS JUST AS GOOD AS SHAKESPEARE IN HIS PRIME. IF YOU STUDIED FOR 20 YEARS TO LEARN HOW TO COMPOSE, THAT'S BECOMING LESS VALUABLE — JUST LIKE WRITING A 100-PAGE LEGAL DOCUMENT.” “HUMAN CAPITAL IS GETTING DEMONETIZED.”

Who should not become a quant trader? Start with the kind of network you'll build — and the kind you won't: Andrew Courtney (@andrewcourt1). 15 years as a quant trader and market maker at Susquehanna International Group. "Let's frame it as who this fits. Contrast it with another elite career." "A banker gathers a wide network across firms. A consultant meets C-suites a year out of college." "My primary relationships were my coworkers — fantastic people, but that was most of it." "A quant trader isn't at conferences, isn't networking, isn't telling anyone what they do." "I can't call a company's C-suite for advice the way a consultant could." "Narrow, concentrated, dense. A different kind of career entirely."

“If you set your goals ridiculously high and it’s a failure, you will fail above everyone else’s success.” — James Cameron

A rule that will improve your life: Never assume bad intent, but don’t ignore repeated patterns. Everyone gets grace. Nobody gets unlimited passes.

This paragraph by Haruki Murakami hits very hard: “Once the storm is over, you won’t remember how you made it through, how you managed to survive. You won’t even be sure, whether the storm is really over. But one thing is certain. When you come out of the storm, you won’t be the same person who walked in. That’s what this storm’s all about.”

Investing is a compression problem

Everyone needs to hear this…

A wise mentor once told me: "Before you play the game, study the winners. If you don't want their life, don't play their game."

stop trying to predict price with machine learning because you are just learning noise patterns that will fail tomorrow the top reason most machine learning hedge funds fail is that they use ai to predict direction instead of managing risk this ibm researcher left to join the most successful quantitative hedge fund ever and his strategy doesn't use ai to find alpha at all

Here's an example of how an alpha decays as we trade less and less liquid assets. This the pre-fees performance of an alpha `cross_ex_std_1w`. It is the standard deviation of volume across exchanges for a given timestamp, and then averaged over the past week. Basically how well do exchanges agree, you can also use OI it is about 0.86 correlation if you swap it out for OI. It is very clear that as we trade less liquid and higher spread assets where it becomes increasingly harder and harder to monetise the signal gets worse. It starts out at 2 sharpe and ends off at 0.3 Sharpe for top 30 (by marketcap). An example of efficiency of alphas.

As someone who builds institutional level quant systems, this Stanford paper on Market Making is the closest thing to an HFT desk I have ever seen publicly shared. 19 pages. Hedge Fund level Market Making Algorithm. Bookmark & get this before someone takes it down.

Documenting the headwinds I now see for AI. It won't seem like it, but I love AI and am long-term positive. But when "math doesn't math" I take note. 1. The core thesis for foundation model lab investment has been high upfront investment made worthwhile by significant long-term profits. 2. These are capital intensive businesses and the compute commitments are very high relative to revenue and require strong growth over long time periods. The "leverage" (commitments versus revenue) is extremely high. 3. The fundamentals are not as positive as they previously were: • Input costs are higher (commodities, chips, power) • Interest rates are higher • Competition is more intense • Scaling Laws are now problematic: exponential costs/power cannot continue 4. Forecasting compute spend is challenging and high risk due to (a) revenue uncertainty and (b) algorithm uncertainty 5. Revenue growth appears to be slowing. The technology is valuable, but ROI is proving to be more expensive and take longer than anticipated. 6. The future is likely "different models for different use cases" with the lower end of the market being highly competitive. 7. Core use cases such as agentic software engineering are likely to need approaches beyond next-token prediction. They are Σ₂ᴾ complexity problems requiring multi-objective optimization and likely a combination of Transformers and other methods. 8. Current forecasts in memory makers are built largely on quadratic attention. That will not persist: we are already seeing work from DeepSeek, Minimax and Nvidia that can cut RAM needs by 80% or more. 9. This means semiconductor valuations are substantially overinflated and will go through the traditional glut versus shortage cycle. 10. For foundation model providers: lower costs with competitive differentiation is good. However, lower costs with a lack of differentiation would mean lower revenues. This makes it harder to (a) service commitments and (b) pay back investors. 11. Leverage is substantially higher than in previous cycles, evidenced by leveraged ETFs, call option activity and margin loans. Korea is particularly susceptible. 12. 0DTE options create a profile that has stronger parallels to portfolio insurance and 1987 than any other point I can remember. 13. The combination of exponential increases in call activity coupled with the ties of semiconductors to structured products means there is a non-trivial systemic risk to the financial system. 14. Implied earnings growth rates are inconsistent with other periods in history. 15. Macroeconomically we cannot and should not fund exponential cost increases. History has shown us repeatedly that there are better ways (see Quick Sort and Simplex). 16. Significant supply is hitting the market via IPOs. –– Taken together: costs and competition are increasing while revenue growth is likely slowing. Valuations are fragile and prone to technology disruptions that are already here. Systemic financial market risk is extremely high.

在 vibe 了五个多月之后，我不觉得 SOTA 模型公司有能力吃下所有软件，我也不站在使用 FDE 改造现的企业来挣点快钱的那一侧，相反的，我是坚定的 AI Native company 的支持者，我认为所有企业都会被 AI Native 的组织所替代，而我坚信这一变化会持续数十年，这些未来的软件公司，会创造本世纪最伟大的投资机会。

Every software company just got a second life and Jensen just explained why (Save this). The conventional fear was straightforward, AI agents replace human workers, human workers use software tools, therefore agents destroy SaaS. Jensen Huang stood on stage at Computex 2026 and walked through exactly why that logic is backwards. Agents don't replace software, they consume it at machine speed, around the clock, without weekends. Here's the actual architecture Jensen laid out. An agent isn't just a large language model but rather an LLM sitting inside a harness that manages memory, orchestrates tool use, routes context, and plans iterative actions. That harness has to constantly call tools, spreadsheets, databases, browsers, and code engines, with every reasoning loop triggering another tool call. A human might use Salesforce 40 hours a week, an agent running inside a company uses it 168 hours a week and never misses a context window. The GitHub data Jensen showed on stage makes it tangible, 90 million pull requests merged, 1.4 billion commits, and 20 million new repositories created every month. As of April 2026, GitHub is processing 275 million commits per week on pace for roughly 14 billion by year end, a 14x explosion in a single year and AI agents are the source. Pull requests opened by AI agents went from 4 million in September 2025 to 17 million in March 2026 more than 4x in six months. That's AI becoming the largest software user on earth. Goldman Sachs quantified the downstream effect last month, token consumption is expected to multiply 24x by 2030, reaching 120 quadrillion tokens per month globally. A traditional chatbot consumes roughly 1,000 tokens per session, an embedded copilot burns 5,000 tokens per day while a continuously running enterprise agent? Over 100,000 tokens per day. The software companies that figured this out first are already printing money, Salesforce Agentforce hit $800 million ARR growing 169% year over year, with 29,000 deals closed. ServiceNow's Now Assist crossed $600 million in ACV, just raised its full year target to $1.5 billion, and told investors that when its agents replace a 20-person support team, total ServiceNow spend by that customer grows more than 5x even after accounting for reduced seat licenses. Workday delivered 1.7 billion AI actions across its platform in fiscal 2026. The key unlock Jensen pointed to and what investors need to understand is MCP, the model context protocol is the interface layer that makes software agent-readable. Software that supports MCP can be called by any agent, from any model, through any harness. Anthropic created it, OpenAI, Microsoft, and Google all adopted it and it was donated to the Linux Foundation. It is effectively becoming the HTTP of agentic computing. Software companies with native MCP support are plugged into the agent economy. Software companies still waiting are one product cycle away from becoming invisible to the fastest-growing category of software users in history.

There is a brand new AI infrastructure company that just hit the public market and only a few people on X are talking about it. $BRUN. I think this is one of the most overlooked opportunities in the entire AI buildout right now. Here is what Boost Run actually is. They deliver scalable AI and high performance computing infrastructure. GPU compute, CPU nodes, managed Kubernetes, and storage for enterprises that need control, speed, reliability, and compliance. They just began trading on Nasdaq in May after coming public through a SPAC. And the relationships they have already locked in are what make this so compelling. Start with NVIDIA. $BRUN is an NVIDIA Preferred Cloud Service Provider and just achieved NVIDIA Exemplar Cloud validation. In a supply constrained market where everyone is fighting for GPU allocation, being a preferred NVIDIA partner means you get access to next generation chips that competitors cannot easily get. That is a massive structural advantage for a company this size. Then look at Dell. $BRUN entered into a $1.44 billion purchase agreement with Dell Technologies for AI compute and storage infrastructure. They also expanded their relationship with Dell Financial Services to align flexible capital with customer contracts. Dell just posted one of the best quarters in its history with AI server revenue up 757% year over year. $BRUN is plugged directly into that supply chain. And the contracts are already coming. They signed orders for 5,000 NVIDIA B300 GPUs across their facilities with a combined contract value of roughly $471.7 million over 36 months. Take or pay structure. The customer pays for the full term regardless of usage. That is locked in recurring revenue. A newly public AI infrastructure company. NVIDIA preferred partner status. A $1.44 billion Dell agreement. $471 million in contracted GPU revenue already signed. And a market cap that almost nobody has discovered yet because it just came public. This is exactly the kind of early stage gem that gets found before it runs.