…of intelligence falling? And what does competition do to that answer?
4. What does the switching cost consist of and does it replenish with AI workload?
5. How much of this thesis requires correctly forecasted frontier model usage?
Some simple questions to evaluate software in the age of AI disruption:
1. When the human user is removed from the loop, where does the value capture sit?
2. Write revenue as units x price. If AI reduces the unit count by 80%, what happens?
3. Who benefits from the cost…
Only if you take analogy to be literally true and without caveat
Analogy helps point towards questions and failure modes may have missed
It’s a computationally cheap way to generate a set of hypotheses, if you are rigorous about what is structural similarity vs language similarity
It does degrade silently as the analogy gets worst and it carries heavy weight of the reasoner’s priors
But dismissing it is a lazy analogy
…franchise value, marking credibility, sector concentration etc
Convex Tail captures equity stakes, warrants, post-default workouts
That’s sort of it, BDCs not a mystery, they are just basic CLO-like funding structures
A simple framework for thinking about BDC valuations:
BDC % of NAV = sustainable forward ROE/Required ROE + Convex Tail/NAV
sfROE is quantitative and captures book turnover, funding costs, fee structure, defaults etc
rROE captures qualitative factors like manager quality…
*SATA does not feel like a proper direct substitute and should trade 100-200bps wide of STRC anyhow given its size, liquidity, history, and lesser governance controls/auditing
*apxUSD and sUSDat were the creations of brain dead individuals, and the reflexivity is painful
Some thoughts on STRC:
*feasibly much cheaper to hike STRC rates by 100bps (annualized cost around $100mm per year to MSTR) than sell BTC and buyback STRC (a 1.6% BTC move also worth $100mm)
*32 BTC sale was meant to be a ceremonial crossing of a red line, in favor of pref…
….Especially to encourage more institutional buying activity given STRC is 80% retail owned
*STRC is the main MSTR funding source given the convert NAV trade (which used to go to funds like Jane Street) is dead with low NAV premium, Saylor has every incentive to protect
After a long hiatus, I am returning to the twitter sphere
Will be focusing on fundamental investing across asset classes - equities, special sits credit, structured credit, and general market commentary.
Need an outlet for brain dumping but don’t have the time for Substack
@maxresnick Not at all
“Hey we see that there’s clearly $90 of assets with probability ~1 to cover your $100 deposit, so we feel fine giving you $40 early. They’re all accounting entries anyhow. When we sell the $90 of assets we will pay ourselves back the $40 and you get the rest”
Quick look at SIVB probably caps losses at 15% MAX if I were betting
Most likely outcome in my mind is 0%, given reg. bodies involved + a decent chance of a sale over the weekend. It’s good politics.
Even at 15% loss USDC at 96-97 is mispriced. 15% * 8.5% assets = 1.2% loss.
@maxresnick Bails who out? The FDIC covers their insured portion and then market resolves the rest.. there’s no gov bailing out required if a full sale of the bank goes through
@maxresnick Meaning, for example, if they’re confident depositors will recovery 90%, they could advance everyone 40% while they wait to liquidate assets
@maxresnick That’s partially a misconception.
If there’s a sale of the bank over the wknd, may be able to withdraw next week.
Even if not, FDIC insured deposits will be withdrawable next week according to my understanding with counsel. FDIC can also issue “advances” on uninsured portions