“I love to trade on insider information,” a senior partner told me on my first day at the firm. “It’s the most reliable way to make money in private markets.” He was half joking. But I knew the other half was part of the job description.
Private markets reward people who know things other people don’t. Analysis is mostly a facade: CYA for the LPs and intimidation for the competition. The tower of MBAs looks unbeatable from outside, and from inside the citadel that was part of the marketing. For the founder or operator across the table, the obvious play was to sell rather than compete: these people would outrun you on diligence, outprice you on capital, outhire you on talent. Better to take the check than try to fight. The premium on analysis was real because analysis was expensive and slow since a good research analyst was scarce and a good DCF took a week while a defensible market sizing took two weeks and an expert call you had to pay for.
That world ended sometime in the last eighteen months. It releveled the playing field for operators, founders, and allocators with the right conditions. The same conditions that came with doing their job well over the past 20 years.
Five Decades
The era we’re leaving is about five decades old. KKR was founded in 1976 and the first MBA boom hit shortly after. McKinsey grew from a regional firm into a global consulting franchise across the 1980s and the Bloomberg terminal was launched in 1982. The combination of credentialed analysts, computing tools, and broadly distributed public data made it possible to believe — for the first time in capitalism’s history — that you could simply analyze your way to an edge in private markets.
However, for centuries before that, private market finance ran on conditions.
Nathan Rothschild built a private courier network across the early nineteenth century that systematically gave him news from continental Europe days before London had it. The advantage compounded into one of the great fortunes of the era. JP Morgan personally stopped the Panic of 1907, not because he had a better model of the banking system, but because he could convene the major banks in his library and make their commitments stick. The Medici weren’t better analysts than their banking competitors. They had the better network, both papal and royal, and the franchise was access, not insight.
Even at the peak of this sunsetting era, the people who made the most money still operated on conditions. Warren Buffett bought GEICO because he showed up at the company’s offices on a Saturday in 1951 and a junior employee, Lorimer Davidson, gave him hours of his time. The original KKR deals were sourced through relationships with Drexel and management teams, not through analytical edge over public comparables. The best deals in private equity have consistently been the proprietary ones, sourced through conditions, won before a process formed.
The analytical industrial complex spent five decades selling a story that didn’t entirely match its own results. It just took the model getting cheap enough to expose the gap.
The era we're leaving wasn't the new normal. It was a brief window during which the analytical layer looked like it could substitute for the conditions layer, enabling new institutional scale due to the elegant story of MBAs in towers. It couldn't. It was just expensive enough to look like a moat. What it produced was institutional blandness, that needed to be quietly backfilled by the conditions the analytical layer claimed to have replaced.
The facade is fading as the cost of analysis falls, and the market is going back to the way it always was, an insider’s game.
The Four Conditions
These conditions weren’t dormant the past five decades and they aren’t newly important now; they were always the real work. Even at the peak of the analytical era, the deals that closed were the ones where someone in the room had real conditions on the asset. The model didn’t win the deal, rather: the relationship got you in the room, the context told you what to underwrite, and the analysis was the formality you ran after you’d already decided.
The conditions always mattered, but up until now, they’ve been modeled over. We’re returning to an age where they are all that matter.
Relationships. Who picks up your call. Who returns your text on a Saturday. Who introduces you to the founder before the banker does. Relationships never went away during the analytical era. They quietly kept doing the work while the analytical layer got the credit. The shift now is that the spotlight is gone. Access lives here too: the room you can walk into, the deal you see before it becomes a process, the supplier who quotes you a real number instead of a list price.
Information. What you know that the model doesn’t. The conversation that hasn’t been transcribed. The pattern you saw three jobs ago that nobody else in the room saw. The supplier whisper, the customer churn signal that hasn’t shown up in the data yet, the regulatory rumor from someone two beers in at a conference in Vegas. Information that lives in a person, not a database.
Context. What you’ve already been through. The cycle you survived. The mistake you don’t have to make again because you made it in 2018. The deal that went sideways the way this one is starting to. It’s a library of pattern matches the model can’t borrow. The reason a customer in this niche behaves one way and a customer in the adjacent niche behaves the opposite way, even though the spreadsheet says they should be identical. The model can simulate context. It cannot live in one.
Resources. What you can deploy right now, not in three weeks after committee. Cash, team, attention, time. Capital plus the ability to act fast is a different instrument than capital alone. A check that lands on Tuesday is not the same instrument as a check that lands in February. A team that can do the work alongside management is the most impactful in this new era.
Together they constitute position — where you’re standing on the board. For twenty years the smart money invested in better models. The next decade rewards investing in better positions.
The figure who operated this way before the analytical era arrived had a name. They knew to read the room, knew the terrain, kept his rolodex in his head, moved before consensus formed, and traded on context the institutional players hadn’t seen yet. They were The Gunslinger. Then the railroads came, the federal marshals showed up, the institutions arrived, and the era ended. For about a century.
Prediction Markets Are the Public Proof
The analytical era’s logic is being tested in public right now, and it’s playing out on Polymarket and Kalshi.
I don’t trade these. They’re a casino with a facade of being “better at analysis” bolted on, and I’m not in the business of pricing geopolitical events in two-week windows. But I read them, because they’re the first venue where a generation of analytical assumptions is being stress-tested in front of an audience.
My favorite pitch for prediction markets is the wisdom of crowds: that aggregated expectations from thousands of motivated participants would surface truer signal than expert forecasts. In categories where participants run comparable analysis on comparable data, that’s roughly what happens. The market converges and the prices look efficient.
But in categories where someone has actual information — a relationship inside the campaign, an early read on a regulatory decision, a source in the room — the market is systematically wrong until the asymmetric signal arrives, at which point it snaps. When everyone runs the same model on the same data, common analysis stops being a discovery mechanism. It becomes a coordination mechanism toward the consensus answer.
Being smart the same way everyone else is smart is the definition of having no edge.
In the run-up to the 2024 US presidential election, the consensus was uncertainty. Public polls had Harris and Trump in a statistical tie. Major forecasting models produced odds in the 50-55% range either direction. Polymarket and Kalshi drifted in roughly the same band. The crowd had converged on a single answer: “it’s too close to call”.
Despite this, three weeks before the election, a French former bank trader operating under the name “Théo” hired the polling firm YouGov to run a custom survey across Pennsylvania, Michigan, and Wisconsin. The methodology wasn’t standard. Instead of asking respondents who they intended to vote for, the survey asked them who they thought their neighbors would vote for — a construction designed to bypass social desirability bias, the “shy Trump voter” effect that public pollsters had largely written off as a 2016 anomaly.
When the results came back, Théo described them to the Wall Street Journal as “mind-blowing to the favor of Trump.” He sold most of his liquid assets, scaled his Polymarket positions across eleven anonymous accounts, and ultimately wagered roughly $80 million on Trump — the largest single position in the market’s history. He cleared approximately $85 million on election night.
The structure of the trade is what matters. Théo didn’t beat the market with a better model. Every well-funded participant on Polymarket had access to the same public polls, news, and statistical tools. He beat it with two conditions. Context: he knew the shy-Trump-voter bias the consensus had written off as a 2016 fluke was a live problem, so he knew where the public data was blind. Resources: he was willing to spend his own money to commission a poll that would see into the blind spot. The edge was knowing to source the right private input to run it on.
Now hold that mechanism in your head and look at private markets, because it’s the same one.
A competitive deal process is a prediction market. The banker assembles the data room, every firm runs the same diligence on the same inputs with the same tools, and the valuation that diligence converges on is the consensus price. There’s no ticker, so it may not look like a market. But the clearing bid in a banked auction is the Polymarket line: the answer the shared inputs were always going to produce, rendered three weeks later instead of live. The analytical layer is the consensus price of private markets. It just took a venue with a settlement date to make the mechanism visible.
Which means the edge in private markets is the edge Théo had. Not a better model of the shared inputs — a private input the process never ingested. The proprietary deal seen before the data room exists. The operating partner that has done 2 of these roll ups in the past and knows that market is more seasonal than the materials would imply. The context that tells you the consensus is mispricing the asset. The relationship that gets you the real number instead of the list price. Over the next five years you’re going to watch the Théo dynamic play out across the lower middle market, deal after deal, with gunslingers taking shots that look like luck from outside the room and like arithmetic from inside it, because the position that justified them was never in the data room to begin with.
How to Be a Gunslinger
Most people working in private markets today were trained to be analysts, IB for a couple years, MBA for three, Buy-side after that. Showing up means showing your work: the deck, the memo, the model, the comp set. The seat justifies itself through analytical output.
However, the seat is increasingly rewarding what you do with the company after close: the operating insight, the years of context, the relationships you bring to bear that nobody else can manufacture in a quarter. Four moves to start executing on your conditions:
Take inventory. Most operators and allocators can’t actually list their own conditions. Spend an afternoon writing them down: relationships that pick up your call, information lives in your head, access that other people in your category can’t. What you’ve been through that others haven’t. The list is shorter than you think and more valuable than you’ve been pricing it.
Refuse the consensus trade. If everyone in the room has access to the same model and the model is producing the same answer, you’re not in a gunslinger seat. You’re in the crowd. Pass on those deals. Take the ones where your conditions tell you the consensus is wrong, or where the consensus hasn’t formed yet because the deal hasn’t entered the analytical funnel. This is harder than it sounds. It feels safer to be wrong with the crowd than right alone. Get over it. The analytical era rewarded crowd-rightness. The gunslinger era doesn’t.
Be in the room. Conditions compound through physical and temporal proximity. The deal flow that matters travels through dinners, not data rooms. The information that matters arrives in conversations, not on screens. You can’t build conditions remotely on principle, but you can build them faster in person. Pick the niches and geographies you want to know cold and get embedded. The model will do the analytical work. You do the work that requires a body in a place.
Move when the moment opens. Position is necessary but not sufficient. The gunslinger edge is position plus the ability to fire when the moment opens. Most analytical-era institutions are built to slow that down — committees, processes, diligence cycles, approvals. If you can structure yourself or your firm to act in days rather than weeks when a condition-driven opportunity appears, you’ve built an edge no model can match.
How the Gunslinger Survives
The objection the thesis has to survive is survivorship bias. Théo is a name because he won. The trader who commissioned a similar survey, read it wrong, and lost his stake is a story nobody tells. If you only study the ones who hit, you’ll learn the wrong lesson, and if you can’t tell a gunslinger from a lucky winner, then you’re bound to make the wrong bets.
The answer is that you can tell them apart — but not by looking at the outcome. A win from genuine edge and a win from variance are identical from the outside, and nearly identical from the inside. You can only separate them by looking at the process, before the result lands. The test is whether you could have written down, in advance, exactly why your information was asymmetric and exactly who in the market didn’t have it. Théo could: his neighbor-polling construction bypassed a known, documented bias. That was true on November 4th regardless of what happened on November 5th. The methodology was the condition. The $85 million was the outcome. The gunslinger who confuses the two is already dead.
There are three ways the Gunslinger model fails:
The first death is the phantom edge. You never had a condition. You had a contrarian model and mistook it for information: ran the survey that told you what you wanted to hear, felt out of consensus and assumed that meant you were ahead of it. A contrarian model and a genuine condition produce the identical feeling of seeing what the crowd doesn’t, and identical-looking wins when they pay off. That’s the whole trap. Being wrong alone is not an edge over being wrong with the crowd, it just feels like one. The only defense is to name the asymmetry out loud, in advance, and to notice when you can’t.
The second death is the decayed edge, and it’s the one a thesis about durable conditions has to be honest about since conditions rot quietly. A relationship that’s cooled, a context that’s shifted under you, an information channel that’s gone silent: none of these announce themselves. The most ironic death in this era belongs to the gunslinger who had real conditions, kept trading on them, and never noticed the terrain moved. He isn’t fooling himself and he isn’t over-betting. He’s pricing an expired edge at full value. Conditions are durable, but durable is not permanent, and the work of keeping them live is continuous.
The third death is the over-concentrated edge. Here the condition was real and current — and it still ruined you, because you bet the firm on a single roll. A real edge is an edge in probability, not a guarantee; the dice still roll. The public lesson of Théo — bet everything — is precisely the lesson most likely to put a gunslinger in the ground. He survived. The survivorship bias is that you only hear about the version that did. Size every position so that being wrong is survivable, however strong the read.
Three deaths, one discipline each: name the asymmetry, refresh the conditions, never bet the firm. Notice that only the third is luck, and even the third is mostly sizing. The gunslinger era rewards conditions. It does not forgive the failure to know which kind you actually hold.
The Table is Set
Analysis fell to the floor, the conditions layer it had been quietly standing on came back into view, and the people with real position: relationships, information, access, exposure, context, resources — got the edge handed back to them. The gunslinger isn’t a new figure. He’s the old one, returning to a market that spent fifty years pretending it had outgrown him.
But knowing the era changed is not the same as knowing what to do on Monday. And here the essay has been carrying a quiet oversimplification, because “build your conditions” is not one instruction. It’s three.
The conditions are worth different things depending on which chair you’re sitting in. The operator’s decade of customer intimacy, the allocator’s operating bench, the founder’s read on a market the analysts can’t see from outside — same four conditions, three completely different games. The operator is defending a niche from people who suddenly want it. The allocator is watching the thing he charged a fee for turn into a commodity. The founder is learning that the pitch isn’t the market anymore, it’s where he’s standing in it. Each of them is about to sit down at the same table and discover the other two have changed.
That’s the next piece. Three sides of the table, and what the return of the insider’s game means for each seat, including which one I left, and which one I bet my own next decade on. Next Tuesday.
If you’re an operator trying to figure out which of your conditions are durable, an allocator trying to build the execution bench the next decade rewards, or a founder building something the model can’t see, I’d like to talk. duncan@saorsapartners.com.
Subscribe to Conduit of Value for the ongoing thread. Companion pieces: HumanScale, Jobs Are Dead. Long Live the $10 Million Niche.