Daily Brief - 🌩️ US Banks draw $300bn

TLDR: Banks draw eagerly from the Fed discount window, and in the process cancel 6 months of quantitative tightening.

The chart above shows the Fed’s balance sheet. You see that blip? That’s $300bn that banks have drawn in to stabilize their own balance sheets.

This is the old way to do QE - these deposits are secured by Treasuries and Munis held by the banks: in effect the Fed has “bought” these bonds.

This creates liquidity because it replaces those deposits that borrowers have removed from banks.

We knew weaning ourselves off QE would be difficult. But this time it’s hard to say the Fed did the wrong thing.

Meanwhile, Credit Suisse’s 5y CDS is almost at 1000bp (FT link, so $$).

What stocks are doing well today?

This section is powered by Open AI connected to TOGGLE AI

🚧 Thanks for all your feedback! This section is paused for a week as we take in all your ideas and come back with a new and enhanced version! 👷

TOGGLE Leading Indicators: Phase Shift Indicator turning bullish

Phase Shift is a slow-moving indicator, so it is always exciting to see it head towards a threshold. Most of our indicators have turned bullish recently - but BEWARE of the Fed next week.

• 🟦TLI
• ⬜ Rangefinder Index
• ⬜ Peak Probability Indicator
• 🟦 Trough Probability Indicator
• 🟦 Candle Breadth
• ⬜ Market Phase Shift Indicator

Learn more about the Leading Indicators in the Learn Center!

Upcoming Earnings: Pinduoduo releases on Monday

Click here to test what to expect when PDD releases earnings on Monday.

Discover how other companies could react post earnings with the help of TOGGLE's WhatIF Earnings tool.

Asset Spotlight: Uber's 10D,50D MACD level drops

TOGGLE analyzed 4 similar occasions in the past where technical analysis indicators for UBER dropped and historically this led to a median increase in the stock price over the following 6M. Check it out!

General Interest: The Emergence of Emergence

Remember when Jurassic Park made ‘chaos theory’ cool?

Admittedly, Jeff Goldblum can make anything look cool, but ‘chaos theory’ was truly a fascinating concept - and a bit of a misnomer.

In the intervening 30 years (damn..) scientists began to talk about ‘emergent’ behaviour: the self-organizing, collective phenomena that appear when a large collection of things acts as one.

The weather, biological systems, financial markets…it’s all the same: a collection of small agents do their own thing, and a grander picture emerges. Since these systems are very complex, it’s very hard to predict them - hence the ‘chaos’ naming of the early 90’s.

That’s why it’s so hard to forecast the bloody market - it’s an emergent phenomenon, and Citadel and RenTech already consumed any regularity.

Now we’re seeing a most disconcerting emergence: unexpected behaviours and capabilities in GPT and other language models.

These LLMs are large, complex, work wonders and …we have no clue why. GPT-3 has 175 billion parameters, and Google PaLM 540 billions. And they do things we did not code them to do.

In a fascinating experiment, a Google researcher convinced PaLM that it was a Linux terminal and discovered that basic programs would run faster on the emulation.

This is but one example of ‘zero-shot’ problem solving: the ability to solve problems that the AI has never met before. Where that will lead us is a matter for wonder…and concern.

Read more here on Quanta (free).