JMo sent me this photo of him prematurely celebrating the CPI print at 8:29am Wednesday morning.
Obviously I’m kidding.
Jmo isn’t up at 8:29am.
GenAI Experiment
I’m a huge believer in genAI. Each week I will drop the newsletter into an AI model and ask it to succinctly summarize the key takeaways. You can click on that to read it as an alternative to the full newsletter, with some of my thoughts on how well it did. I won’t touch the summary but will call out anything I believe it misinterpreted.
We’ll start with Claude, the AI model from Anthropic. Anthropic was started in 2021 by some OpenAI (ChatGPT) employees that were concerned about the direction it was taking. Until recently, ChatGPT crushed Claude for my typical uses. Then Claude released its newest model a few months ago and effectively closed the gap. If I am doing a blind taste test, Claude is starting to pull ahead…for now.
Because ChatGPT was so much better, the only reason I even stayed on top of Claude developments is because all of our technology is housed on an an AWS cloud platform. Amazon is a big investor in Anthropic, so even if I had personal preferences for ChatGPT, my businesses ran on AWS and our business use cases for genAI would be driven by Claude. Alternatively, Microsoft is a big investor in OpenAI. When you hear techies describe their companies as “AWS shops” or “Azure shops”, they are describing the cloud platform they run on. Azure is Microsoft, so if you are an “Azure shop” this likely means your business use cases will be dictated by ChatGPT developments.
These are my layperson thoughts, not intended to be a highly technical review. About a dozen people in each of my companies uses both free and paid versions of each as we continue to test options and stay in the loop. I have ChatGPT, Claude, and Gemini opened as three tabs on my browser and phone at all times.
Also, with Kendrick Lamar and Drake in the news, I may have included an AI generated rap battle between me and our competitors.
And maybe that rap battle concludes with a guest appearance from Jay$.
GenAI Cheat Sheet
ChatGPT (OpenAI)
Class leader for most basic uses. Very focused on speed to market and new features. Just released a new model with voice. More focused on users and retail usage than corporate uses. They have awesome free models. Generally more creative than the other models, but also more prone to hallucinations.
Claude (Anthropic)
Caught up to ChatGPT recently for classic use cases and I actually prefer it for reading PDF attachments. Any of our loan, SREO, and hedge confirm AI abstracts are done via a highly secured Claude backend application we had to build with AWS. Don’t go dropping confidential docs into Claude - ours was a year long and insanely expensive build. Less creative and more conservative in its answers, which makes it less fun to play with but better for business cases.
Gemini (Google)
Late to the party but catching up quickly when they aren’t getting busted for rigging demos. I mostly use this for classic “Google search” functions since it is real time, whereas ChatGPT and Claude have training cut off dates to control spiraling computing costs. If you use Google search, try using Gemini as an alternative to start dipping your toes in the AI game.
Ajax (Apple)
Not available to the public yet.
This Week
- 10T: 4.42%
- German Bund: 2.51%
- 2T: 4.83%
- SOFR 5.32%
- Term SOFR 5.32%
A Small Exhale
Last week’s CPI data should help alleviate concerns about inflation reaccelerating. The reports largely came out as expected, lower across the board than Q1. If only someone had been saying to not get overly excited about Q1 CPI data…
By now we all understand the lag effect from shelter, so let’s look at the BLS’s New Tenant Rent Index (NTRI). That’s not a sarcastic name, it actually focuses on just new tenants. It peaked in mid-2022 at 12% and is now just above 0%.
Look how much Owner’s Equivalent Rent (OER) lags behind, both on the way up and on the way down. That lag is why I wagered JMo in the first place.
Isn’t it possible the Fed was slow too move because the OER lag and now they are repeating the same mistake all over again?
Plus, the quality of the survey data has deteriorated substantially as response rates have fallen. Which do you trust more - surveys or actual stats?
I’ve gotten a lot of smack talk from people about being wrong about inflation and rate cuts this year. But we’ve only had 3 months of data so far this year, and Q1 is notoriously seasonal.
Next week’s Core PCE will be the fourth reading of the year (so it’s released a month behind). Let’s assume it comes out at 2.7% - that’s just 0.1% from the Fed’s year end projection when they were expecting 3 cuts. With 8 months to go.
Just four months into the year and Core PCE will be at 2.7%-ish. And that’s with the BLS increasing the OER weighting, which slowed the progress because of that lag.
How quickly we forget last year’s inflation story.
- April 2023’s Core PCE print? 4.8%
- March 2023 (month prior)? 4.8%
- Year end 2023? 2.9%
Rates
Odds of a rate cut in September stand at 65%. Sept/Nov/Dec are the last three meetings, so if they start in September they could do three cuts by year end, putting Fed Funds sub-5%. I know many believe that sounds crazy, is it when you consider the inflation stuff above?
Bloomberg had a good article on demand for US Treasurys. I get this question a lot as people freak out about debt to GDP ratio and deficits and no one will buy our debt and rates to the moon. That’s not how it works. Treasurys are different than any other asset class in the world.
As the article notes, demand for US Treasurys has been consistent for a decade. The “bid to cover” ratio is a measure of Treasury auction demand.
- “Bid” is the measure of bids to buy Treasurys
- “Cover” is what was awarded
This creates a measure of the ratio of demand vs supply. A ratio of 0 would mean demand exactly matches supply. Look how consistent demand has been, with a spike in flight to safety following the GFC.
Oh, guess what? That dashed line isn’t 0. It’s 2.5, meaning demand for Treasurys is 2.5x greater than supply.
Zero doesn’t even fit on the graph.
Can huge deficits create less demand for Treasurys? Sure, but I wouldn’t count on it materially moving yields anytime soon.
Random Graph of the Day
Perhaps the most overlooked data point from last week was Retail Sales, which came in substantially weaker than expected. Since consumer spending comprises 2/3rds of GDP, it could suggest the consumer is finally buckling.
Courtesy of SMBC, credit card and auto loan delinquencies hit their highest level since 2011.
The Week Ahead
Tons of Fed speeches, which will all be versions of “We need more confidence on inflation progress before cutting rates.”
Wednesday brings us the minutes from the last FOMC meeting. It will be interesting to see if any members were discussing the potential for hikes. I doubt it, but the market will be on guard for that.
Hedge Advisor Rap Battle (Listen Here)
Given Kendrick Lamar and Drake’s rap battle, I thought it would be interesting to see how Claude would have us battle with the other hedge advisors out there.
Pensford
Yo, it’s Pensford in the ring, call me the rate sage,
Missed a prediction or two? Yeah, but I turn the page.
Two years back, Fed moves had me in a daze,
But I evolve, adapt, and still set the stage.
Competitors
Hey Pensford, you think you’re the guru of the rate,
Missed the hikes, now you’re just trying to skate.
Chatham here, we’re giants, you’re just bait,
Sports in finance? C’mon, bruh, that’s second rate.
Pensford
Chatham’s got the size, but where’s the agility?
Forced out your founder, lost more than stability.
I keep it real, take a stand with ability,
While you sit back, no opinion, no liability.
Competitors
Pensford’s small time, flying in that tiny plane,
No market sway, just playing a losing game.
Your jokes fall flat, just like your rate claim,
Stick to sports, ‘cause in rates, you ain’t got no fame.
Pensford
CIRM, oh please, your strategy’s a maze,
Complex for no reason, clients lost in a haze.
And RCA? That flat fee’s just a phrase,
Blindsiding custys at closing? Your ethics, a craze.
Competitors
You’re just jealous, Pensford, of the leads we acquire,
Cold calls, maybe, but they ignite the fire.
Your data obsession? That’s just you in the mire,
While we adapt and soar, ever higher.
Pensford
Derivative Logic, you’re just copy and paste,
Stealing my style, but you can’t match my taste.
I’m the original, in this game, I’m laced,
With clever comebacks, watch how you’re displaced.
Competitors
Pensford, all talk, but where’s the proof?
Your forecasts fail, they’re just aloof.
We might borrow, but we stay under the roof,
Of innovation and truth, not just glib reproof.
Pensford
So here’s to Pensford, where cleverness reigns,
We miss a beat, but our humor sustains.
I’ll keep throwing verses, breaking these chains,
’Cause in this rate battle, it’s clear who gains.
Jay$
Hold up, hold up, let me step in this cypher,
Jay$ on the mic, about to make it hyper.
Y'all acting like children, throwing tantrums and shade,
While I'm steering this economy, making tough trade.
Pensford, you question my moves, but where's your degree?
State school educated, thinking you can school me?
I've got the cred, the experience, the clout,
My decisions are sound, no need for you to doubt.
Two years of rate hikes, and we're still standing strong,
Soft landing achieved, proving y'all doubters wrong.
I'm tired of being judged by those who can't see,
The bigger picture, the complexity, the responsibility.
We're in the last mile, inflation's final stand,
Like 8 Mile's battle, it's time to take command.
The Fed's my 313, Detroit of finance,
We'll slay this inflation beast, no matter the chance.
So keep your petty squabbles, your juvenile rhymes,
I've got a job to do, guiding through uncertain times.
Jay$ out, dropping the mic with finality,
'Cause when it comes rates, I'm the ultimate authority.
.png?width=640&height=366&name=image%20(12).png)
