Nick Latinovic

Building the future of asset management @voltaire_fi ex global macro at Drobny Global Advisors, Point72, CWOC

[How To Prompt @HowToPrompt__]

Yann LeCun was right the entire time. And generative AI might be a dead end.

For the last three years, the entire industry has been obsessed with building bigger LLMs. Trillions of parameters. Billions in compute.

The theory was simple: if you make the model big enough, it will eventually understand how the world works.

Yann LeCun said that was stupid.

He argued that generative AI is fundamentally inefficient.

When an AI predicts the next word, or generates the next pixel, it wastes massive amounts of compute on surface-level details.

It memorizes patterns instead of learning the actual physics of reality.

He proposed a different path: JEPA (Joint-Embedding Predictive Architecture).

Instead of forcing the AI to paint the world pixel by pixel, JEPA forces it to predict abstract concepts. It predicts what happens next in a compressed "thought space."

But for years, JEPA had a fatal flaw.

It suffered from "representation collapse."

Because the AI was allowed to simplify reality, it would cheat. It would simplify everything so much that a dog, a car, and a human all looked identical.

It learned nothing.

To fix it, engineers had to use insanely complex hacks, frozen encoders, and massive compute overheads.

Until today.

Researchers just dropped a paper called "LeWorldModel" (LeWM).

They completely solved the collapse problem.

They replaced the complex engineering hacks with a single, elegant mathematical regularizer.

It forces the AI's internal "thoughts" into a perfect Gaussian distribution.

The AI can no longer cheat. It is forced to understand the physical structure of reality to make its predictions.

The results completely rewrite the economics of AI.

LeWM didn't need a massive, centralized supercomputer.

It has just 15 million parameters.

It trains on a single, standard GPU in a few hours.

Yet it plans 48x faster than massive foundation world models. It intrinsically understands physics. It instantly detects impossible events.

We spent billions trying to force massive server farms to memorize the internet.

Now, a tiny model running locally on a single graphics card is actually learning how the real world works.

[Nick Latinovic @nicklatinovic]

@BowGuppy29365 @systematicls Education, science and tech => innovation => higher productivity growth. Create a scatter plot of PISA rankings and innovation, you get a very high fit. In the US, we benefit from market size and network effects, but education quality is declining, cost rising. See 6 killer apps

[Nick Latinovic @nicklatinovic]

@systematicls Most efficient way of addressing eigenvector sign flipping and ordering issues when using rolling windows?

One approach: https://t.co/c1SjMuE6Uy

[Nick Latinovic @nicklatinovic]

@BoringBiz_ Low uncorrelated returns + capital efficient leverage (low financing cost) = higher absolute returns. Most hedge funds can access cheap leverage and target higher absolute returns if they want to target higher vol. Question you should be asking is why do HFs target lower vol?

[Nick Latinovic @nicklatinovic]

@GordonGekko 364d from Oct 2025 ATH to ATL would mean Oct 2026 ATL yet the next picture has 2026 ATH at $250k? Seems as credible as the stocks to flow ratio/4 year cycle…

[Nick Latinovic @nicklatinovic]

@macrocephalopod @choffstein And how to think about that correlation risk? Though not a portable alpha, equity/bond risk parity was exposed to a flip to higher correlation (and higher risk) via a rise in inflation risk in 2022. What’s the driver of the 60/40 + trend correlation, and is it stable around 0?

[Plume @plumenetwork]

The future of finance is tokenized.

Join leaders from @voltaire_fi, @SentoraHQ, our CEO @chriseyin, and host @jackratko_ as they discuss tokenization and institutional needs in this week’s Ascend 𝕏 Space spotlighting Voltaire.

Live Thursday - 9AM EST ⏰

[Nick Latinovic @nicklatinovic]

@benjamincowen @CuttyRanks There are a few reasons why BTC dominance may continue to trend lower: 1) the Fed cutting cycle is good for small caps/alts, 2) BTC is reaching more mature asset status (vol has declined with the advent of derivatives products), 3) Gold is outshining BTC as a store of value.

[Nick Latinovic @nicklatinovic]

@Noahpinion Is there any credible evidence to support that interventions don’t lead to unintended consequences (negative second order effects) when trying to solve market failures? It’s especially difficult to resolve when those effects play out “over the long run when we are all dead”

[Nick Latinovic @nicklatinovic]

@APompliano It’s underperformed gold over the last 8 years, in risk-adjusted terms. Anybody can access gold futures these days, lever up a less volatile asset to make more return for the same risk as Bitcoin. You don’t seem to understand this basic financial math. Insane

[Nick Latinovic @nicklatinovic]

@APompliano Talking your book 101:

1- Bitcoin is down, “Fed should cut interest rates”

2- Bitcoin is up, “Fed is printing too much money”

[Nick Latinovic @nicklatinovic]

@nntaleb @nfergus makes good comparison to Nixon. Populist of the right.

[Nick Latinovic @nicklatinovic]

@markoinny Why not real yields (TIPS)? Long nominal rates with inflation kicker. We know historically large debt overhangs mean financial repression. Given current real yield levels, long real yields and gold seem like good bets. Hat tip @BobEUnlimited

[Nick Latinovic @nicklatinovic]

@bryce_m88 @PrometheusCHT @RaoulGMI Tariffs are typically growth negative/inflation positive. They raise production costs and import prices, reduce output and competition. Share many of the characteristics of supply shocks.

[Nick Latinovic @nicklatinovic]

@chamath Flawed assumption: large debtor/deficit countries like the US hold all the cards. They also rely on foreign capital—and when foreign asset holders dump, it hits equities, bonds, and the USD hard as we saw yesterday. Nobody can negotiate with the bond market.

[Nick Latinovic @nicklatinovic]

@bennpeifert As much predictive info as a P/E ratio wrt 10 year forward returns

[Nick Latinovic @nicklatinovic]

@bennpeifert Benn, don't overestimate our "collective knowledge". "We" now think that a recession is good so we can lower interest rates and debt service costs. "We" think trade is zero-sum. And "we" think protectionism and tariffs will work this time. Welcome to the age of stupidity.

[Nick Latinovic @nicklatinovic]

Sad day when a macro legend has to clarify that catalyzing the largest trade shock in history isn’t his idea of good policy.

[Gregory Zuckerman @GZuckerman]

Stan Druckenmiller almost never tweets, fewer than 5 times in his life.
Today, he's making clear his opposition to the administration's tariff plan.

[Nick Latinovic @nicklatinovic]

@bryce_m88 @PrometheusCHT @RaoulGMI Fair point, esp short term. Inflation expectations falling sharply, real yields rising, and gold selling off over last couple days are consistent with that also. I think medium term there’s a supply shock effect but that may be dominated by the short term demand shock.