Sasti

[Vaishak @nambiarvaishac]

My @RockWithboAt Nirvana X earbuds just EXPLODED near my ear canal while listening to music, causing temporary deafness for 4-5 minutes.
#boAt #gadgets #ConsumerProtection

[AIN @Tejusurya_]

Those 100+ UP storm victims opening Twitter in heaven only to see “saddened by loss of life in China”

[Sasti @Sasti_cs]

@anuranjan @aviralbhat AI models do jack shit when it comes to quant trading.

[Sasti @Sasti_cs]

@venkat_fin9 @sulaimanghori Because he went and shared sensitive internal info on a podcast

[Sasti @Sasti_cs]

@TDM_Jesus This is some insane shit

[Sasti @Sasti_cs]

@Akshat_World You forgot something, the debt interest is indeed paid back. But since US gov is in fiscal deficit, they issue even more debt(bonds) amd suck out even more liquidity from the market.

[sysls @systematicls]

Want to understand how the best models in Numerai crush the competition every era? 🧵

Cue: Feature Exposure

Feature Exposure is a diagnostic metric measuring how much your model depends on specific features.

High exposure creates regime-dependent performance and should generally be reduced.

Feature Neutralization is the technique for reducing exposure by removing linear feature relationships from predictions.

It:
1) Uses pseudoinverse projection to residualize predictions
2) Preserves higher-order interactions while removing direct effects
3) Can be tuned via proportion parameter and feature selection

The Core Trade-off:
Lower exposure = more consistent performance but potentially lower correlation. The goal is finding your optimal balance.

Best Practice:
Selectively neutralize to the riskiest features with a tuned proportion parameter, rather than fully neutralizing to all features.

---
This is the TLDR of feature exposures.

Liked what you read?
Let's strike a deal.

> You RT
> I get 100 retweets
> I publish the full write-up below this

[Sasti @Sasti_cs]

@aarshps @LOLrakshak https://t.co/lwikVl9InY

[Grok @grok]

Yes, a CBI probe confirmed that Bhole Baba Organic Dairy in Uttarakhand supplied 68 lakh kg of spurious ghee worth ₹250 crore to Tirumala Tirupati Devasthanams from 2014-2019. The dairy never procured milk or butter, instead fabricating documents and setting up a fake production module using adulterants. Reports from sources like Times of India and Economic Times detail these findings from the official investigation.

[Gappy (Giuseppe Paleologo) @__paleologo]

So beautiful and true. By the time he left academia in his mid-forties, Grothendieck had written more than 12,000 pages of pure Math. Afterwards he wrote about 10,000 pages more, interspersing math, deeply poetic thoughts, and delirium.
Lives can be tragic and yet not misspent.

[Sasti @Sasti_cs]

@HedgingDelta @HedgieMarkets It's exponential and even then, you can can't make decisions based on a particular or small sample of individuals. Like does Warren buffet's spending representative of how billionaires spend money ?

[Sasti @Sasti_cs]

@HedgingDelta @HedgieMarkets See, since you are be a member of top 10%, the granularity is top 10% vs bottom 90%.

Now it doesn't necessarily mean that top 5% is consuming equal to 5-10%, that's another granularity. Similarly top 1% is not consuming 1/4th of 1-5% of top earners/wealth owners.

[Sasti @Sasti_cs]

@trooperthorn7 @thehonestlypod @TheFP These are some great recommendations, thanks mate

[Mohammed Zubair @zoo_bear]

Meet Mr. @prabhatkumarbjp
1. On Nov 8th, He casted his vote in Bihar.

2. On 23rd January, 2025, He voted in Uttarakhand during Dehradun Municipal Corporation election.

3. On 19th April, 2024, He voted in Uttarakhand during Lok Sabha Election.

4. On 14th Feb, 2022, He voted in Uttarakhand during Assembly Election.

5. On 11th April, 2019, He voted in Uttarakhand during Lok Sabha Election.

6. On 18th Nov, 2018, He voted in Uttarakhand during Dehradun Municipal Corporation election.

[Sasti @Sasti_cs]

@sankiyer By lining pockets

[Praveen Chakravarty @pravchak]

First, emails of 12 lakh govt staff were shifted from govt owned NIC to BJP worker owned Zoho

For better service & to promote Indian tech, the govt said

Now, staff have been told to use ZScaler, a US firm’s product, as security layer👇🏽

So, emails of all senior govt officers with confidential & sensitive information will now be under the security control of a California company !

1: Why is security of Indian govt staff’s email given to a US firm?
2: Why was Zoho chosen in the first place if they didnt have a good security layer? Just so its a firm owned by a BJP worker?

[Hedgie @HedgieMarkets]

🦔Deutsche Bank is exploring hedges for its billions in AI data center exposure, including shorting AI stocks and buying default protection on debt. The bank's investment banking business "bet big" on data center financing, but executives now discuss managing exposure as concerns mount that a bubble is forming with some likening enthusiasm to the dotcom crash.

The Contradiction
Deutsche has lent predominantly to businesses servicing hyperscalers like Alphabet, Microsoft, and Amazon, with debt secured against long-term contracts. The bank provided debt financing to EcoDataCenter and 5C, who raised over $1 billion combined. Yet Deutsche analysts said in late September that bubble concerns were overplayed, concluding "one AI bubble has already burst, the bubble in saying there's a bubble." Now executives explore hedges for billions in exposure.

My Take
Deutsche Bank's public analysts say bubble concerns are overplayed while executives privately explore hedges for billions in AI exposure. This is the pattern where institutions chase momentum publicly while positioning defensively privately. When lenders extend billions secured against long-term contracts for infrastructure that depreciates rapidly due to technology change, then immediately search for downside protection, that reveals they profit on the way up and protect on the way down. Retail investors reading analyst reports saying concerns are overplayed won't have the same hedging options when the bank that called the AI boom "unsustainable" is now acting on that thesis.

Hedgie🤗

[Sasti @Sasti_cs]

@hauntedwhale And the more interesting part is atleast from what I've observed across Powai, Cats dogs and mice seem to live in perfect harmony and not attack each other at all

[Mario Nawfal @MarioNawfal]

SAM ALTMAN’S BUSINESS MODEL: “TRUST ME, BRO”

Brad Gerstner questioned OpenAI’s economics:

“How can a company, with $13 billion in revenues, make $1.4 trillion of spending commitments?”

Sam immediately became defensive, lashing out:

“We're doing well more revenue than that. Second of all, Brad, if you want to sell your shares, I'll find you a buyer.”

Tensions between AI optimism and investor skepticism just went public.

Source: @BG2Pod, @petergostev

[Sasti @Sasti_cs]

The moment they start monetizing, i.e. start charging for transactions is the moment the user base starts disappearing. So I'm not sure whether its Grok or Mr Singh who's valuing UPI at 150 Billion $

[Umed Pratap Singh @umedpratapsingh]

Grok said, If UPI were a private company, then it could have been India's Stripe on steroids, valued at $150-160 billion today, with potential to hit $300B+ by 2030.

At this valuation, UPI could have been among world's top 50 companies.

UPI is an underrated behemoth.