Markets, Macro, AI, Tech. Head of AI + Strategy at
@SmarshInc
. Previously CPO/Head of AI Research
@dreasoning
, Global Macro
@nomura
. All opinions my own.
2023 Investment Recap
What Worked
• OSCR (+167%)
• TALK (+186%)
• COIN (60% on structure, 400% on spot)
• Grayscale Convergence (350%)
• TLT (12%)
• Municipal CEFs (18%)
• Bank Preferreds (73%)
• Gen AI (this got stupid)
• PINS (78%)
What Didn't
• AGNC Preferred vs
@DavidSacks
@elonmusk
That’s not right David. You can test the error with a binomial distribution. This is the Law of Large Numbers.
Twitter samples 9,000 per quarter or 36,000 per year.
At 5% sampled mean the standard error is sqrt(5% x 95%/36K) = 0.36%.
Twitter is right.
One of the most illuminating papers you can read on The Great Depression is Martha Olney's
"Avoiding Default: The Role of Credit in the Consumption Collapse of 1930"
By putting yourself in the shoes of then-consumers, things start to make sense...
1/5
@FedGuy12
Thanks for sharing
@FedGuy12
. Here’s another view of it from their investor presentation.
Notice that Portfolio duration and Hedge-adjusted are nearly the same which indicates nearly no hedging.
A few very interesting add-ons reading through the investor deck for JPM/FRC.
This deal appears to be a *fascinating* feat of financial engineering.
1/5
The game theory around First Republic, the FDIC, the banks and the White House is a bit crazy.
When you walk through it, today's announcements from the FDIC and the US Government make sense and may be an attempt to get the banks to recapitalize FRC.
1/12
When you look at the Fed's H8 data for Small, Domestically Chartered Banks, you can see exactly what happened...
First, the run on deposits: $250 billion over two weeks.
1/5
@rscotteisenberg
@DavidSacks
@elonmusk
The standard error describes the margin of error for a sampled mean.
Population size isn’t part of the computation. The size of the sample is.
A multiple of your standard error gives you the confidence interval.
The Law of Large Numbers is why it converges. Hope that helps.
So many people on Twitter are confusing coupons with convexity right now.
Yes - there’s an asymmetry to 20 years falling by 20 bps vs rising. But the majority of this is due to coupon, not convexity.
Moreover - if you want to do proper bond math you need to account for:
•
Let's take a quick look through First Republic, and how they're navigating.
To begin, uninsured deposits fell hard (by nearly $100 billion), but are now stabilizing after banks added $30 billion.
Why? Only $19.8bb is now uninsured and the flight risk is now much lower.
1/10
But consumption decreased a lot.
The small print matters. And Prisoner's Dilemmas can have major impacts.
By 1938 the terms of default had largely changed and we prevented a similar situation.
Worth a read.
The math on this is wildly incorrect. I would encourage you to correct and retract this
@mikealfred
.
If you look at the increase in the actual debt maturing the payment goes up by 22% versus 100%.
If you look at the interest payments versus Federal Tax Receipts (total, not just
Roughly half of all Treasuries outstanding today will mature by the end of 2025.
The current average interest rate on federal debt is 2.6%.
The current prevailing interest rate is over 5%.
If nothing changes, the US government will be forced to refinance this debt at roughly
@maxjanderson
Interesting analysis but you don’t need Net Liquidity. You’re largely just looking at FRB Reserves Balances.
Assets - RRP - TGA
= (Liabilities - RRP - TGA) + Capital
= Reserves + Currency + Other + Capital
Reserve changes is how policy is designed to work.
CC:
@dampedspring
People tightened their belts on consumption in order to not lose their equity in key items.
This began a spiral: decreased consumption led to economic slowdown led to layoffs led to decreased consumption.
Default rates on cars increased, but not remarkably.
4/5
How Microstrategy funds for so cheap…
1. Buyer buys the converts
2. Sells long-dated high strike calls against it
3. Receives small coupon
Because vol is so high buyer covers all of credit risk.
Pure arbitrage at the expense of retail traders.
Hat tip
@macrogliblyglo1
In the late 1920s people had taken out a large number of loans for autos, appliances and pianos.
These loans were typically 12 to 24 months in duration, with 10 to 40% down payments, and remarkably high interest rates.
Debt was new.
2/5
This is one of my favorite charts to look at.
Deposits, Stocks + Bonds held as a % of Disposable Income vs the 10Y inverted Treasury Average Yield. Data is through Q2: recently would be 4.50+.
You can immediately see a few key things:
(
@concodanomics
inspired by your 🧵s)
1/6
The issue was the structure of debt.
Contracts typically did not allow for you to sell your durable good to pay off the loan.
Moreover, if your durable good was repossessed, you often lost the surplus.
This is actually the system working the way it's supposed to against exogenous shocks.
I expect there will be earnings hits this quarter but that the sales may also help with capitalization ratios.
5/5
People that are laser focused on JOLTS misunderstand the economic situation. There are structural (vs business cycle) factors at play. Let me outline the specifics of why and what to do.
1. The Labor Force has returned to pre-pandemic levels.
@alexstanczyk
@joerogan
@PeterZeihan
Alex - much of what he is saying is correct. These are fundamental economic principles.
The technology is awesome - but he’s right that fixed supply does not support growing economies.
I may write more on this, but if you fast-forward the following economic equation it highlights inherent issues with Bitcoin as *the* universal currency
Money Supply x Velocity of Money = Real GDP x Prices
This is typically expressed as MV = PY
@michaeljburry
@balajis
1/5
@CloudsGalore
@DavidSacks
@elonmusk
That’s only true if the entire population turns over each day which is not the case. There will absolutely be correlation.
There will be some attenuation, but each day is not it’s own analysis. That would assume daily population independence.
Fathers be good to your daughters.
As the father of an 8 year old girl this has my attention more than ever.
Thanks
@donnelly_brent
for first tipping me off to this.
By far, my favorite question at any point in the investment cycle is this:
“How much has to go right to justify this valuation?”
Or more mathematically:
“What percent of plausible future states would justify this valuation?”
This isn't a prediction. But there's so much fear-mongering on Twitter that I wanted to walk through solutions.
None of this is easy, but it's not impossible. You've got to concurrently keep your culture in place and staff motivated.
I wish the leadership team the best.
10/10
@DavidSacks
As you know - the securities can be worth less than the deposits in a rising rate environment.
So the product you’re describing would still have risk of a run.
Feat 2: Re-orienting the creditor stack.
Uninsured deposits being made whole while the FDIC is taking losses.
Technically you'd go: insured deposits, FDIC, uninsured deposits.
Instead we are going: insured deposits, uninsured deposits, FDIC.
Aligns with previous deals.
3/5
There are a few things that are easy to overlook in these sorts of analyses. The thread below is well-intended but misses the mark.
1. The tech sector is now much more mature and diversified than it was in the late 90s.
You need an inferred multiplier:
P/E Observed =
P/E for
2000 peak vs now:
Tech sector traded at 2x its profit share vs 1.25x now.
SPX was at a 25x forward P/E vs 20x now.
Translation:
S&P 500 would need to reach 6250 to price-in same level of irrational exuberance as 1999-esque - per Soc Gen
Here's an ominous indicator for the Housing Market.
The 10-year government bond now yields a higher return than the Cap Rate, or profit from operating a rental property.
No wonder real estate investor demand is collapsing. More profitable to buy bonds.
4. Why aren’t these workers returning? We can see that disabled individuals increased during COVID by 3 million, which more than explains the shortage.
Remember that such individuals often need caretaking (which also helps explain the need for healthcare workers).
@BillAckman
@federalreserve
Bill - you were actively pushing for large hikes in Q3.
Were you expecting a different outcome than what's transpired? Honest question.
is a fiction. Rates are going up a lot soon. The sooner the Fed quells inflation, the better for longer-term bonds and long-term financial assets like equities. Don’t be misled by short-term, technically driven market movements. Stocks of high quality businesses with long-term
@Mattfox68569363
Fabozzi's Fixed Income is the canonical work.
However there's a 6 part series by Antii Ilmanen many years back that I found to be the best work.
There are three possible responses to this:
1. Sufficiently drop aggregate demand / GDP so that the workers aren’t needed.
2. Public programs to rehabilitate the disabled.
3. CapEx investment and automation.
There’s too much focus on
#1
. Shrinking an economy is bad policy.
$880 billion of potential fraud from pandemic relief programs.
This is the size of 2008 bailout but entirely fraudulent.
It’s no wonder prices of Rolexes, Lambos and Airbnb properties went to the moon.
As the bond market rallied, this enabled them to begin selling off some of their securities portfolio: $168 billion.
There were also circa $70 billion in loans sold, ostensibly in private transactions.
3/5
For full sake of clarity.
1. If the stock rallies like crazy they have a free call spread.
2. If it stays put and Microstrategy doesn't default they're paid both principal and call premium for massive ROI.
3. If they default they recoup most if not all losses via calls sold.
Feat 3: Lots of wins
- FDIC paid $10.6B by JPM
- JPM recognizes gains of $2.6B
- JPM 20% IRR and $500mm net income accretion
- Tangible book value accretive
- Capital ratios intact
- Everybody gets deposits back
4/5
Feat 1: "Transforming" high risk, high funding to low risk, low funding.
The FDIC loss-share reduces risk-weighting on covered loans to 25% - *far* lower than typical for these assets.
The FDIC is also providing fixed-term financing of $50 billion.
2/5
Sigh, major factual error.
The specific coffee series that was removed was last published in 1988. It’s still in the CPI, just not with a specific can size.
Again: be careful when people start with the narrative and use data to back it. Details matter.
Why is the
@BLS_gov
removing the price of coffee from the inflation report?
Coffee is an important part of America prices, almost as important as gas prices (we already exclude gas) and we are removing it from the CPI INFLATION REPORT!
H/t
@DonMiami3
Many people are talking past each other on SVB.
Why? The situation is complicated - and both the Points and Counterpoints can be true at once. I’ve compiled many I’ve heard below.
What’s important is to resolve this in a manner that mitigates financial and technology contagion.
An important lesson I saw on Wall Street.
Markets teach humility. Everybody is wrong, a lot. Those who realize this, are willing to change their mind, and adapt: survive. Those who do not, do not.
Very few career paths force this kind of adaptation.
Great article by
@FedGuy12
.
This aligns with the point I've been making. IOER and massive excess reserves stunted the transmission mechanism of hikes similar to the way that Reg Q stunted Burns' hikes.
For hikes to impact, Powell needs to increase QT instead of more hikes.
To put the potential China bailout into context.
$280 billion is 4 times the record ETF inflow of last year.
It is about 9x the largest net investment inflow via the Stock Connect program.
As a percent of total market capitalization it’s much higher yet.
Breaking News: Bloomberg has learned Chinese officials are considering a market rescue package in the order of $280B. Markets responding nicely initially.
Details to follow. Stand by.
There are two questions it’s important for you to answer
@balajis
:
In your Bitcoin maximalist world:
1. How do you prevent debt-deflation cycles?
2. How do you propose to address trade imbalances with a fixed currency?
VOTE FOR BITCOIN
The post below by
@MacroScope17
is excellent. Let me extend his point one step further: the real election is BTC vs USD, the primaries have already started across the world, and every ballot counts. So make sure to vote early and often. Here's why this is more
As far back as I can remember I have been able to see macro trends clearly. However - most of that time I have been a terrible portfolio manager.
In the last few years I've changed markedly. My wife asked me lessons learned – which I thought I'd share here. I'm not sure why
Post-mortem on First Republic:
- JPMorgan acquires/assumes all deposits
- Doesn't assume preferred/unsecured notes
- Loss-sharing w/FDIC (details below)
- Est. FDIC loss of $13 billion
- One-time gains of $2.6 billion for JPM
1/7
The reason that this can work is because most of the assets on First Republic's balance sheet are likely money good. This is in huge contrast to the 2008 Global Financial Crisis.
Most assets will revert to par over time.
9/10
There is a lot of misunderstanding about how bonds and mortgage securities work.
I put together this diagram so that people understand:
bank "mark-to-market" losses on their Held to Maturity Portfolios will self-correct, even if rates stay high.
1/10
Liabilities: the bank is heavily funded by the Fed, FHLB and big banks.
To put it in content: $135.9 billion of $215.0 billion (63%) of liabilities come from government- and GSIB-related activities.
This will compress NIM, but it will be sticky + reduce liquidity risk.
3/10
@LTCM_ETF
Coupon being the recurring payment that the bond pays out,
In bond markets carry is calculated as the difference between short-term borrowing costs on a security vs the cash paid out via coupons.
Since short rates are higher than long-dated yields that is negative right now.
@taobanker
This comes from the Taylor series decomposition.
Duration is the first order approximation of the change in price versus the change in yield. First order linear approximations are by definition symmetric.
The convexity term is the second order part, which is asymmetric.
Finally - we can also normalize by assets to look at the composition.
You can see that:
1. Deposits have trended lower over time
2. Stocks are near secular highs
3. Bonds are near secular lows
6/6
To begin: Wedbush had estimated that First Republic's tangible book value if marked fully to market would be negative $73/share.
If correct, the $13.5 billion capital hole is what creates an issue for any prospective buyer.
2/12
Short note on how the First Republic Bank “rescue” works. There is a ton of confusion out there: my goal is to help quell panic.
As of Dec 31, 2022 the bank had $4.23 billion of cash and cash equivalents.
This makes sense: as a bank your deposits are pretty “sticky”.
“What’s going on nationwide is every one of these banks has either frozen their loan-to-deposit ratio or, more likely, is very intent on shrinking it”
Former Dallas Fed President Robert Kaplan
The market is up 11% since this call.
Tech outperformed to the upside with AI being the major driver.
Short-term, the easy money has been made on this. Long-term the thesis holds. I'm largely in uncorrelated bets that should trend higher over time.
@rleshner
This is patently false.
Study history. DeFi does not work.
The reason is human nature: greed creates leverage and fear creates contagion.
There must be an uneconomic buyer of last resort. The longer it takes to relearn this lesson, the more painful.
@parthiv645
@DavidSacks
@elonmusk
We know as sample sizes goes up the margin of error goes down.
In stats the population size isn’t relevant in the margin of error.
If you have a new method or reason to overturn stat theory, please do post.
In March, a bank consortium deposited $30 billion in uninsured deposits (this is key) for an initial term of 120 days.
This mitigates liquidity issues for First Republic. As of April 21, 2023, they had $45.1 billion of cash/equivalents + unused available borrowing capacity
3/12
Impressive results out of JPM
Consumer/Community Banking Net Income +80% YoY
Importantly, look at Loan Loss Reserves: they increased, but decelerating
This tells you that JPM had room within their earnings to increase them by even more, and chose not to
@BickerinBrattle
👏
Option Selling helps me generate $1000s per week in cash flow
My sweet spot for selling options is Delta 0.15 to 0.25
Let's talk about the benefits of selling options with a delta of 0.15 to 0.25
👇👇👇
4 step plan to fix the economy:
1. Fed to start leading + stop reacting to data
2. Launch Supply Chain Lending Facility w/ 0% interest rates to mitigate capacity issues
3. Drain excess cash in a targeted manner (tax hikes on very wealthy)
4. Fix issues w/ OER/CPI + housing
The net effect is that you recapitalize some of the $13.5 billion – wait out the losses on the assets (which eventually will be made good), and try to profit from upside on the warrants.
Which certainly sounds better than losing up to $30 billion.
11/12
This is a bank doing banking things.
They were hit because of VC deposit flight, which they replaced with brokered deposits (smart) and began sale of certain assets (smart).
Banks by their nature are prone to speculative attacks. Don't put people out of business to make a buck.
This is where it gets interesting.
If at the end of 120 days the banks chose not to roll over their deposits, First Republic would pledge assets to the FHLB and Fed to tap additional borrowing.
The collateralization is a key fact.
4/12
In May I'd indicated PacWest was under speculative attack and w/time they'd find a path forward.
Today marked their merger with Banc of California. All deposits honored, all liabilities honored, and preferreds honored. Common diluted - which is the way the system should work.
This looks to be a pre-written essay by Russian Media celebrating a successful military outcome in Ukraine restoring the unity of Russia, Ukraine and Belarusia. Sources at the end of the thread.
"Russia's military operation in Ukraine has ushered in a new era." 1/5
I'll share a contrarian view
Predictions of a near-term major recession will be wrong. Companies leaning into growth will be rewarded.
• Consumer, corporate + bank balance sheets are strong
• Consumer + corporate debt servicing is low
• Corporate balance sheets are strong
The recent banking situation is going to have broader impacts than most realize.
I’ll outline 4 key changes below and how this will drag on growth.
In a different thread I’ll highlight why this should translate into the Fed being done.
No pictures: this is dense.
1/6
It’s very hard to understand how this 30%+ return was not insider trading.
“On March 17, Representative Nicole Malliotakis, Republican of New York, bought shares of New York Community Bancorp after private discussions with New York State bank regulators.”
From NYT and WSJ
In this thread, I'll make a case for why the Fed should be done hiking, covering:
1. Importance of transmission mechanisms
2. A surprise reason Arthur Burns failed
3. The shift from fractional banking
4. How the bank runs unclogged policy
5. The danger of hiking more
1/14
So the government leaks that they are currently unwilling to intervene on First Republic as a clear signal to the banks.
As the banks you now have two options: see who blinks first, or act.
8/12
The Fed’s number one job is stability. Price stability is one part of a larger stability equation.
To overtly try to drive equities lower creates instability. Companies cannot plan with instability.
If the market gives the Fed what they’re asking for history will not be kind.
What’s happening in housing right now reminds me a lot of oil in 2007.
At the time people thought that oil was under supplied and that the replacement cost meant we wouldn’t go below $80.
Three things they missed:
1. Investors had bought up a ton of the inventory
2. Those
So if you're the FDIC you threaten to downgrade the scoring condition so as to limit borrowing from the Fed/FHLB.
Now in the event of receivership: the banks lose up to $30 billion because they are unsecured.
The FDIC gets away rather unscathed.
6/12
If you wait and the government blinks, then you may get your deposits back or a better funding deal on asset purchases.
If they don't blink, you're back to losing up to $30 billion.
And in either case, you've upset the hand that feeds you IOER.
9/12
If I were in Powell's seat, this is what I would convey:
"We are seeing broadly that banks plan to tighten their lending conditions. This does a good bit of work for us and places rates further into restrictive territory than they were in April.
As such, at the current time we
Tell me you don’t understand math without telling me you don’t understand math.
Let’s break it down:
$3mm/BTC x 21mm BTC = $63 trillion
This is:
- 2x the total amount of Treasuries
- or 3x the total M2 money supply.
Let’s assume for the moment this is a currency. The
This is both chart crime and a crime of understanding.
Chart crime: you have to normalize by Nominal GDP, Bank Assets or Money Supply. The number will still show an increase but much more metered.
Understanding crime: the BTFP created a major funding arbitrage where banks
The GFC bailout looks like just a blip on the radar vs what the Fed is doing today to support the banks with liquidity via loans/facilities.
And we wonder why the market is up.
This chart blew me away.
This is terrible analysis, misinformation viewed 200K times.
Deposits fell from $375mm to $343mm QoQ or 8.5%, not 30%. YoY analyses conflates shifts to MM funds.
The stock rallied because CET1 ratios are strong.
In the event of receivership (not expressing a view), the Fed and FHLB can (to my knowledge) seize the collateral against the loans.
This leaves fewer assets for the FDIC to sell – which would translate through into the FDIC carrying more losses.
5/12
WaFd one of the first regional banks to report w/loan originations -55%+ YoY ($1bb vs $2.2bb)
“Net loans outstanding grew…due to funding of…loans previously originated” but “the Company has intentionally slowed new loan production given the uncertain…environment”
The recent banking situation is going to have broader impacts than most realize.
I’ll outline 4 key changes below and how this will drag on growth.
In a different thread I’ll highlight why this should translate into the Fed being done.
No pictures: this is dense.
1/6
Generative AI will prove to be one of the most deflationary forces of our lifetime.
Governments will spend against it in order to balance the social impacts.
I get a very different valuation metric for S&P than
@RayDalio
has outlined here.
He indicates a similar methodology to my own but with very different results.
This measure calculates the earnings growth rate that is required to produce equity returns in excess of bond
Having been through many bubbles over my 50+ years of investing, about 10 years ago I described what in my mind makes a bubble, and I use that to identify them in markets—all markets, not just stocks.
Read what my bubble gauge says about the current
@Regenarian
@DavidSacks
@elonmusk
You’re misunderstanding.
5% is the sample mean.
= # of reviewed accounts deemed to be a bot / # of reviewed accounts
You’re calculating the proportion sampled (samples / population). We don’t need that for any of the math we’re discussing.
One key thing Volcker taught us.
When you need to break the back of inflation you have to maintain high rates farther into a slowdown than is normal and comfortable.
It’s the opposite of the Fed’s most recent guidance and begs the question as to political maneuvering.
While shopping for a new family car I pulled up some data that we know - but powerful to see.
Cars went from depreciating to appreciating to deprecating again.
During the appreciation phase it’s common for people to start to believe that the vehicle will keep going up and thus