November 2025 Random Ramblings
Summary
A winning streak can be a hot craps table masquerading as investment skill. Walker’s friend made roughly 5x on night one before losing every dollar of winnings on night three; similarly, Walker had very few losing legal situations for seven years, followed by three years when seemingly every one lost. He is now asking whether he ever had an edge, or whether early success weakened his diligence.
John Malone’s “levered buyback model” may have owed more to its operating environment than value investors admit. Growing EBITDA while maintaining roughly 4x leverage creates extra borrowing capacity and cash for repurchases, but the model’s 1980-2010 peak also enjoyed falling interest rates and extraordinary media economics. With Formula 1 the notable exception, SiriusXM, Charter, Warner Bros. Discovery, and QVC suggest the tailwinds may have mattered enormously.
Walker’s perhaps worst-performing bucket is stocks he initially rejected, then bought after they fell—often from $50 to $30—when his identified risk materialized. Correctly spotting the risk can create a dangerous belief that he knows the company “stone cold,” lowering the underwriting bar just when new diligence is most necessary. His unresolved question is: “What step am I skipping?”
Shorting Palantir, CoreWeave, or another perceived AI-bubble stock on valuation and disputed economics carries especially high risk. Their “nosebleed valuations” and debatable economics, TAMs, and moats resemble the late-1999 Cisco setup, but Walker says one would have to be “crazy” to short them naked and stresses that shorting is especially risky.
The neglected AI analogy is not dot-com collapse but the failed 2010-2015 shorts of Tesla, Netflix, Amazon, and Salesforce. Those companies drew persuasive accounting and valuation critiques—Netflix’s content depreciation, Amazon’s lack of economic profit, Salesforce’s multiple—yet shorts “got their faces ripped off” as several became exceptional businesses. Walker is not predicting the same outcome for today’s names; he wants investors to acknowledge it as a real alternative path.
Management-provided targets and NAVs are not, by themselves, an investment thesis. Three-year projections routinely disappoint, corporate overhead erodes headline NAV, and a discount already advertised by management is known to the market. Liberty SiriusXM is Walker’s warning specimen: the gap closed largely through SiriusXM falling toward the tracker, not the tracker rising toward management’s implied value.
Deep dive
1. A hot streak can manufacture confidence without creating edge
Walker’s opening specimen is craps: a novice friend hit a hot table, called it “the infinite money hack” and “a free ATM,” made roughly 5x on night one, did well again on night two, then lost all the winnings on night three. The casino’s edge had never disappeared; variance had merely delayed its arrival.
His investor test: after five consecutive winners, “is it evidence of skill,” or did a hot strategy manufacture confidence before its odds caught up? The same question applies to themes, repeatable-looking playbooks, and corporate strategies.
Walker’s legal special situations sharpen the point. For roughly his first seven years of professional investing—“to the extent that I am a professional”—antitrust, MAE, and other legal bets produced very few losers; over the last three, he felt that seemingly every legal situation he looked at was a loser, leaving him unsure whether skill, selectivity, luck, or weaker diligence explains either regime.
He never hid the limitation—during the Twitter situation he repeatedly said, “I am not a lawyer”—yet winning may have made parachuting into specialists’ territory feel justified. Believing a generalist can rapidly learn a sector and outinvest trained specialists may be “the height of ego.”
2. Malone’s levered-buyback machine may have been regime-dependent
The classic John Malone engine combined a growing subscription business with roughly constant leverage. If EBITDA rose 10%, debt could also rise 10% without increasing the leverage ratio; that incremental borrowing capacity, plus free cash flow, funded progressively larger repurchases.
Walker wonders whether the model’s 1980-2010 triumph reflected two exceptional tailwinds: declining interest rates and “the best period for media in history.” The spreadsheet logic still makes intuitive sense to him, but mathematically elegant leverage does not establish that the strategy works across regimes.
The last decade supplies uncomfortable evidence. Formula 1 worked, but SiriusXM, Charter, Warner Bros. Discovery, and QVC performed poorly; several businesses stopped repurchases, raised equity, or approached bankruptcy. Walker retains enormous respect for Malone’s intelligence and discloses a small Charter position, while conceding, “I was very wrong.”
3. Correctly spotting a risk can lower the bar at the wrong moment
The recurring setup begins when smart friends pitch a company at $50 and Walker passes because of a specific risk—say, Facebook entering a dating app’s market. The risk then materializes, the stock falls to $30, and he buys; instead of rebounding, it gets “crushed” again.
This has been “perhaps my worst-performing bucket.” His suspected mental error is that identifying the initial problem makes him feel he understands the company “stone cold,” even though new diligence may still be necessary after the adverse event.
Permanently refusing every company once rejected cannot be the general answer. Walker leaves the problem unresolved, asking how to reframe the setup and, “What diligence am I skipping?”
4. The AI short has both a dot-com precedent and a successful-tech counterexample
Walker sees a heavy recent drumbeat around an “AI bubble,” intensified when Sam Altman “basically” hung up after someone asked how a $10 billion company promising $1 trillion of capex over ten years would fund it.
Palantir and CoreWeave are the shorts he hears most often. Both carry “nosebleed valuations” alongside questions about economics, TAM, and competitive moats, although Walker explicitly allows that they might still be great companies and says naked shorting would be “crazy.”
The bearish pattern match is late-1999 or early-2000 Cisco: a real company at an unsustainable valuation. But from 2010-2015, hedge-fund managers with strong track records made similarly compelling valuation and accounting cases against Tesla, Netflix, Amazon, and Salesforce—only to get “their faces ripped off.”
Netflix’s content depreciation drew scrutiny, Amazon seemed never to produce economic profit, and Salesforce looked extraordinarily expensive. Walker’s underheard third path is that today’s AI leaders might be future-defining businesses; he is not asserting that outcome, and their current market values might already constrain the upside.
5. Management’s NAV is public arithmetic, not hidden value
Walker first separates numbers from personal trust. Friendly executives may promise never to issue undervalued equity, then discover “the deal of the century” six months later and issue equity amounting to 50% of the company—while also giving management a larger empire and greater compensation.
Long-term targets deserve similar skepticism. Without claiming a formal study, Walker is “pretty sure” most three-year investor-day goals are missed; he rhetorically asks how many 2021 SPACs projected $5 billion of revenue by late 2025, even though many still had not generated any revenue, while IBM famously missed its early-2010s EPS objective.
A typical NAV slide turns $1 billion of assets and 100 million shares into $10 per share, versus a $6 stock and a 40% discount. Even after checking every component, Walker’s results have been poor: management NAVs do not account for corporate overhead, and the supposed bargain is already visible to the market and potentially to quantitative systems.
Liberty SiriusXM is the clean specimen. SiriusXM at $4 implied roughly $40 of tracker NAV, if Walker remembers correctly, while Liberty SiriusXM traded near $25; Greg Maffei highlighted the discount and promised repurchases, but stress halted buybacks and brought a rights offering to support the rest of Malone’s empire during COVID. Ultimately SiriusXM fell toward the tracker as the float normalized—the “cheap” security did not rise to management’s NAV.