Alexandr Tor

@amitisinvesting The market doesn't collapse because everyone expects it. It collapses because no one can imagine it. While the bears chew over the same arguments about CapEx and inflation, the "Black Swan" quietly grows in the shadows, where no one is looking.

Qullamaggie on Anyone Can Do This “Yeah, I started with $3.5 million this year, I’m up like 800 or 900%. It’s insane and I made so many mistakes. Like I can tell you guys, any fucking idiot can do this. Like literally, just don’t be a moron — you have to learn how the market works. You have to stick to rules. You need to have rules in the first place. Like if I can do it, anyone can do it. Literally nothing special about me, absolutely nothing. Or well… my mom says I’m special but I’m not sure it’s a good thing, to be honest.”

One thing I've learned in sports, business, and media: when people can't attack the argument, they attack the messenger. I showed three clips. The conversation became about my résumé. That's usually a sign we're no longer discussing the actual point.

@FlowbyBobby You missed one: you should not be chasing power, you let it come to you. IYKYK

You should not be chasing women, you let them chase you. You should not be chasing money, you let it come to you. Make your gravitational force so strong that you attract everything you dream off. The world is yours, you just have to believe it.

The market is no longer buying AI exposure, it's buying scarcity by layer. The book I keep is up 111.83% over the last 6 months, while the S&P 500 is up 11.02% over the same window. That leaves the book with +100.81% alpha (only tracking 112 companies rn) But the headline return is not really the point, as the more useful signal is what is happening underneath it. Across the book, 106 of 113 names are up, so this is not a single-stock story or one lucky basket carrying the whole thing. At the same time, the dispersion between baskets is extreme: InP & Substrates is up 438.79%, while Connectors & Interconnect is down 0.14%. That is a huge spread inside the same AI infrastructure trade. That is what I wanted the dashboard to show. AI Bottlenecks is not meant to be a stock list. It is a way to rank the physical layers of the AI buildout: substrates, photonics, memory, packaging, power, cooling, networking, cloud, storage, construction, and the rest of the supply chain that sits underneath the model headlines. Each basket is equal-weighted, and each basket gets one vote in the book. The point is not just to ask whether AI infrastructure is working as a theme. The point is to ask which layer the market is re-rating first, which layer is lagging, and where the constraint may be moving next. The 6-month leaderboard now looks like this: InP & Substrates: +438.79% Memory Supercycle: +382.30% Photonics / CPO: +295.90% HBM / Packaging: +176.79% Custom Silicon: +102.31% Storage: +87.36% Lithography & Fab Tools: +80.86% Construction & MEP: +61.35% AI Cloud / Neoclouds: +58.07% Networking / Retimers: +56.89% Power & Grid: +54.71% Cooling: +45.11% Rare Earths: +41.18% Outliers: +10.96% Connectors & Interconnect: -0.14% The top of the table is telling a pretty clear story: the strongest layers are the ones tied to substrates, memory, photonics, advanced packaging, and custom silicon, the things that physically gate the next step of the AI buildout. These are the pieces that matter when you start asking what has to be true for Rubin Ultra, HBM4E, higher-density clusters, and the next round of inference capacity to actually ship. The bottom of the table is just as important. A weak basket does not necessarily mean the layer is irrelevant. It may mean the market already paid for the story, the chokepoint did not bind this quarter, or the catalyst was more geopolitical than operational. That is why the rank order matters more than the headline return. A refinery report gives a useful analogy. A barrel of crude is a barrel of crude, but the yield by cut, gasoline, diesel, kerosene, asphalt, residue, tells you where the constraint was in that quarter. The commodity is the same. The bottleneck moves. The report ranks the pressure points. That's what this book is trying to do for AI infrastructure. The visible AI trade is usually shown as one line. This is the same trade broken into 15 physical lines, each one tied to a different part of the route card. The book is up because most layers are up. It is interesting because the ordering of those layers tells you what the market thinks is scarce.

Ain’t so fun when the rabbit’s got the gun

@KempeLab Congratulations? I’m not an expert Dr. Kempe but are the posts I keep seeing claiming that this is half baked slop true? @ylecun

We did it! Thrilled to announce that with my team at FAIR Meta we released 25+ auto-formalized mathematics textbooks covering analysis, algebra, geometry, topology, combinatorics, probability, statistics, PDEs, number theory, and theoretical computer science - the largest such effort to date.

Honestly an embarrassing thread. Why not hire 5 ppl to look at your work before doing something public like this. Would actually feel bad if I was a PhD student with my name attached to this project.

@BrammerAyse Easily AI!!

I’ve been getting emails from high school students at some of the most prestigious schools in New York (Dalton, for example). What strikes me is how specific these emails are, they all reference my research in detail and explain exactly how it connects to the projects they want to pursue. The level of detail makes me think these are not simply AI-generated emails. Given that a high school student interested in, let's say, China or Germany already knows this much about my work, I wonder if there are people paid for crafting highly tailored outreach emails. Or am I underestimating AI?

@Glitch_Trades @Simply0DTE

@Glitch_Trades Can you explain why msft, meta had the pump while intc, google, nvda had a dump? I’m pretty sure it is the MSCI rebalancing but how to figure out what will pop and what will dump? Asking not just for me

That EOD pump was too powerful on $MSFT. Swung full.

Opus 4.8 is insane, nothing will be the same after this model 💀

500k lines of code, but the statements are just wrong or excessively simplified… Thanks to Christian Merten and Jack McCarthy for looking at this in detail on Zulip (link below). Note for example that their definition of a compact orientable manifold is "a compact manifold" 😂

Was talking with a younger trader at Postman Agents meetup this week- he asked what to focus on starting out trading. I loved this question because I can objectively look back now on my journey. Here’s how I framed it from my playbook for this young man: first: do not treat this as a way to make money, this is the wrong mental process and just doesn’t work with human psychology. View this more as a math problem that you are trying to solve- really focus on training your brain to be constantly solving these little micro problems that emerge every day, but keep vision of the bigger problem at hand. Second: be in this for the long run, and dig deep on your mental relationship with money. Do you have an abundance or scarcity mindset? Is it just a digit or does it represent your physical time or labor? This is so important to be able to detach your mind from thinking this is anything else but a database full of numbers. Third: learn how to loose and mentally move on quickly. Become completely indifferent to any one outcome. When the event is over move on to the next one as if nothing has happened in the past.

@LongFormMath I would go with Riemann...Riemann hypothesis, Riemannian geometry. Or maybe Grothendieck, for his way of thinking more so than his spectacular work. Erdos problems have become popular recently to benchmark AI, but I think most math research is not really about Erdos problems.

I ended my 2025 math history book discussing Erdős and how his conjectures continue to stimulate math research. The book’s final sentence was: “Of all the dead mathematicians, Erdős is the most alive.” Given the last couple weeks, I suppose that holds up well? At least for the time being…

With everything we are hearing right now about ticks this seems like good information to share. “Here’s what I’ve learned after more ticks than I care to count. First, whatever your uncle told you, forget it. No matches. No nail polish. No Vaseline. No soap on a cotton ball. All of those do the same terrible thing, they stress the tick out, and a stressed tick empties its gut back into the bite before letting go. Which, if you think about what that actually means for a second, is literally how Lyme and the rest get transmitted so you’re not speeding up its exit. You’re making it throw up into you. Fine-tipped tweezers. Grip right where the mouthparts enter the skin, not the body, the head. Pull straight up, steady, no twisting, no jerking. It’ll feel like it’s resisting because it is, the mouthparts are barbed. Just keep the pressure on and it lets go in a few seconds. If a piece breaks off in the skin, leave it alone. Your body pushes splinters out. Digging around with a needle does more damage then the fragment ever would. Clean it with alcohol or soap. Wash your hands. Now here’s the part most people skip: don’t flush the tick. Tape it to an index card. Clear packing tape right over the body, write the date and where on your body it was, and stick the card in a drawer. If you come down with anything weird in the next 30 days, rash, fever, joint pain, that flu-that-isn’t-flu feeling, that tick goes with you to the doctor. Some labs will test the tick itself, which is faster and often more reliable than waiting for antibodies to show up in your own blood. A dated tick taped to a card is one of the most useful things you can hand a doctor who’s trying to figure out what’s wrong with you. The other thing worth saying out loud: if the tick was engorged when you pulled it, and you can’t swear it was off your body within 24 hours, call your doctor that same day. Don’t wait for a rash. Fewer than three out of four Lyme cases even produce the classic bullseye. A single preventive dose of doxycycline within 72 hours of a deer tick bite cuts the Lyme odds way down, and most docs in tick country will write that prescription without giving you a hard time, especially if you walk in with the tick taped to a card and a clear timeline.”

@Glitch_Trades Skylit on Top

Being ahead of the rotations is the pinnacle of trading. If you want my full guide on how I pick and identify where big money is poised to rotate their money next, LIKE + RETWEET + COMMENT ‘Skylit On Top’ and I’ll send you the full guide on how you can front run the whales like I do. Yesterday was $META, today is $MSFT.

@Glitch_Trades Aha, I see thanks I’m new and learning so much from you!

@9hOily Bottom of pennant yesterday. Meaning if reversal and breakout, parabolic move incoming.

Take some trims or do whatever you want. Sniped absolute lows. $MSFT When others become doubtful, that’s when you pounce.