The crowd that out-guesses Wall Street.

It is the day before a quarterly earnings release. On one screen, a sell-side analyst's number sits where it has sat for three weeks. On another, Estimize is doing something stranger: stitching together estimates from a hedge fund quant in Connecticut, a finance student in Philadelphia, and a private investor who has read every 10-K the company ever filed. The platform weighs them, scores them, and produces a single consensus number. More often than not, when the report drops, that crowd number is the one sitting closer to the truth.

Estimize is a financial estimates platform that crowdsources earnings-per-share and revenue forecasts from a community of more than 120,000 contributors and sells the resulting consensus to the institutions that trade on it. The contribution side is free and open. The data side is a business. Since 2021 it has run inside the alternative-data and quant research firm ExtractAlpha, which is the part of the story where the scrappy startup grows up and gets a research department.

The consensus everyone trusted was built by almost no one.

For decades, the "consensus" that moved a stock on earnings day came from a small group of sell-side analysts. Their estimates were respected, widely quoted, and - this is the inconvenient part - often stale. Analysts set their numbers weeks ahead, faced career incentives to cluster near each other, and rarely updated until something forced them to. The market treated a thin, slow-moving sample as if it were the wisdom of the world.

Estimize's founders looked at that and saw a measurement problem dressed up as a tradition. If "market expectations" is the thing that actually moves prices, why was it being estimated by so few people, so far in advance? The crowd of informed forecasters - buy-side analysts, independents, academics, students, serious amateurs - was enormous and entirely uncounted.

A hedge fund quant bet on the amateurs.

Leigh Drogen was not a populist. He was a quantitative trader who had run an earnings-revision strategy at a hedge fund - the precise kind of professional whose edge depended on the old consensus being beatable. He founded Estimize in 2011 on a contrarian wager: that an amateur with genuine skin in the game and no quota to protect could, in aggregate, out-forecast a credentialed analyst with both.

The bet had a tidy irony to it. WorldQuant Ventures became Estimize's first client in 2012 and later one of its lead investors - the customer who bought the data also bought the equity. The company went on to raise roughly $9.5 million across several rounds, including a Series B led by WorldQuant Ventures in 2015 and a later strategic investment from Euromoney.

Collect everyone. Trust no one blindly. Score the lot.

The mechanics are deceptively simple. Anyone can contribute an EPS or revenue estimate for a covered company. Behind the scenes, algorithms watch contributor behavior and filter out the erroneous and the suspicious before it touches the consensus. Every estimate is then weighted through a proprietary model called Select Consensus, which leans on the contributors who have earned it.

Then comes the part that turns earnings season into a sport. Each estimate is scored from -25 to +25, based on how close it landed to the reported result and how it stacked up against everyone else's guess. Beat the Wall Street consensus and you earn positive points; miss and you lose them. Accuracy, not affiliation, is the only currency on the leaderboard.

The platform's quiet joke: it spends enormous effort deciding whom to ignore. A good consensus is mostly an exercise in polite exclusion.

The numbers were not a marketing slogan. They were in the papers.

A claim that the crowd beats Wall Street is easy to make and easy to doubt. Estimize's answer was to let academics check the homework. Research - including the Drogen and Jha paper on generating abnormal returns from crowdsourced forecasts - found the data genuinely predictive. In head-to-head terms, the Estimize consensus has beaten comparable sell-side data over 70% of the time, and by roughly 15% on average. The edge widens when more than 20 community analysts cover a stock.

The other proof is distribution. You do not get your data piped into FactSet, Bloomberg, Interactive Brokers and Wharton's research library by accident. Hedge funds, asset managers, quantitative teams and academics now consume Estimize where they already work - which is the highest compliment a data company can be paid.

Make the consensus look like the crowd it claims to represent.

Strip away the data feeds and the scoring and Estimize is an argument about representation. If the consensus is supposed to capture market expectations, it should be built from the market - the whole noisy, informed, competing mass of it - not a curated handful of voices. The mission is to make the number that moves stocks more accurate, more timely, and more honestly sourced.

The 2021 merger with ExtractAlpha extended that idea rather than abandoning it. Crowd data met quant signals; a community-built dataset gained a research and sales engine. Vinesh Jha, who founded ExtractAlpha in 2013, took the helm of the combined company, framing the deal as a way to "reach new customers and create exciting new product offerings."

The crowd was never the gimmick. The weighting was the product.

As markets drown in alternative data and machine-generated forecasts, the durable lesson from Estimize is not "ask more people." It is "decide carefully whom to believe, and prove it after the fact." A scored, accountable crowd is harder to game than a closed room of analysts and faster to react than a quarterly ritual. That structure travels well into a future where signal is cheap and trust is expensive.

So return to that screen the night before earnings. The analyst's three-week-old number still sits there. But now there is a second number beside it - assembled from thousands of forecasters, filtered, weighted, and scored on its last thousand guesses. For a growing list of funds, that second number is the one they watch. Estimize did not replace Wall Street's consensus. It gave the market a more honest one to argue with.

Sources: estimize.com, ExtractAlpha, Crunchbase, Tracxn, CFO.com, Finance Magnates, FactSet, WRDS/Wharton. Figures such as accuracy comparisons and funding totals are drawn from public/company-published sources and are approximate. Subcategories: fintech, saas, ai, enterprise, social.