hermaeus @polyhermaeus

OKAY THIS IS GETTING STUPID, OPUS 5.5 + GROK + JEV READ 12,040 COMMENTS ON X IN 24 HOURS AND PROMO CAMPAIGNS PAID ME $2,316.40 FOR WHAT THEY FOUND 34 payouts from 7 promos. the whole run cost $4.12 promo campaign = a project that pays people for useful info about it on X. hype gets nothing, a real number gets paid grok = the reader. it pulls every post and every reply from the last 24 hours opus 5.5 = the finder. it reads the replies and marks the ones with real info in them jev = the gate. it matches each find to a live promo and submits only when the numbers hold how the run went → grok pulled 1,670 posts and 12,040 replies → 312 accounts were bots, jev dropped them before anything got submitted → opus 5.5 marked 412 replies as real info → jev matched 36 of them to 7 live promos → 2 got rejected, the source turned out to be a bot thread → 34 got paid, between $37.80 and $110.80 each where the money came from 01 AGENT WEEK · 9 payouts · $742.80 02 AI CODING · 7 payouts · $512.40 03 L2 SEASON · 5 payouts · $318.00 04 WALLET APP · 4 payouts · $264.60 05 GPU CLOUD · 4 payouts · $221.40 06 DEFI DATA · 3 payouts · $168.20 07 MEME WATCH · 2 payouts · $89.00 the first one paid $60.60. a post said L2 fees hit a new low, a reply under it said fees are low because the blocks are empty. that reply was the info honestly the replies that paid best looked boring. one guy just noticed that 41 accounts in a meme thread were made on the same day by hand that's 61 hours of reading replies. jev made 9,617 calls and the run cost $4.12, so every $1 of compute brought back $562 meme watch paid the least, $89 for 2 finds. most of that thread was bots saying gm at 0:04 the first reply gets paid at 0:35 the wallet stops at $2,316.40 ↓ bookmark it, the promos refresh every week send this to whoever still farms promos with AI slop replies would you read 12,040 comments yourself for $2,316, or let jev do it?

Noisy @noisyb0y1

6 days ago

x.com/i/article/2105…

9 23 197 394K 498

@issatheblows treating it like a new hire actually makes a lot of sense

@thegreatest_sv so basically a bouncer for claude, i kinda like that

@Atenov_D Two percent keeps one bad trade from ever being a bad month

Gilbert Strang taught math at MIT for decades. Now his lectures are free. In this 13 minute lecture he gives an overview of differential equations. He starts with money in a bank account. Interest pushes it up, withdrawals pull it down. Then he moves on to springs and matrices, and ends with the heat equation. Wall Street runs on the same math. The Black-Scholes formula that prices options is a PDE, and it looks a lot like the heat equation he writes on the board at the end. Algo trading uses it too. A real market has hundreds of assets moving at once, so the equations become one big matrix. Eigenvalues and eigenvectors, the same tools from the lecture, split that mess into simple pieces. Quants use them to find stocks that move together. Then a bot buys the cheap one, sells the expensive one and trades the gap between them. It makes money whether the index goes up or down. Quants are among the best paid people in finance. The starting point is one free video.

Lumen @lumenxbt

2 months ago

x.com/i/article/2082…

79 469 3K 12.6M 9K

@e_sofia9 Glad you liked it. The full lecture is worth the 14 minutes

@verumidest Crowding kills market-neutral when every quant holds the same pairs

@verumidest diversified on paper, but every dollar still runs through the same pipes

@Benderxxx7 Most people sell the dip. Buffett sells only when the thesis breaks

@kobaHUB how many shorts would survive paying premiums while housing kept rising?

Jim Simons was a math professor. His Medallion Fund averaged about 66% a year before fees for three decades and left him a fortune above $30B. Quant models lean heavily on linear algebra, and MIT put a full course on it online for free. Gilbert Strang has taught at MIT since 1962. Lecture one starts with a matrix times a vector, Ax, and he reads it as a combination of columns. Ax = x₁·col₁ + x₂·col₂ + x₃·col₃ Take every possible x and you get a whole space, the column space. Its size depends on how many columns are independent. That count is the rank. Then he factors the matrix. A = CR C keeps only the independent columns. R tells you how to rebuild the rest from them. Read the same product by rows and you get a theorem that lands fast: independent columns and independent rows always come out equal in number. The smallest piece is a rank one matrix, a column times a row, u·vᵀ. Any matrix is a sum of those. Real data matrices are too big for memory, so you don't read them. Multiply by a random x and Ax gives you a typical vector from the column space. A hundred random x give you a good feel for the whole matrix.

@0_oRai lol turning $10 million into $1 billion from one simple pattern is wild

@0xjakke 42 days is insane, but the customer angle makes total sense

@vontrixai admitting on camera that Llama 4 missed is rare. most CEO would have quietly changed the subject

@AI__Avenger asking for 30 days instead of the whole account is the smartest part. it makes saying no feel strange.

@mindarchx moving a dollar into your own category is the best line here

@ai_with_jyoti only betting when your number beats the bookies odds is the whole edge. most people never get that disciplined

@AI__Avenger the sharding part is what got me lol 10,000 people per database instead of 6 million, such a simple fix

@venmilxbt honestly I'd have panicked and taken the cash lol good thing he's built different

@0_oRai Inaction is a mistake too, just one that's harder to notice