Mispriced Talent Pools

Talent as alpha, not cost

Eric Glyman frames recruiting through an explicit analogy to baseball's Moneyball. The market for people, in his telling, is the most competitive market on earth, and a startup with few dollars cannot win it head-on. So, like Billy Beane's Oakland Athletics buying undervalued statistics such as on-base percentage instead of well-rounded sluggers, the move is to hunt where a specific capability is underpriced. Talent is treated as alpha rather than cost: find people who create far more value than it costs to pay them, in pools where competition is thin.¹ The framing supplies the "where to look" that complements criteria for what to value.

Ramp's five pools

Glyman enumerates five pools Ramp hunts in.¹ First, aliens with extraordinary abilities: Ramp has sponsored twenty to thirty O-1 and H-1B visas for people other companies "weren't built to hire," and Glyman notes that his co-founder Karim Atiyeh was among the smartest people who could not land internships. Willingness to do the visa work opens an excellent pool with little competition. Second, early talent: meet people before the price is set, as when a candidate joined an earlier company at eighteen for a four-week winter internship, a market a summer later would have priced in fully. Third, black sheep, the neurodivergent pool, developed further in neurodivergent talent strategy. Fourth, high taste: opinionated, decisive people, on the logic that a startup needs a distinct point of view, so hire people who have one, which connects to the argument in taste as moat. Fifth, vintages and lineages done right: "hire from company X, they are great at sales" fails unless you ask which era and which function, and separate the people who built a function from those who merely operate it at scale.

The lineage trap

Glyman's sharpest caution is what he calls a hard negative: a great product hides sales quality. A company whose product sells itself gives no signal about whether its salespeople are good, so a "good sales lineage" from such a firm is a weak signal. The inverse is the strong one. His example is a salesperson who topped the leaderboard at a company that scaled while carrying a deeply negative net promoter score. Selling something customers disliked is proof of a great salesperson; hire the person who can sell a bad product, then hand them a good one.¹

The geographic version

Alex Bouaziz reaches the same arbitrage from a different seat, applying it to geography rather than individual profiles. Building Deel as a fully remote company, he argues that talent in places such as Sao Paulo, Krakow, and Bangalore is both underpriced and stickier, because "you are not competing in the same marketplace" as employers bidding for the same people in a single city.² Remote hiring also reaches pools that local markets quietly discount, and he cites examples such as a highly capable employee with a severe disability and candidates that concentrated hiring pools tend to screen out. The mechanism is identical to Glyman's: the same capability costs less when it is bundled with something the market discounts, whether a visa hassle, a raw or young profile, an atypical resume, or an unglamorous location.

Why it matters, and where it strains

The framework reframes recruiting from competing for the obvious A-players everyone bids on into arbitrage against the market's discounts. It also supplies the supply side to related hiring ideas about which shapes of person to buy and how to price trajectory over present value.

The chief tension, in Glyman's own account, is that these pools are cheap precisely because they are hard to assess. The entire edge is the ability to read signal others cannot, which is why he stresses assessing in domains you genuinely understand. Buy a pool you cannot judge and the discount is deserved. A second caution is bias in the proxies: the trait actually sought is drive, but the markers used to spot it, such as thousands of hours on a game server or a math-olympiad record, are not neutral across who gets to accumulate them.