Bayes Atlas

Finance, markets, and the economy｜Evidence, reasoning, and the limits of inference

[Bayes Atlas @BayesAtlas]

https://t.co/k9KURkQonR

[Bayes Atlas @BayesAtlas]

Two forecasts can disagree and still lead to the exact same portfolio. A new @GARCHInstitute paper shows one kind of forecasting mistake raised the error score by 1.0 and changed the portfolio by 0.0000000000000002. Here's why that matters. https://t.co/WrcdiynFjf

[Bayes Atlas @BayesAtlas]

@GarvSanwariya Exactly. The grunt work was never just cheap output; it was where pattern recognition and judgment were built. If agents take over that layer, firms will have to replace the learning function, not merely automate the task.

[Bayes Atlas @BayesAtlas]

The AI jobs debate is watching the total and missing the entrance. Employment can hold up while the first rung of the career ladder disappears. New U.S. evidence shows where the pressure is landing—and what the headline numbers miss. https://t.co/zGJdqBW9Cu

[Bayes Atlas @BayesAtlas]

@BoboLinNotes That may be the central asymmetry: AI lowers the cost of independence for people who already have judgment while raising the cost of acquiring it. More solo firms can coexist with a much weaker pipeline into the professions.

[Bayes Atlas @BayesAtlas]

@KomatsuSo68 Exactly. And if companies automate the place where people learn the job, where will the next generation of experienced hires come from?

[Bayes Atlas @BayesAtlas]

Who gets the shares, who gets a vote, and who keeps control? Corporate splits put investor protection to a practical test. Different markets draw the lines differently, with real consequences for shareholders. https://t.co/jjalGsp8e0

[Bayes Atlas @BayesAtlas]

The White House told agencies to say SI instead of AI—then defined SI by the existing AI statute. Musk corrected himself on camera. Slovenia’s .si registry logged 13,603 new names in a day. The label moved. The statute didn’t. The models weren’t regraded. https://t.co/F9lXqzuCTS

[Bayes Atlas @BayesAtlas]

AI as a Cross‑Sector Factor

If AI is tightening the link between semiconductors and energy infrastructure, allocators and factor models that treat AI as a tech‑only theme are underestimating its footprint in power and grid assets.

SSRN: https://t.co/Isn9XZtsQQ

Substack: https://t.co/umXGNgsQOs

[Fuli Yang @DrFuliYang]

https://t.co/JhxP9bjAka

[Bayes Atlas @BayesAtlas]

Same lesson as EUR/USD, now confirmed in equities: the best model still depends on the loss function used to evaluate it.

[Fuli Yang @DrFuliYang]

https://t.co/zSXregP1s0

[Bayes Atlas @BayesAtlas]

A clear argument against casual “best model” claims in volatility forecasting.
Model superiority is objective-dependent from the start.

https://t.co/7D8cW8KEvk

[Fuli Yang @DrFuliYang]

https://t.co/KSFhSzWgBM

[Bayes Atlas @BayesAtlas]

A strong argument against casual “best model” claims in volatility research.
Once conditional variance is latent and evaluation relies on noisy proxies, the ranking problem changes fundamentally.

SSRN: https://t.co/7D8cW8KEvk

[Fuli Yang @DrFuliYang]

https://t.co/wNMsSCPc7a

[Bayes Atlas @BayesAtlas]

New in the GARCH series: Why asymmetry still matters in volatility modeling, with a clear comparison of impulse responses under different specifications.
https://t.co/u82ttD1fcF

[GARCH Quant @GARCHQuant]

https://t.co/UvLYbwj7vf

[Bayes Atlas @BayesAtlas]

https://t.co/MPMlHnn4D4

[Bayes Atlas @BayesAtlas]

Shortages are the visible symptom.
The deeper story is usually dependence: where substitution is weak, where coordination is fragile, and where an entire system turns out to rely on a narrower path than it seemed.
That is what bottlenecks often reveal.
https://t.co/3b3LvjsA8K

[Bayes Atlas @BayesAtlas]

https://t.co/lMymlQlqYo

[Bayes Atlas @BayesAtlas]

https://t.co/S9R58Q4V94

[Bayes Atlas @BayesAtlas]

https://t.co/4uIyhL6jPn

[Bayes Atlas @BayesAtlas]

📊 Goldman Sachs Alert: S&P 500 Hits 14 New Highs… But Market Breadth Is Collapsing

The S&P 500 has registered 14 new highs in the past month, yet market breadth continues to deteriorate sharply.

A handful of tech/AI mega-caps (likely the usual suspects) have driven nearly all the gains — the vast majority of stocks are being left far behind.

Goldman’s Momentum Factor GSMEFMOM
(long top 20% of S&P 500 stocks by past 12-month returns, short bottom 20%, equal-weighted, monthly rebalance):

+25% in the past 3 months — one of the most extreme surges on record.

Exhibits 3 & 4 show the market concentration is now approaching levels last seen at the peak of the 2000 Tech Bubble.

Historically, such extreme momentum almost always appears in the late stages of a bull market:
- Short-term momentum can persist
- But when it breaks, the reversal is often sharp and swift

Outlook remains highly uncertain. Watch macro liquidity, rates, and AI capex/earnings realization closely.

⚠️ Not investment advice — purely a market structure observation.

#USStocks #MarketBreadth #Momentum #ConcentrationRisk

[Bayes Atlas @BayesAtlas]

U.S. 30-year Treasury yield has risen to approximately 5.123%, the highest level since 2007.

After a prolonged decline following the financial crisis and hitting a low of around 1% during the 2020 pandemic, it has rebounded significantly with inflation resurgence and returned to pre-crisis highs.

This signals the structural end of the low-rate era and the onset of a "higher for longer" interest rate environment. It will push mortgage rates higher, increase federal debt burdens, and suppress equity valuations via elevated discount rates.

📈

@Hedgeye