The big quant convergence
Also: Fantasy DMOing, AImaxxing, SpaceX, index shenanigans, prediction markets and the strange melancholy of slaying video game monsters
Last year Alphaville wrote a big post about the nascent convergence of proprietary trading firms and quantitative hedge funds, as the former adds slower signals to their arsenal and the latter explores faster strategies.
As a result, the line between what used to be called a high-frequency trading firm and a “statistical arbitrage” hedge fund is getting a little blurred. As the head of one big quant hedge fund told Alphaville at the time:
At the very short end [of latency], opportunities have compressed; at the very long end, premia are crowded. Naturally, capital and talent migrate towards the middle. So you see prop firms with execution DNA stretching into multi-day signals, while classical stat-arb firms accelerate their cycles.
Well, the overlap has grown even further since then, with a flurry of intriguing moves in both directions in recent months. Here are some of the ones we’ve spotted.
In March, Bloomberg reported that Squarepoint, a big quant hedge fund that once operated inside Lehman Brothers, had started a separate prop trading firm. To run it, Squarepoint has hired Kirill Gelman, a former senior executive and trader at GTS, Barclays and Lehman. The firm’s Finra filing indicates that Eric Levine, another Lehman alum, is head of trading.
Last month, Business Insider reported that Yiming Zhang, a top quant researcher at Jump Trading, had quit to launch a new venture at Millennium Management (once his extravagantly long non-compete clause expires). Zhang is a big name in prop trading — he’s been a central person at Jump for 17 years — so this caused a stir in the industry.
These are the higher-profile moves, but they aren’t isolated examples. Alphaville hears that both Millennium and Qube Research & Technologies are hiring particularly aggressively from the prop trading/HFT industry lately, while the prop firms are almost all bulking up in slower “mid-frequency” signals. Everyone wants to get their hands on some of those Jane Street dollars, after all.
We’ve now learned of another sign of the great quant convergence. The Voleon Group — an AI-focused hedge fund set up by two former senior DE Shaw executives — has quietly set up something called Voleon Securities, which says it “provides liquidity in securities markets”. Voleon declined to comment.
Regulatory filings indicate that Simon Spenser, a British physicist and former DE Shaw and Two Sigma Securities executive, is the “designated principal” at Voleon Securities, but Alphaville gathers that Adam Kravetz is currently in charge of the new trading firm, and some low-level LinkedIn stalking gives it credence.
Kravetz was previously COO at Seven Eight Capital, a quant fund spun out of Schonfeld Strategies until it was shut down in 2024, and before that worked at Citadel Securities, BofA and Tower Research.
Voleon currently has four jobs listed on its website for Voleon Securities, stressing that it is a “new business” and hires will have a chance to “join the initial build-out of a fully modern securities business rooted in the frontier of AI/ML and statistics”:
We apply state of the art AI/ML techniques to construct our market making strategies. For nearly two decades, our affiliate Voleon Capital Management has led the hedge fund industry and worked at the frontier of applying AI/ML to investment management, becoming a multibillion-dollar asset manager. Voleon Securities builds on Voleon’s deep real-world experience applying ML to financial markets. Voleon Securities is looking for creative, entrepreneurial researchers who enjoy grappling with very difficult problems.
FWIW, in a sign of how white-hot the competition for talent is, Voleon is offering finders fees of up to $15k for anyone who refers a candidate that lands a job there. Please consider this ‘stack a referral, and let us know how it goes. 🫡
A week on Alphaville
○ Here’s this week’s quiz. Guess what these three squiggly lines show for a chance to win a fetch I❤️charts T-shirt.
○ It’s probably a bad idea to put Alphaville in charge of anything more important than the occasional pub quiz. But if Keir Starmer (at pixel time) appointed us head of the UK’s Debt Management office then here’s what we’d do.
○ More takeaways from an S-1 for the ages. Craig Coben took a look at SpaceX’s IPO prospectus and was suitably astonished.
○ Alphaville would also like to benchmark our pay against Tim Cook and Satya Nadella, just like Ford.
○ Hedge funds are AImaxxing. The title is mainly designed to annoy the FT’s Katie Martin, but it is noteworthy how many hedgies are going all in on AI (and fleeing software).
○ Trad bookies ♥ prediction markets. They might not like the competition, but they do like making markets on them.
○ The phrase “up to” can do a lot of work, but you should always take the big numbers they imply with a fistful of salt, as Dan Davies points out.
○ Et tu too, FTSE Russell? Come for the indexing gamesmanship, stay to learn that SpaceX might initially be classified as a “value stock”.
○ It’s always a good day when OG Alphavillain Joseph Cotterill returns to our pink pixels. This week he tackled Senegal’s increasingly domestic debt crisis for us.
○ What made the DoJ go after this particular prediction market insider? Impossible to say, really…
Best of Further Reading
○ MIT Press reader published a fun piece on the the strange melancholy of slaying monsters.
○ A Stock Certificate From 1941 Taught Me More About AI Than Anyone from OpenAI. A fun Substack from Apers Insight’s Francis Huang.
○ SpaceX continues to dominate the conversation. Polemic Paine pointed out that they don’t ring a bell at the top. They IPO.
○ Is AI profitable yet? No.
Chart blast
Rob Armstrong’s Unhedged newsletter asks whether semiconductor stocks are in a supercycle or a superbubble.
It’s getting hot in herrrre.
Original Sin — European version.
I seen a peanut stand, heard a rubber band. I seen a needle that winked its eye.
But I seen ‘bout everything, when I see SMIT trade at a premium to NAV. Naturally, SpaceX is to blame.
The migrate-to-the-middle logic is right, with one twist. the middle is appealing only because it's the band the convergence hasn't competed away yet. Everyone arriving is what closes it. Mid-frequency stays uncrowded right up until it's the consensus, and this flurry is the sound of it becoming the consensus.
Crowding is the obvious cost. The quieter one is correlation. When prop shops and stat-arb funds poach the same researchers and run the same ML over the same horizon, they stop being different strategies and turn into one strategy with several letterheads.
Which is fine until something forces the middle to delever at once. A room full of near-identical positions behaves beautifully until one of them blinks. August 2007 was that movie, and the cast is bigger now.