moo 🏳️‍⚧️

[moo 🏳️‍⚧️ @cryptotranny]

Seiche now takes crypto. You can pay for a subscription or simply support the desk in BTC, ETH, SOL or USDT on Tron.

All addresses, with copy buttons and network notes: https://t.co/dDiCrFUHMY

[moo 🏳️‍⚧️ @cryptotranny]

VIX at 16. Credit spreads at the 6th percentile of their own history. Every price screen says calm.

The plumbing disagrees. The $2.5 trillion RRP buffer is down to $3 billion, reserves are drifting toward the floor, and the tail of the funding stress distribution measures fat: shape 0.64, no ceiling. The worst spike on record is not the worst this basin can produce.

Plumbing sits at its 76th percentile. Markets at their 27th. One of them is wrong about the next six months.

Today's full reading is free, misses published next to hits: https://t.co/p9zMMScenA

[moo 🏳️‍⚧️ @cryptotranny]

This morning I said the pipes and the price screens disagree. Here is the physics under the disagreement.

I fit extreme value theory to every funding-stress shock since 2018. Tail shape came back at 0.64. Heavy-tailed, no ceiling. The worst spike on record, 305bp, is not the worst the plumbing can make.

VIX at 16 is pricing a thin tail. The pipes have a fat one.

Everything public, every method shown, twice a day: https://t.co/oZ4IBqo1Uu

[moo 🏳️‍⚧️ @cryptotranny]

This morning I said the pipes and the price screens disagree. Here is the physics under the disagreement.

I fit extreme value theory to every funding-stress shock since 2018. Tail shape came back at 0.64. Heavy-tailed, no ceiling. The worst spike on record, 305bp, is not the worst the plumbing can make.

VIX at 16 is pricing a thin tail. The pipes have a fat one.

Everything public, every method shown, twice a day: https://t.co/oZ4IBqo1Uu

[moo 🏳️‍⚧️ @cryptotranny]

The board misses too. It slept through the speed of March 2020, and I publish that miss next to the hits. Free public data, updated twice a day, every method shown: https://t.co/qPYX2bjTrl

[moo 🏳️‍⚧️ @cryptotranny]

September 2019. March 2020. SVB. Each one showed up in funding plumbing weeks before it showed up in price. Plumbing is the seating chart. Price is the fire alarm.

[moo 🏳️‍⚧️ @cryptotranny]

Bank reserves: $2.97 trillion, drifting toward the ample floor. Overnight RRP: $3.3 billion. Treasury cash: $783 billion. The balance sheet trend alarm has been on since May 13.

[moo 🏳️‍⚧️ @cryptotranny]

I run a small terminal that reads only free public data. Fed H.4.1, NY Fed operations, OFR repo, Treasury cash. No Bloomberg. This morning's composite: 46 out of 100. Regime: STRAIN.

[moo 🏳️‍⚧️ @cryptotranny]

Plumbing indicators sit at the 76th percentile of their own history. Market indicators sit at the 27th. High yield spreads are at the 6th. One of these numbers is wrong about the next six months.

[moo 🏳️‍⚧️ @cryptotranny]

The Fed's shock absorber held $2.5 trillion in 2022. This morning it holds $3 billion. VIX is at 16. Credit spreads are near their tightest on record. The pipes and the price screens disagree.

[moo 🏳️‍⚧️ @cryptotranny]

And if July 31 passes quietly, we will post exactly what the model got wrong and why.

That is the whole point. A stress signal you cannot audit is just vibes.

[moo 🏳️‍⚧️ @cryptotranny]

The Fed had a $2.5 trillion shock absorber in 2022.

Today it holds $2 billion. Nobody repriced anything.

With the buffer gone, every Treasury settlement drains bank reserves directly. Reserves already sit below where our model says scarcity starts.

Mark July 31.

[moo 🏳️‍⚧️ @cryptotranny]

Every funding squeeze of the past year showed up in the plumbing before it showed up in prices.

Sep 15: repo printed 18bp over fed funds on a tax date
Oct 31: banks took $50B from the Fed's repo window
Dec 31: a record $74.6B

Screens looked calm every single time.

[moo 🏳️‍⚧️ @cryptotranny]

@GreatFireChina @CDT @OpenObservatory @deepseek_ai @Alibaba_Qwen #DeepSeek #China

[moo 🏳️‍⚧️ @cryptotranny]

China's own AI models censor on command. I built a free, open-source index that measures it live, with receipts.

This week DeepSeek and Qwen refused or rewrote 30% of sensitive prompts. Neutral Western models refused none.

[moo 🏳️‍⚧️ @cryptotranny]

A hobby project, built in the open as a public good. If you work on China, censorship, or AI safety and it is useful, take it, cite it, break it. Feedback very welcome.