Organisms need, to use the metaphor of Marcus Aurelius, to turn obstacles into fuel—just as fire does
– Nassim Nicholas Taleb, American mathematical statistician.
On Friday, I had never heard of an investment banker named Liang Wenfeng.
By Monday, I was scrambling to decode an algorithm he’d built — one deeply rooted in the Jacobi model (a mathematical approach named after the German mathematician Carl Jacobi) I’d mastered years ago.
How on earth did I miss this?
I haven’t slept properly since.
The frustration’s been eating at me, so much so that I ended up throwing a drinking glass into one of my Bloomberg Terminal screens.
I’m livid.
Liang’s work is astonishing, the kind of mathematics that not only challenges the limits of my understanding but makes me question why I ever stopped pushing those limits in the first place.
Could this be the blueprint for the next-generation trading architecture?
The question wouldn’t leave my mind.
Liang is the driving force behind DeepSeek, an artificial intelligence (AI) model born out of High-Flyer, a hedge fund managing $8 billion in assets and one of the largest quantitative funds in China.
High-Flyer is famous for using AI to push trading strategies to their maximums, but DeepSeek is on another level entirely.
A Beijing-based associate told me that Liang insists on being thought of as an engineer, not a trader.
He isn’t one for the trappings of wealth or fame but once said he wanted the respect of the U.S.-led investment banking world.
Well, he’s certainly earned it.
He’s a pioneer in China, applying deep learning—an AI technique modelled on the human brain—to create predictive models that leave even seasoned quants in awe.
Unlike high-frequency traders, who chase microsecond advantages, Liang and his team have taken a different approach—medium-frequency trading.
Their models hold positions longer, finding an edge in a space most quants overlook.
It all took me back to 2011 when I was among 20 top analysts handpicked by former Goldman Sachs CEO Lloyd Blankfein to craft a series of equations and code breakers for Jim Simons, the quant godfather who started Renaissance Technologies, the hedge fund with the ridiculous average annual returns (66 per cent or so, for decades. And stop emailing, it wasn’t a Ponzi).
I devoted several years of my life to the project and it was rewarding, though a bit of a nightmare that pushed me dangerously close to committing suicide.
Simons ran Wall Street’s most secretive firm.
Even his rivals didn’t want to speak with me, out of fear of offending Simons.
I’ll never forget the 200-hour quant boot camp Simons put us through.
Out of 155 participants, only six of us including me managed to make it to the end.
For the uninitiated, “quant” or quantitative analyst is a fancy way of describing the systemised, algorithmic approach to trading—ordered thinking distilled into mathematical models.
Simons’s directive was brutal: build everything from scratch.
No shortcuts.
No imitating human grandmasters.
Just pure, original thinking.
Rest wasn’t part of the equation. Neither was mercy.
Simons, who died last year, would chain-smoke his way through the sessions, unfazed by our growing exhaustion.
Fast forward to last weekend, I kept seeing mentions in our circles of me, Simons and something called DeepSeek.
Every time I asked about it, I got blocked or ignored.
Should I have known what DeepSeek was? Probably.
For the past few days, I’ve been buried in simulation models, analysing scenarios around tariffs and crypto.
So when I first heard about it, I dismissed it as just another altcoin.
Then something strange happened.
One of my old equations perked up on a math database (we creators don’t usually obsess over those rankings, we generally only check every hour or so).
By Sunday, I read the coverage of DeepSeek’s model through a link shared by a colleague and elsewhere.
Fascinating article, but I still didn’t know what this had to do with me.
That changed at 3.11 am on Monday.
A colleague forwarded me an email with the subject line:
“DeepSeek—have you seen this?”
The message read:
“That new China-made AI model crushing our tech stocks’ positioning today? The guy behind it, Liang Wenfeng, also runs a quant hedge fund.
Interestingly, he wrote the preface to the Chinese edition of the Simons’ research paper.”
And just like that, the puzzle pieces clicked.
DeepSeek’s breakthrough sent shockwaves through the market, posing a direct threat to nearly every company carrying a market capitalisation of more than $1 trillion.
The selloff that followed seemed overdone and deeply consequential.
I started digging into the details and confirmed Liang’s connection to Simons.
His profile was right there on a Chinese government website.
“Born in 1985 …”
I froze. He’s one year older than me.
“Liang drew inspiration from Simons and his training using advanced mathematical techniques in 2011 …”
Wait, what?
I stared at the screen, baffled.
That’s when my wife video-called me after learning the news.
Sceptical that anyone could be so obsessed with my work, she asked if it was the same person.
“You’re not just seeing what you want to see, are you?”
It was undeniable.
I tracked down a translation of Liang’s preface to Simons’ paper, and it was obvious the man had studied every decimal point of my execution algorithms.
“Whenever I encounter difficulties at work,” Liang wrote, “I recall Simons’s words: ‘There must be a way to model prices.’”
Liang continued, “The publication of this research paper unravels many previously unresolved mysteries and provides a wealth of lessons to learn from.”
Even my best and brightest teams didn’t pull that much out of it.
Absolutely impressive.
Nonetheless, his algorithm won’t get to rest on its laurels.
Coordinating with other battle-hardened quants, I’ll be unleashing far more depth — ‘firepower,’ if you will — with a level of precision that, for the first time in my career, is truly intimidating.
I spent the rest of the day trying to get in touch with Liang, or even someone who works with him or has interacted with him.
Hours later, no luck.
I began having flashbacks to dealing with Simons — the mental chess, the dead ends, the creeping realisation that someone was always two steps ahead.
No wonder Liang idolises him so much.
So, Wenfeng, if you’re reading this, send me a note.
Give me a call.
Shoot me an email. Send a mail pigeon.
I don’t care how — just send me something.
I’ll be waiting.
The views expressed here are those of the writer and do not necessarily represent the views of the Sarawak Tribune.
