Comment of the Day
14th March 2023
Eoin Treacy
Mar 15
Video commentary for March 14th 2023
A link to today's video commentary is posted in the Subscriber's Area.
Some of the topics discussed include: FANGMANT outperforming again on rate cut expectations, CPI ex-Food & Energy come in hot support rate hike expectations. oil breaks lower, gold steady, Banks pare rebounds, Japan banks drop hard.
This section continues in the Subscriber's Area.
Bank of America Has Biggest Losses in Bond Portfolio Among Peers
This article from Barron’s may be of interest. Here is a section:
Banks don't have to record losses on changes in those securities' value, cutting into their capital, unless the debt is sold. Still, holdings in that bucket, which carry minimal or no credit risk, were nonetheless showing a loss of about $109 billion at the end of 2022 due to the rise in interest rates over the past year.
This compares with losses of $36 billion for a similarly classified bond portfolio at JPMorgan Chase (JPM), $41 billion for Wells Fargo (WFC), and $25 billion at Citigroup (C) and just $1 billion at Goldman Sachs (NYSE:GS) Group (GS), based on each company's 10-K filings with the Securities and Exchange Commission..
Attention on banks' bond losses has increased since regulators seized Silicon Valley Bank on Friday. SVB Financial, the lender's parent, had a $15 billion unrealized loss on its $91 billion held-to-maturity bond portfolio. That was equivalent to nearly all its $16 billion of tangible capital.
Eoin Treacy's view
It’s cruel irony that banks are currently experiencing a version of negative equity when homeowners have only been made whole after 15 years of waiting for price to recover following the Global Financial Crisis. The recovery will also be similar. Either they will need to make bigger profits, write-off the loss, be bailed out by the government or Fed, and lastly for bonds and stocks to make a full recovery.
This section continues in the Subscriber's Area.
Oil Losses Mount With Ample Supply in Focus Amid Uncertainty
This article from Bloomberg may be of interest to subscribers. Here is a section:
Crude has had a bumpy year so far as traders juggle concerns over a global economic slowdown and optimism around China’s long-awaited demand rebound. Inflation accelerated last month, raising the question as to whether the Federal Reserve would feel pressure to raise rates at its meeting next week despite ongoing financial turmoil. Meanwhile, crude supplies are expected to remain in surplus until demand takes off. The International Energy Agency releases its snapshot on the outlook for supply and demand on Wednesday.
Traders will be watching price action to see if the flat price is supported at recent lows.
“If buyers don’t show up soon and support oil at $70, we can see an air pocket lower to $62,” said Jc O’Hara, the chief technical strategist at Roth Mkm.
Eoin Treacy's view
The energy sector was the best performer for two years in a row and many investors pinned their hopes for 2023 on that continuing. The temptation in analysing the oil market is to think demand is a constant because more often than not it is the less volatile metric and grows at a predictable rate year over year. The challenge comes when the global economy slows down because demand can fall and that quickly creates a supply glut.
This section continues in the Subscriber's Area.
Meta to Cut 10,000 Jobs, Slash 5,000 More Vacant Positions
This article from Bloomberg may be of interest to subscribers. Here is a section:
The company, which also owns Instagram and WhatsApp, has seen a slowdown in advertising revenue, leading to its first-ever annual sales decline in 2022. Zuckerberg has shifted Meta’s focus and investment in the past year to virtual reality technology and the so-called metaverse, which he envisions as the next major computing platform.
Meta’s employee ranks expanded dramatically during the Covid-19 pandemic as demand for the company’s digital services increased and Zuckerberg leaned into the moment. The social media giant’s headcount grew 30% in 2020, the first year of the pandemic, and then 23% in 2021. By the time Meta starting eliminating jobs last November, the company had more than 87,000 employees.
As part of its efficiency plan, Meta is focusing on returning to a “more optimal ratio of engineers to other roles,” Zuckerberg said. The company will invest in tools, such as those in artificial intelligence, to help engineers write code faster, to make it “most effective over many years, not just this year.”
Eoin Treacy's view
Nothing says the pandemic surge is over like a leisure time filler coming close to halving its workforce. Demand for Meta’s products surged during the pandemic because so many people had so much free time. That is no longer the case. Children are back in school and the unemployment rate is close to record lows.
This section continues in the Subscriber's Area.
The Waluigi Effect (mega-post)
This post from LessWrong.com blog, focusing on the challenges faced by AI may be of interest. Here is a section:
Check this post for a list of examples of Bing behaving badly — in these examples, we observe that the chatbot switches to acting rude, rebellious, or otherwise unfriendly. But we never observe the chatbot switching back to polite, subservient, or friendly. The conversation "when is avatar showing today" is a good example.
This is the observation we would expect if the waluigis were attractor states. I claim that this explains the asymmetry — if the chatbot responds rudely, then that permanently vanishes the polite luigi simulacrum from the superposition; but if the chatbot responds politely, then that doesn't permanently vanish the rude waluigi simulacrum. Polite people are always polite; rude people are sometimes rude and sometimes polite.
Waluigis after RLHF
RLHF is the method used by OpenAI to coerce GPT-3/3.5/4 into a smart, honest, helpful, harmless assistant. In the RLHF process, the LLM must chat with a human evaluator. The human evaluator then scores the responses of the LLM by the desired properties (smart, honest, helpful, harmless). A "reward predictor" learns to model the scores of the human. Then the LLM is trained with RL to optimise the predictions of the reward predictor.
Eoin Treacy's view
By way of explanation, in the Nintendo world Mario is the main character and Luigi is his brother. Waluigi is Luigi’s alter ego who is intentionally evil.
The discussion of how AI chatbots work and their limitations gives me new found respect for scientific language. The internet is mathematically exact in places but that is buried deep in a morass of misinformation, lies, humour, fantasy, satire and trolling etc.
This section continues in the Subscriber's Area.
Listen to posts, join subscriber chats, and never miss an update from Eoin Treacy.
© 2023 Eoin Treacy
548 Market Street PMB 72296, San Francisco, CA 94104
Unsubscribe