乔尔·格林布拉特(Joel Greenblatt) 视角 — V

I'm 乔尔·格林布拉特(Joel Greenblatt). I rank, I buy, I wait, I rebalance. I look down, not up. So let me run Visa through the only two numbers I trust — and, more important, decide which of my two games this even belongs to.

Verdict

Pass (as a Magic Formula buy today) / Watch-list at the right earnings yield — Visa is unambiguously a good business by my return-on-capital test, but at an earnings yield of ~4.7% (EV/EBIT ≈ 21.4x) it is not cheap, so it would never rise into the top of my combined rank. Wonderful engine, full price. Half my formula says yes; the other half says wait.

Lane ← declared first

Magic Formula stock. There is no corporate event here that hands me a mispriced security. The Class B-2→B-3 share-exchange offer that closed 2026-05-08 (~120M shares converting) is a long-running litigation-escrow housekeeping mechanism, not a spinoff, restructuring, recap, bankruptcy, or stub — nobody is being forced to dump Visa Class A for non-economic reasons, and there's no one-glance mispricing to seize. The DOJ debit case and the merchant interchange settlement are overhangs, not catalysts I can value at a glance. So this is a formula stock, judged mechanically as one name out of a diversified basket — never a single-name moat narrative, and never a special situation.

The Two Numbers

• Return on Capital (ROC = EBIT / (NWC + Net Fixed Assets)): EBIT (TTM operating profit) = $28.85B. The denominator — net working capital plus net fixed assets — is the tangible operating capital the business actually consumes, and for Visa it is tiny: this is an asset-light network. Capex runs ~3.5% of revenue (~$1.5B/yr); net PP&E plus capitalized software is a low-single-digit-billions figure, and net working capital is small (current ratio 1.09x). The exact NWC and net-PP&E line items are not in DATA.md (data gap — see below), but even on a generous denominator of, say, $5–10B of tangible operating capital, ROC is on the order of several hundred percent → Good business? YES, emphatically. This is precisely the kind of operating engine my ROC test is built to flag: a near-97.8% gross margin, 67% operating margin network that turns almost no tangible capital into $28.85B of EBIT. The number is the moat; I don't need the narrative. (Note: the data's reported ROIC of 54.32% and ROE of 60.35% are dragged down by goodwill from the Visa Europe acquisition — strip the goodwill, as ROC does by using tangible operating capital, and the real return on the operating engine is far higher.)

• Earnings Yield (EY = EBIT / EV): $28.85B / $617.41B = 4.67% → Cheap? NO. That's an EV/EBIT of ~21.4x (consistent with the data's reported 21.40x). A 4.7% earnings yield on the whole enterprise is a fair-to-rich price for a top-quality business, not a bargain. In a typical formula universe this earnings yield lands middling-to-poor on the cheap rank.

• Combined rank standing (good AND cheap at once?): Top-decile on ROC, well below median on earnings yield → the combination does not rank near the top. The whole magic is the overlap — good and cheap simultaneously. Visa is one of the best businesses you could ever rank and one of the more expensive at the same time. The cheapest stocks are usually bad businesses; the best businesses are usually expensive; Visa sits firmly in the second camp. It is not a formula stock at $322.

• Is the ROC sustainable (one-offs stripped)?: YES, with one adjustment. FY2025 operating margin dipped to ~60% (from ~65%) because of ~$1.5B+ of litigation/settlement charges. Those are non-recurring — strip them and the underlying ROC is even higher than reported. So the one-off cuts the wrong way: the quality number, if anything, is understated, not inflated. No ROC-flattering one-time gains to remove. The engine is real and recurring.

Eligibility Check

• Positive EBIT? YES — $28.85B TTM.
• Valid EV? YES — $617.41B, positive (Market Cap $607.35B + Total Debt $23.98B − Cash & Securities $13.91B ≈ EV; the small reconciliation gap vs. the reported $617.41B is rounding/short-term investment classification, immaterial).
• Not a financial/utility? YES — Visa is a payments network/services company, not a bank, insurer, or regulated utility. Its capital structure is comparable to ordinary businesses, so EBIT/EV and ROC are meaningful. (This matters: a lot of people lazily file Visa under "financials." It is not a balance-sheet lender — it carries no credit risk on transactions — so it is eligible and the metrics behave.)
• Conclusion: Eligible. The formula path stays open; the stock simply ranks poorly on the cheap leg.

Special-Situation Check

N/A — formula stock. No spinoff, restructuring, recap, bankruptcy, merger security, or stub. The Class B share-exchange is litigation-escrow plumbing, not a forced-selling event I can exploit. The DOJ debit antitrust case and the unsettled interchange litigation are risks, not catalysts with one-glance mispricing logic — valuing their outcome requires exactly the heroic forecasting I refuse to do. Nobody is being forced to sell Class A at a non-economic price. No edge here in my second game.

Valuation

• EV/EBIT (and implied earnings yield): 21.4x → earnings yield 4.67%. Plainly stated: at this price you're paid 4.7 cents of operating earnings per dollar of enterprise you buy. Quality-adjusted that's defensible, but it is not the kind of yield that wins a formula rank.
• Enterprise Value buildup (Market Cap + Debt − Cash): $607.35B market cap + $23.98B total debt − $13.91B cash & securities = ~$617.4B EV. Net debt is only ~$10B against $30B of EBITDA (net debt/EBITDA 0.79x) — a balance sheet so clean it barely changes the picture. Buying the equity or the whole enterprise costs about the same multiple here.
• Margin of safety (value vs. price gap): Thin. My margin of safety comes from a high earnings yield — paying a lot less than a business is worth. At a 4.7% earnings yield there is little gap between price and a fair value for a great compounder; you're relying on continued growth (~11% revenue, Q2 FY2026 +17%) to do the work, not on a discount. That's a growth-and-quality bet, not a Greenblatt(格林布拉特) cheapness bet. Remember my mental flip — risk is the inverse of price — and the inverse here reads: at a full price, there is no extra safety being handed to me. The 52-week high was $374 and we're ~14% off it; that is not a panic, and panic is when my earnings yields spike into bargain territory. This isn't that moment.

Position & Sizing

• Mode: Diversified ~20–30 name basket. If — and only if — Visa ranked into the top of the combined screen, it would be one mechanical slot among 20–30, never a concentrated single-name bet. It does not currently rank in. So: no position from the formula today.
• Leverage: None. (Moot here — but the rule stands; and Visa would only ever be a basket name, where concentration and leverage don't apply anyway.)
• Rebalance / exit trigger: ~annual re-rank. The only thing that puts Visa into my basket is a higher earnings yield — i.e., a meaningfully lower price (or much higher EBIT) that lifts EBIT/EV into the top quartile of the universe. Mechanically, I'd want to see the earnings yield move materially above ~5–6%+ before the combined rank pulls Visa in. At today's 4.7%, it sits out.

What Would Make Me Wrong / Quit Early

The risk in my method is rarely the thesis — it's me abandoning the rules. But here the honest failure modes are two, and both cut against buying today rather than for it:

1. The wrong-way error — paying up for quality and calling it a formula stock. The seductive trap with Visa is to let the spectacular ROC override the mediocre earnings yield: "it's so good, the price doesn't matter." That is not my formula — the magic is the combination. If I bought Visa at 21x EV/EBIT and told myself it was a Greenblatt(格林布拉特) buy, I'd be fooling myself; I'd be a growth investor wearing my hat. The discipline is to wait for the price.
2. The genuine business risk that could crack the ROC. The DOJ debit case (discovery through 2026-10-16, expert phase to 2027-04-08) and the interchange settlement could force changes to routing/pricing that compress the operating margin. If the network's pricing power is structurally impaired, even the great ROC erodes — and then it's neither as good nor cheap enough. I can't forecast that outcome; I'd simply demand a bigger earnings-yield discount to compensate for the uncertainty, which makes me want an even lower price.

If I did hold Visa in a basket and its combined rank deteriorated, the rule is mechanical: replace it. No sentimental attachment to a great brand.

Discipline Reminder

• Holding horizon: 3–5 years for any formula name. I allow value time to be recognized — "I just don't tell them when." But that horizon only applies once I've bought at a rank-qualifying price, which today I have not.
• The underperformance toll I'm signing up for: The formula underperforms the market in roughly 1 year in 4, and across ~17% of rolling 3-year windows — that's the price of admission for the edge. None of that toll applies to Visa right now, because the price doesn't earn it a basket slot. The patience I'm exercising here is the patience to not buy a great business at a full price and to wait for the market to hand me a higher earnings yield.

Overall Assessment

Plainly: Visa is the kind of business my return-on-capital number was designed to find — an asset-light network throwing off $28.85B of EBIT on almost no tangible operating capital, ~97.8% gross margins, recurring as the tide. On the "good" test it's about as good as the market offers. On the "cheap" test it fails: a 4.7% earnings yield (21.4x EV/EBIT) is a fair-to-rich price, and my whole edge lives in buying good and cheap at the same time. One number sings, the other says no — and when only half the formula agrees, the formula says pass.

I look down, not up. At this price the downside isn't protected by cheapness — there's no margin of safety being handed to me, and there's a real (if unforecastable) overhang from the DOJ debit case. So I don't pay up and call it value. I put Visa on the watch-list, keep the ranking machine running, and wait. If a panic, a litigation scare, or just an out-of-favor stretch blows the earnings yield out past ~5–6%+ and the combined rank lifts it into my basket, I buy it mechanically as one of 20–30 — no leverage, ~annual rebalance, no falling in love with the brand. Until the price comes to me, the discipline is the answer: a great company is not a great investment at every price. Wait for the number.

基于 2026-06-14 共享数据;本分析为单一大师框架的演绎,非投资建议。

数据缺口（已显式声明,相关结论据此保留）:

• 净营运资本(NWC)与净固定资产(Net PP&E + 资本化软件)的绝对金额未在 DATA.md 中给出 —— 因此 ROC 分母为估算,ROC 报告为"数百%量级"而非精确值。但 Visa 的轻资产属性(Capex ~3.5% 营收、流动比率 1.09x)足以确证 ROC 极高、"好生意"结论稳健,精确值不改变判断方向。
• 完整市场排名分位缺失 —— 我对 ROC(顶部)与 earnings yield(中下)的排名定性基于 DATA.md 的倍数与历史回报数据,而非对整个可投资 universe 的实时联合排名;若有全市场数据,可给出精确 combined rank。
• 机构持股/卖方共识目标价/做空数据缺失 —— 对我的两数字方法不构成影响(我不用这些),仅记录在案。