โ† All Episodes

Apr 27, 2026 ยท 45:18

Rules for Thee, Alpha for Me

Unqualified Fact Check

71%

held up

4Nailed it
2Close enough
1Whiffed it

Strong on the data, sloppy on the date โ€” the STOCK Act is a 2012 law, not a 2004 one, but the directional argument holds.

Show Notes

This week Dan and I sit with a question that has been bothering me for a while and got sharper this week thanks to a perfect natural experiment: a US Army NCO got charged for making four hundred thousand dollars on Polymarket betting on the Maduro raid he was part of. Around the same time, Senator Markwayne Mullin bought between fifteen and fifty thousand dollars of Chevron โ€” the only US-listed company with Venezuelan oil operations โ€” days before the same operation. The senator's position is up about twenty-two points over the S&P. The soldier is going to prison. The senator is going to a committee hearing.

We get into why that gap exists, what the data on congressional trading actually shows (it's more nuanced than the headlines), why the STOCK Act has a two-hundred-dollar fine and zero prosecutions in roughly a decade, and whether the answer is more rules or more sunlight. Spoiler: I came in believing one thing and Dan talked me partway out of it.

We also fall down a few side roads we couldn't resist โ€” the French guy who spoofed a Polymarket weather contract by holding a hairdryer to the airport temperature sensor (still my favorite grift of 2026), the term-limits and lifetime-appointments question, and a riff about reading David Deutsch and Rory Sutherland in the same week. Dan plants a seed at the end about something we've both been wrestling with: what happens to investing when you bolt a probabilistic machine (LLMs) onto a deterministic one (the legacy quant playbook). That one's coming in a future episode.

Thanks for sticking with us. We are, by Dan's accurate diagnosis, "the weird ones because this is what we choose to do with our weekend." Go build something.

โ€” Sean

Links

Timestamps

  • 12:0812:23): "NANC, plus about 30%... S&P plus 25%... Republicans... trailing the market by a good 8 percentage points." Approximately correct. Trailing-12-month: NANC ~28.78โ€“31.54%; KRUZ ~18.14%; SPY ~25%. Sean was within a point on each.

Predictions from this episode

PendingSean

Elon Musk will engineer a merger between Tesla and SpaceX by 2028. The SpaceX IPO (10x voting-share dual-class structure) is the prerequisite vehicle for keeping Musk in control through the combination, putting "all the robots and all the AI" under one Musk-controlled entity.

โ€œSo I say 20, 28, 2028 that goes down... That's my guess. So that'll end up in our predictions file and in a couple of years we can see if I was close.โ€
Said in: Rules of Thee, Alpha for Me (Apr 27, 2026)Horizon: End of 2028Target: Dec 31, 2028Confidence: Load-bearing โ€” Sean explicitly tagged it for the predictions file on-mic.
Came trueSean

SpaceX will IPO around June 9, 2026 (the "6/9" date is a joke / tonal play, not a literal claim โ€” the underlying prediction is "June 2026"). The offering will use a 10x voting-share dual-class structure ("StocktickerXX" or similar โ€” `X` is taken by US Steel) so Musk retains control through subsequent mergers.

โ€œ6-9, 6-9, what better date for an IPO of SpaceX than 6-9?โ€

Resolution: Effectively confirmed. SpaceX filed a confidential draft S-1 on Apr 1, 2026 and a public S-1 on **May 20, 2026**; the roadshow began **June 4, 2026**, with final pricing set for **June 11** and shares expected to begin trading **June 12, 2026** on Nasdaq under ticker **SPCX**. Sean's operative claim was "June 2026" (the "6/9" date was a tonal joke) โ€” that's a hit. Two deltas from the call: (a) the ticker is **SPCX**, not the dual-class "X"-style play Sean riffed on (X is US Steel), though reporting still emphasizes Musk retaining outsized ownership/control โ€” consistent with the control-retention thesis; (b) reported valuation $1.75โ€“2T, raising ~$40โ€“80B. Mark Came true on final pricing June 11. **Activates the downstream chain** โ€” `tesla-spacex-merger-2028`, `spacex-acquires-cursor`, and `tesla-138x-forward-earnings-unsupportable` are now on live clocks. Sources: CNBC, SEC S-1 (EDGAR), re-verified 2026-06-05. **Listing confirmed (2026-07-19):** SPCX priced at $135 on Jun 11 and began trading Jun 12, 2026 on Nasdaq โ€” the largest IPO in history at a ~$1.77T valuation โ€” opening at $150, touching $176.52, and closing at ~$161 (+19%) on the second-largest IPO-day volume in Nasdaq history. Came true stands, final. Sources: CNBC, NBC, Wikipedia (2026-07-19).

Said in: Rules of Thee, Alpha for Me (Apr 27, 2026)Horizon: End of June 2026 (~2 months from prediction)Target: Jun 30, 2026Updated: Jul 19, 2026Confidence: Load-bearing for the broader merger thesis. The exact 6/9 date is rhetorical; "June 2026" is the operative claim. Note: Sean cites "everybody's saying June" โ€” this is a market-consensus echo, not a unique forecast.
Came trueSean

Post-SpaceX-IPO, SpaceX will acquire Cursor (the AI coding company) to fill the SpaceX data-center capacity that Grok isn't using. Acquisition happens "a few months later" after the IPO. Eventual use case: full-self-driving AI training inside the merged Tesla-SpaceX entity.

โ€œthey're buying cursor, which is for those who aren't aware, an AI coding company who makes actually a pretty nice tool... SpaceX built this giant data center that they don't really have any use for because Grok isn't that popular. So they need something to fill it.โ€

Resolution: Scored โœ… โ€” and faster than even Sean's framing. **SpaceX announced the acquisition of Anysphere (Cursor) on June 16, 2026 โ€” four days after the IPO** โ€” for $60B in an all-stock deal, the largest venture-backed startup acquisition ever, expected to close in Q3. The staging thesis was vindicated in the filings: SpaceX had quietly secured an *option* to buy Anysphere on **April 21, 2026 โ€” pre-IPO** โ€” with a ~$10B breakup/deferred-services fee, exactly the "deal is being staged" structure Sean inferred. Reported strategic rationale centers on AI coding + xAI compute integration (Cursor trained on tens of thousands of xAI chips) rather than Sean's specific "fill the idle data center" mechanism โ€” the market reaction was brutal (SPCX shed ~$600B in the four days after the announcement), but the prediction was the acquisition, not the reception. Company, sequence, and timing: direct hit. Sources: Quartz, DevOps.com, Motley Fool, Yahoo Finance (2026-07-19).

Said in: Rules of Thee, Alpha for Me (Apr 27, 2026)Horizon: End of 2026 (contingent on `spacex-ipo-june-2026` actually happening)Target: Dec 31, 2026Updated: Jul 19, 2026Confidence: Load-bearing for the staged-sequence narrative. Note: the public framing is that Cursor *couldn't* be acquired pre-IPO due to disclosure mechanics; Sean treats this as evidence the deal is being staged.
PartiallySean

Tesla's ~138x forward-earnings multiple is partially "Elon premium" (the only listed Musk-company exposure) and will compress when SpaceX provides a direct-Musk public-market alternative.

โ€œThere's not the value there. Like they have to, mean, some of that's an Elon premium, right? Because it's the only way people can actually buy an Elon company. So you would expect to see that possibly Tesla shares dropโ€

Resolution: The trigger fired (SpaceX priced 11 Jun 2026) and the **directional** call landed hard: **TSLA โˆ’12.6% from 11 Jun to 28 Aug 2026** ($399.15 โ†’ $348.75) against the **S&P 500 +4.3%** โ€” roughly **17 percentage points of underperformance** in eleven weeks, exactly the "Elon premium leaks to the new listing" shape Sean described. **But the metric he named moved the other way.** Forward P/E is now **183x (StockAnalysis, 29 Aug)** to **201x (GuruFocus, 28 Aug)**, against GuruFocus quarterly readings of 179x in Q1 2026 and 193x in Q2. Trailing P/E is 362x. The multiple **expanded while the price fell**, because Tesla's **Q2 2026 print on 22 July** โ€” record revenue, sizeable EPS miss, **operating margin 1.4%** โ€” sent the stock **โˆ’14.5% on 23 July** and forward EPS estimates were cut faster than the price. So the compression thesis is falsified by the very number it was stated in. โš ๏ธ **The baseline is unverified.** No source found supports a **138x** forward reading for TSLA at any point in 2026; every series located starts higher (179x+). Either the figure came from a different forward year or a pre-cut consensus. That makes the compression claim harder to rescue, not easier. **Standing lesson for Canon:** a multiple-compression prediction needs its starting multiple sourced at the time it is made, or it cannot be scored later. Sources: StockAnalysis, GuruFocus, Electrek (Q2 2026 results). (2026-08-29)

Said in: Rules of Thee, Alpha for Me (Apr 27, 2026)Horizon: Contingent on SpaceX IPO; evaluable 30โ€“90 days post-IPO via TSLA price-action vs. macro-controlled benchmarksUpdated: Aug 29, 2026 (2 revisits)Confidence: Medium โ€” directional, not a magnitude claim. Hosts acknowledge they personally got out of TSLA at low prices and missed the run, so this prediction is offered with self-aware "stupidity for the trees" framing.
PendingSean

The top-decile alpha of congressional traders persists at roughly +30% over market in any given era, even as the *average*-member alpha hovers near zero post-STOCK-Act.

โ€œRegardless of the time era, there's the top decile of Congress that trades at about plus 30%, which is impressive that that's over every era.โ€
Said in: Rules of Thee, Alpha for Me (Apr 27, 2026)Horizon: Ongoing โ€” measurable annually; reasonable review at year-end 2026 and onwardTarget: Dec 31, 2026Updated: Aug 29, 2026 (8 revisits)Confidence: Load-bearing for the segment's thesis. The exact +30% figure is more of a stylized claim than a precise forecast โ€” the *direction* (top-decile alpha persists despite STOCK Act) is the testable part.