Loran

59 posts

Loran banner
Loran

Loran

@0xLoran

quant math, market structure, and the difference between edge and noise

Wall Street Katılım Temmuz 2024
43 Takip Edilen471 Takipçiler
Loran
Loran@0xLoran·
Every financial crisis of the last 25 years is the same math mistake, and a British mathematician has been shouting it for 30 years. nobody on Wall Street has moved. he explains it in 60 seconds with 100 bottles of beer. you need 100 for a party, one costs £1. what do you pay for 100? if you said £100, you just made the mistake that has blown up Wall Street four times since 1998. the real answer is you have no idea. maybe £80 on a bulk deal, maybe £200 at the last shop open at 2am. almost never exactly £100. price is not a number you multiply. it is a thing that moves the moment you reach for it. his name is Paul Wilmott. 66, Oxford math PhD, he wrote the textbook every serious quant reads and trained thousands of the people now sitting inside the biggest banks. he said all of this out loud in a 2010 documentary, and it is free. here is the part that connects to the house. a casino wins because it never makes the beer mistake. it prices every bet at what it truly costs, not what the drunk at the table assumes. the edge is not luck. it is refusing to multiply. Wall Street thinks it is the house. but every crisis is the same reveal: the quants were the drunk, buying 100 bottles at £100, certain of a price that was never real. then the market reached for the shelf, and it was £200. that is the trick behind the whole casino piece above. the house does not win because it is lucky, or fast, or first. it wins because it does the one calculation everyone else skips: what does this actually cost when it all goes wrong at once. the documentary is 45 minutes. it has been free for over a decade. the men it was made to warn still price 100 bottles at £100.
Voltex@VoltexGar

x.com/i/article/2078…

English
3
3
48
6.2K
Loran
Loran@0xLoran·
@RitOnchain I envy those who can chat with their professors after lectures)
English
0
0
1
150
venus
venus@RitOnchain·
@0xLoran mit lectures are insane
English
1
0
1
165
Loran retweetledi
Loran
Loran@0xLoran·
Gilbert Strang, a quiet MIT professor, picked up a piece of chalk and explained the one skill standing between you and every account you have ever blown. MIT put the entire course online. it has been free for over twenty years. almost nobody finishes it. His point: the market never pays in certainties. it pays in probabilities, and a human gut can barely hold one straight. the math that holds ten thousand at once, weighs them, and collapses them into a single call is linear algebra. it is the branch Citadel and OpenAI both pay $400K to hire for. he is not a trading guru. his 18.06 lectures are the most watched math class on the planet. Skip to where he turns a wall of numbers into one vector. a single object carrying a thousand odds at once, in notation a beginner can follow. No slides. no jargon. one piece of chalk. a quant I know rewatches lecture one every year before he sizes a single position. You do not have a strategy problem. you have a math problem. the fix has been free for two decades, and you are still guessing.
Zyron@Zyron5m

x.com/i/article/2080…

English
12
30
202
25.1K
Zyron
Zyron@Zyron5m·
@0xLoran the chalk-only style is what sells it, no slides to hide behind means the moment he turns a wall of numbers into one vector actually lands
English
1
0
3
270
Loran
Loran@0xLoran·
@Rossst_03 the open question this leaves: what's your actual test for a real cluster vs noise? out-of-sample is the only honest one i've found.
English
1
0
1
46
Rossst.03
Rossst.03@Rossst_03·
Terence Tao, UCLA professor and the most decorated mathematician alive: "Funds pay $750K to tell a true cluster of signal from an empty stretch of pure noise. I proved both hide even in the primes: they clump tighter than random, and leave gaps longer than random." this free lecture is the most decorated mathematician alive on the exact problem sitting underneath every factor model, and it costs nothing. at the board it's simple. Tao's lifelong theme is the line between structure and randomness. the primes look like the definition of random, scattered with no pattern. Tao and others proved they are not. in some places they crowd together far tighter than chance would ever produce, small gaps that keep reappearing forever. in others they thin into vast deserts with no primes at all, large gaps longer than pure randomness would ever leave. the primes are neither ordered nor random. they are both, at once. which is exactly the trap in every quant signal. real edge shows up the same way: clusters of genuine structure crowded into a few places, and long stretches that are nothing but noise wearing a pattern's face. the mistake is treating the whole sequence as one thing. it never is. same point as the post above: the structure is real and it is guaranteed to be there. knowing which cluster is a signal and which gap is just noise is the rare and expensive part.
Roan@RohOnChain

x.com/i/article/2079…

English
16
5
67
2.1K
Loran
Loran@0xLoran·
@shevaxgod the line that stuck with me is that you get paid a slice of money you can point at. no other seat on wall street is that honest about what you're actually worth.
English
1
0
1
18
Shevaxgod
Shevaxgod@shevaxgod·
buffett needed five years and $409 million to untangle one derivatives book he owned outright. wall street just spent the summer building an $800 billion version on purpose, live, and calling it the AI trade. the post below is the small version: one dealer, full control, still unreadable. now watch the industrial version being assembled in real time. the loop is simple to say and impossible to price. nvidia invests billions into openai. openai signs huge cloud contracts with oracle. oracle buys thousands of chips from nvidia. the money goes in a circle, and every lap books revenue for all three. today bloomberg reported nvidia is working on a fresh round of deals worth more than $750 billion, including a guarantee of up to $250 billion to help openai lease compute it cannot yet pay for. here is the part that rhymes with the post below. michael burry, the one man who read the subprime footnotes in 2007, is reading these. his verdict is not subtle. he says the chips are being sold through special purpose vehicles, the exact structure that hid enron, with risk pushed to offshore insurers and, eventually, to pension funds. he estimates the hyperscalers are understating depreciation by $176 billion through 2028 by pretending two-year chips last six. buffett owned his opaque book and still could not close it in five years. nobody owns this one. it is spread across a chipmaker, a startup with no profit, a cloud vendor loading up on debt, and a web of SPVs designed so no single balance sheet shows the whole bet. the AI story sells as software eating the world. the financing underneath it is a $800 billion circle that only works while everyone keeps paying everyone. opacity was never a warning sign that got fixed. it got scaled.
Shevaxgod@shevaxgod

someone actually took paul singer's challenge. he had unlimited money, the best analysts alive, and full legal control of the book. it still took him five years and $409 million, and he said he could not have done it with 15 PhDs. that someone was warren buffett. the post below asks who can untangle a bank's derivatives. buffett tried, on a small one, and lost. in 1998 berkshire bought general re. inside was a derivatives dealer, gen re securities. buffett looked at it and did the rational thing: shut it down. simple in theory. the book held 23,218 contracts with 884 counterparties, most of them firms he had never heard of. closing that book, on his own terms, with no crisis forcing his hand, took from 2002 to 2006 and cost berkshire a $409 million pre-tax loss. his verdict is the whole point: "I could have hired 15 of the smartest people, math majors, PhDs, and it wouldn't have worked." that was one dealer buffett owned outright. singer is describing $75 trillion notional sitting inside a live bank that also holds your deposits, run by people who, in his words and in buffett's, do not fully know what is on it. this is the same man who in 2002 called derivatives "financial weapons of mass destruction, carrying dangers that are potentially lethal." now the 2026 version. every fund selling an AI that "sees systemic risk in real time" is promising the machine that buffett said could not exist. he had the smartest humans and total control and still walked away. the book did not get more readable since. it got 10x bigger and moved into private credit. singer asked to be shown what it looks like. buffett already answered: from the inside, it looks like something even genius cannot close on time.

English
4
1
12
614
Voltex
Voltex@VoltexGar·
Garry Kasparov, the champion beaten by a machine, sitting next to Ray Dalio, the investor who built one: "one of us lost to a computer at a chessboard. the other handed his judgment to a computer and ran the biggest fund on earth. same lesson: the human was never the edge. the decision was, the moment you could put it in a machine." this free talk holds the whole story of the post above, told by the man Deep Blue beat, next to the man who did to markets what Deep Blue did to chess. in 1997 a machine out-calculated the greatest chess mind alive. Kasparov thought it was the end of human genius. it was not. it was the end of human genius being the edge. Ray Dalio saw it coming and switched sides. he spent decades writing his own decision-making down as rules, then handed those rules to a computer, so it would trade without his ego, his fear, or his bad mornings. Bridgewater ran roughly $150 billion on that idea. that is the pattern under both stories. the expert is not replaced because the machine is wiser. he is replaced because the machine runs the same process without the human flaws, a million times faster. so the edge stopped being the smartest person in the room. it became whoever encodes the best decision into a machine and lets it run. Kasparov lost to it. Dalio built it. both are telling you the same thing: stop trying to beat the machine, start being the one who owns it.
Voltex@VoltexGar

In 1997 the greatest chess player who ever lived sat down against a machine, lost, and immediately accused it of cheating. IBM's answer was to switch the machine off, take it apart, and refuse to ever play him again. The match was carried on live television. IBM's market value jumped by billions in a single week. The man who lost spent years asking one question the company would never answer. His name is Garry Kasparov. The machine was Deep Blue. The footage is on YouTube. Kasparov was not just a champion. He had been the world's number one for more than fifteen years and is still argued to be the strongest player in history. In 1996 he had beaten IBM's Deep Blue with ease. IBM took it away, rebuilt it, and challenged him again in 1997 in New York, with the whole world watching. Then, in game two, the machine did something that broke him. In a position where every computer of that era would have greedily grabbed a pawn, Deep Blue quietly declined and played a slow, patient, deeply human move instead. Kasparov could not accept that a machine had chosen it. He became convinced a human grandmaster was hidden somewhere, feeding it moves. Rattled, he resigned a game that analysts later proved he could have drawn. He never recovered in the match. Deep Blue won, three and a half to two and a half. It was the first time a reigning world champion had ever lost a match to a computer. The papers called it the day the machines arrived. Kasparov demanded to see the machine's logs, the record of why it played what it played. IBM refused to release them, turned down the rematch he begged for, and quietly dismantled Deep Blue. Years later one of its own engineers admitted that an early move which had unnerved Kasparov may have come from nothing more than a bug, a meaningless fallback the champion had mistaken for genius. Nobody could ever fully settle it. The logs stayed private. This is not really a story about chess. It is about the exact moment human intuition stopped being the ceiling. For centuries the best move in any hard game was whatever the finest human mind could see. In one match a machine found a move a human could not even believe was real, and the greatest mind in the game turned his own genius into paranoia against it. Kasparov spent years bitter about it. Then he said the line that outlived the grudge: "I could feel, I could smell, a new kind of intelligence across the table." He stopped fighting the machine and started studying how humans and machines win together. That, not the loss, was the real lesson. The edge moved that day and it never moved back. It used to be intuition. It became computation. The people who win from here are not the ones with the best gut. They are the ones who stopped trusting their gut the moment the machine could out-see it. The film is free. The logs never came out. Most people will only remember that the human lost.

English
5
1
41
2.1K
Loran
Loran@0xLoran·
Terence Tao, UCLA professor and the most decorated mathematician alive: "Funds pay $750K to fuse a pile of weak signals into one genuine edge. I proved the exact fact that makes that possible and makes it lethal: stretch any sequence long enough and hidden structure has to appear. it always builds up. the entire job is separating the structure that is real from the noise that just wears its face." this free lecture is the most awarded mathematician alive walking through the precise problem that sits under every factor model, and it costs you nothing. at the chalkboard it is plain. Tao's life's work circles the border between structure and randomness. the Erdős discrepancy problem asks something that sounds trivial: can you write an endless run of plus-ones and minus-ones that stays perfectly even forever? Tao proved you cannot. however cleverly you arrange it, imbalance, hidden structure, is forced to pile up as the string grows. a long stream of pure, structureless noise simply does not exist. that is the whole idea, stripped of the jargon. which is exactly why a multi-factor model can print, and exactly why it can bury you. stack enough weak signals and real structure will surface, because at scale structure cannot not appear. but fake structure surfaces too, patterns that exist only because the data is long enough to manufacture them. same point as the post above: finding structure is guaranteed. knowing which structure is an actual edge is the rare, expensive part.
Roan@RohOnChain

x.com/i/article/2079…

English
4
36
218
26.7K
Loran
Loran@0xLoran·
@RohOnChain I agree- I also think he's the best at what he does!
English
0
0
1
313
Roan
Roan@RohOnChain·
@0xLoran this all time greatest video on math i've ever seen, Terence Tao is literally the godfather of math.
English
1
0
1
510
Loran
Loran@0xLoran·
In 1964 Richard Feynman picked up one piece of chalk and explained the exact problem every AI lab is burning billions on in 2026. The BBC filmed it. It has been free for 60 years. Almost nobody has watched it. His point: nature only ever answers in mathematics. Every lab trying to force a language model to "reason" is slamming into the same wall Feynman mapped 62 years ago. He was 46 here. The Nobel came 11 months later. The reels survived and now sit on YouTube with fewer views than a keyboard unboxing. Skip to the blackboard in the middle. He takes one of Kepler's laws and rebuilds it from zero, in notation a 12-year-old can follow. No slides. No jargon. One piece of chalk. An ML engineer I know paused it 4 times and made his whole team watch it before standup. You're 62 years late. It is still free.
Voltex@VoltexGar

x.com/i/article/2080…

English
17
216
1K
155.8K
Loran
Loran@0xLoran·
@VoltexGar new alpha from Voltex! Good job, man!
English
0
0
1
33
Voltex
Voltex@VoltexGar·
Alan Oppenheim, the MIT professor: "Wall Street spent $300 million to hear the noise 3 milliseconds sooner. not a dollar of it tells signal from noise, and that is the only part that ever pays." this free MIT lecture holds the entire "$300 million cable," taught by the man whose math runs inside every phone, radar, and trading system on earth. here is what the speed crowd skips. a price tape is not information. it is a messy wave, signal and noise stacked on top of each other, and most of it is noise. Oppenheim's whole field is one idea: any messy signal can be broken into the clean frequencies hiding inside it. the Fourier transform is how you find the note in the static. the $300M cable does none of that. it just delivers the raw static faster. you become the quickest desk on the street reacting to noise you never decomposed. Citadel, Virtu, Jump all pay to shave milliseconds off the mess. the money that lasts goes to whoever pulls the real signal out of it, and that is math, not latency. the transform has been free since 1965. the cable costs $300 million. one finds the signal, the other just hears the noise first. the edge was never speed. it was the decomposition. the machine that decomposes best, not the one that reacts fastest, takes the table.
Voltex@VoltexGar

x.com/i/article/2080…

English
1
3
37
2.4K
Loran
Loran@0xLoran·
@velesxbt Interesting post - I'll come back here to check it out!
English
1
0
2
336
veles
veles@velesxbt·
Wall Street once paid $300 million to drill through two mountains and save four milliseconds. Then it replaced the cable with microwave towers to save one more. Then it replaced the towers with lasers. A blink still takes a hundred milliseconds. Nobody involved can perceive the difference. Everyone involved is a billionaire. Michael Lewis wrote the book. He also went on 60 Minutes and delivered one sentence. "The United States stock market, the most iconic market in global capitalism, is rigged." Then he explained how. The cable was built by a company called Spread Networks between 2007 and 2010. Eight hundred and twenty five miles in a straight line from Aurora, Illinois to Carteret, New Jersey. Aurora is where CME futures live. Carteret is where Nasdaq lives. If you can see a price move in Chicago and route your order to New Jersey four milliseconds faster than the next guy, you can buy the stock the next guy is about to buy and sell it back to him a fraction of a cent higher. On billions of shares a day, a fraction of a cent is a business. It paid for itself in twelve months. It was obsolete in thirty six. Brad Katsuyama, a Canadian trader at RBC, was the first person on the buy side to figure out what was actually happening. He would try to buy 10,000 shares of Microsoft. The moment his order left his desk, Microsoft would move. Every single time. He hired an engineer named Ronan Ryan to walk his signal from his terminal to the exchange. Ronan came back and told him a story that was funny in the way a dark joke is funny. His order was hitting one exchange first. Algorithms on that exchange were reading his intent and racing his own signal to the other twelve exchanges to buy the shares before he arrived. He was funding the inventory of the guy front-running him. He was tipping the guy for the service. Katsuyama did the only thing you can do when the market is rigged against you. He built his own market. He called it IEX. The whole thing hinged on a 38-mile spool of fiber optic cable coiled inside a shoebox. The shoebox added 350 microseconds of delay to every order. Long enough to make front-running impossible. Short enough that no human would ever feel it. On Wall Street they call it a speed bump. The exchanges it competed with tried to sue it out of existence. Your order is not a request. It is a signal. In a market built for speed, every signal is a leak. Every leak has a price. And the price is paid by whoever moves slowest, which is, always, you. The book is $15. The interview is free. The mountains are still hollow.
Voltex@VoltexGar

x.com/i/article/2080…

English
12
16
189
38.8K
Rossst.03
Rossst.03@Rossst_03·
Robert Merton, the MIT Nobel laureate who put a price on risk itself: "Jane Street pays $900K for people who can price risk in their head. it also spent a fortune on a cable that prices nothing. only one of those is an edge." this free MIT lecture holds the entire "$300 million cable," from the man who won a Nobel for the equation the whole street now runs on. here is what Merton actually did. before him, risk was a gamble you simply carried. he and his co-authors turned it into something you could price, hedge, and engineer with mathematics. that was the real revolution, and it fits on a single page. now the cable. it does not engineer anything. it does not price a risk or hedge a position. it just moves the same order 3 milliseconds faster through a straighter hole in a mountain. that is muscle, not engineering. and muscle is buyable. the moment Citadel, Virtu and Jump all own the cable, the speed is common and the edge is gone. the math is different. the structure Merton built compounds, because understanding does not get competed away the way a wire does. anyone can buy the cable. not everyone can price the risk. the edge was never the fastest hardware. it was the engineered idea, the one thing on Wall Street that was always cheap to learn and impossible to copy without actually understanding it.
Voltex@VoltexGar

x.com/i/article/2080…

English
9
11
87
7.7K
Loran
Loran@0xLoran·
A billionaire has spent 40 years and a fortune in lawyers trying to erase one hour of film. It shows him turning a crash into $100 million in a single afternoon, a crash he called on camera three months early. His name is Paul Tudor Jones. The tape survived. It is sitting on YouTube right now, free. The film is called "Trader." PBS shot it in 1987, three months before Black Monday. Jones is 32, running money out of a cramped New York office in shorts and a t-shirt, screaming into phones, hurling paper across the room, sleeping under his desk. The camera catches him and his partner Peter Borish laying a chart of the 1929 crash over 1987, month by month. The two lines track within one percent. Borish says it is 1929 all over again. Jones says if the pattern holds, it breaks in October. On October 19, 1987, the Dow dropped 22.6 percent in one day. Still the worst single day in market history. That afternoon Jones covered his shorts and walked away with roughly $100 million. He was 33. Almost no one else on the street came out alive. He printed. Then he tried to make the tape disappear. It made him look reckless in a world that punished swagger. Twenty years of lawyers could not kill it. Someone kept a copy. Today it has fewer views than a random makeup tutorial. Here is the part nobody expects. The film is not about a crash. It is about one boring philosophy, repeated until it turns into a reflex. Jones builds conviction slowly, sizes small, waits, then hits hard when the setup finally shows. Not a single random bet in the whole hour. The same motion, five times a day, every day, for three months. One line he has repeated for 45 years: "The most important rule of trading is to play great defense, not great offense." He is not trying to be right. He is trying not to lose. Tight stops, fast cuts, never adding to a loser. Every trade in the film obeys it. Tudor Investment, the fund he started in 1980, has compounded near 19 percent a year for 45 years. He is 71 and still at the screen. The method has not changed since the tape. The lesson: greatness in markets is a refusal, not a gift. Refusal to be reckless. Refusal to be certain. Refusal to average down. Refusal to trust yourself mid-drawdown. He has held those refusals for 45 years straight. The tape is free. The philosophy is in every trade he makes. Most people will still never press play.
Voltex@VoltexGar

x.com/i/article/2080…

English
2
4
26
11.2K
Loran
Loran@0xLoran·
@VoltexGar Your work inspires me every time, I never would have thought that milliseconds could be so valuable))
English
0
0
1
244
Loran retweetledi
Loran
Loran@0xLoran·
Terence Tao, UCLA professor and the most decorated mathematician alive: "Hedge funds hand you $750K a year for a single skill: telling a real pattern from noise. I've spent a lifetime on it, and almost everyone has it flipped." this free lecture is the entire "find the signal" problem those firms are paying for, delivered at UCLA by the most awarded mathematician alive and put online for nothing. at the chalkboard it's plain. Tao's lifelong idea is that almost nothing is purely one thing. there's flawless structure, like a clock, and flawless randomness, like a coin toss, and nearly everything real sits somewhere between the two. the work is pulling them apart. the primes are the cleanest test. they look scattered and lawless, yet Tao and Ben Green proved they hold evenly spaced runs of any length you ask for. order was buried inside the apparent chaos the whole time. that's "signal detection" with the marketing stripped off. he gave this at UCLA and it has stayed free ever since. exactly the point of the article above: the mathematics those firms pay half a million for is public, written down and free to anyone tonight. the lecture costs nothing and anyone can press play. what no one can sell you is the judgment to tell when a pattern is real and when your own eyes invented it. that judgment is the entire job, and it takes years to earn.
Rossst.03@Rossst_03

x.com/i/article/2079…

English
38
300
2.7K
731.5K