乔尔·格林布拉特(Joel Greenblatt) 视角 — HOOD

Verdict

Pass — excluded from the formula. Robinhood is a broker-dealer, a financial. My Magic Formula explicitly throws financials out before ranking, because EBIT, debt, "operating capital," and EV all stop meaning what they're supposed to mean on a balance sheet stuffed with customer money. I don't have a number I trust here, so I don't have a buy.

Lane

Magic Formula stock — and it fails eligibility at the gate. There is no corporate event that turns HOOD itself into a special-situation security. HOOD is the acquirer (Bitstamp, WonderFi, TradePMR, MIAX) — buying things isn't a spinoff, a stub, a post-bankruptcy equity, or a forced-selling situation. So this is not the special-situations lane either. It's a formula candidate, and the formula won't take it.

The Two Numbers

The honest answer is that I cannot compute my two numbers cleanly for this business, and that is the finding.

Return on Capital (ROC = EBIT / (NWC + Net Fixed Assets)): Not meaningfully computable. The data shows营业利润 (operating profit) of $2.14B TTM, but the denominator — net working capital + net fixed assets — is meaningless for a brokerage. The balance sheet carries $38.1B total assets dominated by customer custodial cash and securities (总现金及证券含客户托管 = $19.27B). The note in DATA.md says it plainly: "Robinhood作为经纪商，资产负债表含大量客户托管资产...分析时需剥离." Once you strip the client float, what's left isn't the "tangible operating capital the business needs to run" in the sense my ratio assumes. The XBRL even refuses to give a clean operating-capital read (CapEx and debt-structure fields are flagged missing/irregular). EV/EBIT and EBITDA are both marked n/a (金融行业). → Cannot certify "good business" on my terms. N.
Earnings Yield (EY = EBIT / EV): Cannot be computed as I require. DATA.md lists EV/EBIT = n/a outright. Even if I improvise — operating profit $2.14B over EV $78.26B ≈ 2.7% earnings yield — that is (a) not the EV-on-clean-operating-EBIT my formula demands, and (b) on its face a thin yield. For context the stock trades at P/E 45x, P/S 18x, EV/Sales 17x. That is not "cheap" by any reading. → N.
Combined rank standing (good AND cheap at once?): Below — not rankable. It can't sit in the universe to be ranked, and on the rough numbers it would rank poorly on the cheap axis anyway.
Is the ROC sustainable (one-offs stripped)?: Moot, because the ratio isn't valid here — but worth flagging that this earnings stream is the opposite of what I want even if it were rankable: FY2021 −$3.69B, FY2022 −$1.03B, FY2023 −$0.54B, then FY2024 +$1.41B, FY2025 +$1.88B. Profits that swing from massive losses to records in three years, driven by trading volume, crypto, and market mood, are not the durable operating engine the ROC test is meant to detect. Q1 2026 crypto revenue was −47% YoY. This is a cyclical, sentiment-geared revenue base wearing a high-margin coat.

Eligibility Check

Positive EBIT? Y (operating profit ~$2.14B TTM).
Valid EV? Technically EV = $78.26B > 0, but EV/EBIT is unusable for a financial — debt and cash here are intertwined with the client/brokerage book, not a clean enterprise-value buildup.
Not a financial/utility? NO. It is a financial — a broker-dealer.
Conclusion: Any single "No" excludes it. HOOD is a financial → excluded from the formula. Formula path stops here. Banks, insurers, brokers — their capital structures make EBIT, debt, and "operating capital" non-comparable to ordinary businesses. They don't belong in my screen, full stop. I am not going to pretend a number means something it doesn't just because the company is interesting.

Special-Situation Check

Catalyst present? No — not for HOOD as a security. The corporate events here (Bitstamp/WonderFi/TradePMR closed; MIAX pending) make HOOD the buyer. That's strategy, not a spinoff, recap, bankruptcy emergence, stub, or merger security that hands me a mispriced piece of paper.
Forced / uninformed selling? No. 2026 YTD −35% is sentiment and a crypto-revenue air pocket, not index-driven forced dumping of an orphaned security. A falling growth stock is not a forced seller; it's just a falling stock.
One-glance logic, no heroic forecast? No. To underwrite HOOD you must forecast PFOF survival (10-K lists its ban as a top risk), crypto cycles, trading volumes, rate-driven NII, and the integration of four acquisitions in two years. That is exactly the "predict the unpredictable" I'm built to avoid.
Insider/management incentives: Super-voting Class B control (Tenev + Bhatt run the company; public shareholders have no governance lever), and DATA.md notes insider net selling > buying over the last three months. Nothing here points an outside small investor toward an obvious, one-glance mispricing in my favor. → Not a special situation.

Valuation

EV/EBIT (and implied earnings yield): n/a per DATA.md. Improvised ~2.7% earnings yield (op. profit / EV) — not a number I'll stand behind for a financial, and unattractive even taken at face value.
Enterprise Value buildup (Market Cap + Debt − Cash): Market cap $83.92B + total debt $13.61B − cash/securities $19.27B ≈ EV $78.26B. But note the debt and cash figures here are entangled with the brokerage/client book — this is precisely why the buildup is unreliable for a financial.
Margin of safety (value vs. price gap): None I can certify. At 45x earnings and 18x sales, with a profit history that was deeply negative as recently as FY2023, there is no demonstrable gap between a conservatively-figured business value and today's price. My whole method is "pay a lot less than it's worth and let the gap protect you." I see no gap I can defend.

Position & Sizing

Mode: Would have been the diversified ~20–30 basket — but the name never enters the basket because it's excluded at eligibility.
Leverage: N/A (no position).
Rebalance / exit trigger: N/A — there is nothing to rebalance; it never qualifies for a slot.

What Would Make Me Wrong / Quit Early

For me the usual failure mode is abandoning a cheap, out-of-favor name during underperformance. That risk doesn't even apply here, because I never own this — so there's no thesis for me to lose faith in. The only way I'm "wrong" is if you redefine my method: if someone built a financials-specific screen (different capital metrics, a banking/brokerage-appropriate return measure), HOOD might be assessable — but that is explicitly not my formula, and I won't bend the formula to fit an exciting story. The classic mistake I'm avoiding is over-complicating / over-forecasting a name to talk myself into it.

Discipline Reminder

Holding horizon: N/A — no position. My formula horizon is 3–5 years for names that do qualify; I'd rather sit in cash on this one than improvise metrics.
The underperformance toll I'm signing up for: None here, because I'm not signing up. The toll I refuse to pay is the one that comes from buying a 45x-earnings financial on a narrative and calling it value. The discipline is knowing what my method can't price — and saying so out loud — instead of forcing a rank where the numbers don't behave.

Overall Assessment

Let me be plain, the way I always am. Robinhood may well be a fine, fast-growing business — 27 million funded customers, $307B in platform assets, real operating profit now after years of red ink. None of that is my question. My question is two numbers: is it a good business by return on the capital it actually consumes, and can I buy it cheap relative to the whole enterprise's operating earnings? For a broker-dealer, I can't answer either one honestly, because the balance sheet is mostly other people's money and the metrics that work on a normal company stop working on a financial. That's why I exclude financials by rule — not out of laziness, but because a number I can't trust is worse than no number.

And even if I squinted past the exclusion, what I'd see is a stock at 45x earnings and 18x sales with a profit stream that was hundreds of millions in the red as recently as 2023 and whose crypto line just fell 47% in a quarter. That is the opposite of "good and cheap at the same time." There's no margin of safety, no forced seller handing me a mispriced security, and no one-glance logic — only a list of things I'd have to forecast (PFOF, crypto cycles, rates, four acquisitions) that I have no business forecasting.

So I look down, not up. The downside isn't protected by cheapness or by a clean quality number, and the company doesn't fit either of my two lanes. Pass. Not because Robinhood is bad — because it is outside the circle where my method works, and the discipline is to know that and walk away rather than dress up a story as a strategy.

基于 2026-06-13 共享数据;本分析为单一大师框架的演绎,非投资建议。

Robinhood Markets, Inc.