Hope Sterling: Michael, hey — okay, how was your weekend, because I spent part of mine genuinely spiraling about something I cannot stop thinking about.
Michael C. Vincent: That tracks. What's the spiral?
Hope Sterling: I was watching someone place a stock order — like, literally just buy some Apple shares — and the price moved before the order even went through. In the time it takes to blink, it shifted. And I kept thinking, wait, who moved it? Who even knew?
Michael C. Vincent: Well. And that's the thing nobody tells you at the open.
Hope Sterling: Right — but it's not just a cost, is it? Like that gap, the bid-ask spread, the highest a buyer will pay versus the lowest a seller accepts — I kept reading about it and I was like, this isn't a toll. This is something weirder. So that's what we're getting into today — what the spread actually is, and why it knows things you don't.
Michael C. Vincent: And here's the scale of it. HFT algorithms — high-frequency trading — now constitute the majority of equity trading volume. Which means price discovery, the whole process of the market turning dispersed beliefs into one number, is happening in microseconds. The spread isn't sitting still. It's breathing.
Hope Sterling: Microseconds — like, not even a fast human reaction, just the time between a thought and a thought.
Michael C. Vincent: Less. And underneath all of it is something called the limit order book — a real-time record of every outstanding buy and sell order, sorted by price and time priority. Every quote you see is pulled from that structure. The spread is just the top of it — the exposed edge.
Hope Sterling: And most people — like most people trading right now — have no idea the spread is doing anything except costing them money.
Michael C. Vincent: That's the question we're sitting with. Is it a cost, or is it a signal — maybe the most information-dense signal in the entire market?
Hope Sterling: Wait — so the spread being a signal, not just a cost — does that mean every time there's a gap, the market is literally saying 'we haven't agreed yet'?
Michael C. Vincent: That's exactly it. And let me set the scene, because the plainest version of this isn't stocks — it's a harbor at dawn.
Hope Sterling: Okay, I'm in. Harbor. Dawn. Go.
Michael C. Vincent: A fisherman wants twelve dollars a pound for his catch. A restaurant buyer shows up willing to pay eleven fifty. That fifty-cent gap — that's the spread. No trade happens. They're both standing there, waiting. The moment one of them moves, a price gets discovered. And now every other seller and buyer on that dock knows something they didn't before.
Hope Sterling: The trade itself is — wait, the trade is the information? Like, just the fact that it happened at eleven seventy-five or whatever tells everyone else where reality is?
Michael C. Vincent: That's the continuous price-discovery machine in one sentence: agreed trades are public announcements. Now — and this is where the limit order book comes alive — the fisherman setting that twelve-dollar price is placing what you'd call a limit order. He's saying, I'll sell, but only if the market reaches me. The restaurant buyer walking in and paying whatever the fisherman accepts? That's a market order. It consumes whatever's sitting there immediately.
Hope Sterling: So limit orders are the patient ones and market orders are like — I need this right now, name your price.
Michael C. Vincent: Exactly. And the limit order book is just — imagine the whole dock. Every fisherman's asking price stacked in one column, every buyer's bid in another, sorted tightest to widest. The spread is the gap between the top of each stack. Electronic, order-driven trading replaced the old system — dealers standing in the middle, quoting both sides themselves — with this centralized structure. That shift made this the structural norm for global markets.
Hope Sterling: Hold on — there used to be, like, an actual person in the middle deciding the price?
Michael C. Vincent: A dealer, yes. A specialist on the floor. They held inventory, they quoted both sides, they were the market. The limit order book essentially automated that role — dispersed it across thousands of resting orders rather than one human judgment. Which is, you see, a profound structural change. The price isn't someone's opinion anymore. It's the aggregate of every outstanding intention.
Hope Sterling: Okay that's — that's actually kind of beautiful? And also terrifying? Because like, if the book is just intentions stacked up, what happens when everyone pulls their intention at the same time? The whole stack just — disappears?
Michael C. Vincent: Now, remember that detail — it matters. Because what happens when the book thins out, when liquidity evaporates, is where the spread stops being a quiet fifty-cent gap and starts doing something much stranger. But that's the mechanism underneath it all — two order types, one book, and every agreed trade broadcasting to the dock.
Hope Sterling: But wait — that stack disappearing is actually what I want to pull on, because I kept reading and the reason the book thins out isn't random. It's the spread doing something almost like — like, it's defending itself? Kyle published this in 1985 and the thing that got me is it's not just about cost, it's about what the market maker can *read* from who's buying.
Michael C. Vincent: Go on.
Hope Sterling: Kyle's whole model is that order flow carries a signal. Like, the *pattern* of who's buying and how much and when — that's information leaking into the market. And the market maker is sitting there reading it. The spread literally reflects the signal-to-noise ratio of that flow. Which means if clean, informed orders start piling in on one side, the spread widens. Not because of anything the trader did wrong, just because — the market maker smells something.
Michael C. Vincent: That's the heart of it. And Glosten and Milgrom the same year — 1985 — made the structure explicit. Three players: informed traders, noise traders, and the market maker in the middle. The spread is exactly the compensation the market maker demands for the risk of trading against someone who knows more than they do. That risk has a name — adverse selection.
Hope Sterling: Adverse selection — wait, that's the one where, like, the very fact someone wants to trade with you is suspicious?
Michael C. Vincent: Precisely. If a hedge fund has an earnings read nobody else has, and they send a buy order — the market maker is now on the wrong side. Every single time. So the market maker widens the spread to price that danger directly into the cost of liquidity.
Hope Sterling: Okay but — no, this is the part that actually made me a little angry — because that means a retail trader, someone who genuinely just wants to buy shares, is paying the wider spread *because* a hedge fund somewhere might know something. Like, the uninformed person is subsidizing the market maker's protection against the informed person.
Michael C. Vincent: That is the uncomfortable math. Exactly as Kyle described it — uninformed traders are the noise in the signal. And without them absorbing losses on the wrong side of spreads, the market maker couldn't afford to quote at all.
Hope Sterling: They're funding the whole operation without knowing it.
Michael C. Vincent: And there's a second layer — inventory risk. A market maker quoting both sides continuously ends up *accumulating* positions. If they've sold a lot, they're holding a short book. Prices move against them. So the spread has to cover that exposure too — adverse selection and inventory risk together determine what the market maker must charge just to remain solvent.
Hope Sterling: So the spread is like — it's doing two jobs at once and we only ever see one of them.
Michael C. Vincent: And now here's where the paradox closes in. When spreads widen to protect against informed traders, uninformed participants face a higher tax. Some of them leave. Order flow imbalance grows. The market maker has *less* noise to hide the signal in — so spreads have to widen further. It's a feedback loop that can hollow out the book.
Hope Sterling: The protection destroys the thing it needs to survive. And honestly — I mean, that's before we even get to dark pools and what price clustering on NYSE and NASDAQ suggests real traders are actually doing, which is going to make this whole model look even messier.
Michael C. Vincent: Well. That it will.
Hope Sterling: Dark pools — individually, you get better prices. But pull back and the whole public signal is getting hollowed out.
Michael C. Vincent: Now, Ye in 2016 mapped exactly how this sorts out. And it's almost elegant in how damaging it is.
Hope Sterling: Wait — sorted how?
Michael C. Vincent: Three tiers of trader. Strong signal — you have real information, you go to the exchange, you want guaranteed execution. Weak signal — you're not confident enough, you stay out entirely. But the moderate-signal traders? They migrate to dark pools. Better price, less certainty. And that middle group is precisely the flow that would sharpen the lit market's price discovery.
Hope Sterling: Oh that's — wait, so the dark pool doesn't steal the dumbest flow or the smartest flow. It steals the middle. The useful middle.
Michael C. Vincent: The calibrating flow. Gone.
Hope Sterling: And every single one of those traders made a perfectly rational choice. Like, no one did anything wrong and somehow collectively they just — degraded the thing everyone depends on.
Michael C. Vincent: Which is why there's no clean policy fix. There's actually an impossibility result buried in the research on privacy-preserving exchange designs — if you give a market maker a coarser signal than the actual order flow, they necessarily transfer welfare to traders. But you cannot simultaneously have that coarseness and achieve efficiency and zero profit. You can't optimize all three. Pick two.
Hope Sterling: That's — I mean, that's not a gap in the theory. That's the theory saying the gap cannot close.
Michael C. Vincent: Fundamental impossibility. Not a calibration problem.
Hope Sterling: Okay but — this is my thing, my actual thing — even if you fixed the dark pool problem, even if everyone traded on NYSE and NASDAQ in the open, the prices themselves are already doing something the models don't account for. Transaction prices cluster at round numbers. Like, fifty dollars. A hundred dollars. Not the theoretically correct price — a psychologically comfortable price.
Michael C. Vincent: The record is murkier than the models want it to be.
Hope Sterling: Right — but it's not just murkier, it's like — wait, if prices are gravitating toward round numbers rather than the informationally efficient price, then price discovery isn't discovering anything? It's just anchoring to whatever feels psychologically legible?
Michael C. Vincent: And that connects directly to what Lu and Abergel found in 2018. Purely Markovian limit order book models — the tractable ones, the ones with closed-form solutions — they fail to reproduce actual participant behavior. You add non-Markovian features, meaning the order book has memory, traders respond to what happened three minutes ago, and suddenly the strategy performance in backtesting improves substantially. But you've lost the clean math.
Hope Sterling: So the model that's useful for thinking is wrong, and the model that's accurate is unusable.
Michael C. Vincent: Permanently. The tractability is the fiction. The messy version is closer to the thing.
Hope Sterling: And meanwhile traders are literally clustering around a hundred dollars because it feels right — like, that's the real market? Humans being humans while the models assume something cleaner?
Michael C. Vincent: You see, that's the uncomfortable core of it. The conversation the market is having — dark pools draining the middle, round numbers substituting for rational prices, models that work until reality walks in — it's failing the way conversations fail. Not catastrophically. Quietly, and by design.
Hope Sterling: I keep — no, actually, the thing I cannot shake is the chain itself. Like, new information comes in, order flow goes imbalanced, price adjusts, equilibrium restores. That's the heartbeat. That works. It's elegant and it genuinely works. And then we just spent an hour finding every way it gets routed around.
Michael C. Vincent: And whether the HFT algorithms — the systems running the majority of that volume now — whether they replicate what the human market maker was doing or quietly hollow it out? That's still an open empirical question. Nobody has closed it.
Hope Sterling: That's — I mean, that genuinely unsettles me? Like, not in a panicked way, just — we don't know. We actually don't know yet.
Michael C. Vincent: Maybe that is the point. Markets are conversations. And conversations don't resolve — they pause. The spread confesses uncertainty, and sometimes it never fully collapses it. That's not a failure of the mechanism. That's what the mechanism actually is.
