The Penny Empire:
How Market Makers Turn Fractions
of a Cent into Billions

The Invisible Machine Behind Every Trade

Every time you buy a share of Apple, sell a position in an ETF, or execute any transaction in any financial market anywhere in the world, something happens in the milliseconds before your order is filled that most investors never think about. A firm — one you have almost certainly never heard of — stepped between you and the market, offered to take the other side of your trade, and earned a fraction of a cent for doing so. By the time you checked your brokerage confirmation, that firm had already done it millions of times again with someone else.

These firms are called market makers. They are not traders in the way most people understand the term. They are not betting that Apple goes up or down. They are not analysts forecasting earnings. They are, in the most precise sense, the plumbing of modern financial markets — the entities that ensure there is always a buyer when you want to sell and always a seller when you want to buy, at a price that is fair and continuously updated. In exchange for that service, they collect a spread: the difference between the price at which they will buy an asset and the price at which they will sell it. That spread, on a heavily traded stock, might be $0.01 — one single cent. It might be even less.

One cent, repeated across ten billion shares of daily US equity trading volume, is $100 million. Per day. That arithmetic is the foundation of some of the most profitable businesses in the history of finance — businesses that most people have never heard of, operating in a layer of the market that is entirely invisible to the average investor, generating returns that dwarf those of most hedge funds and investment banks combined.

Why Markets Need Them: The Liquidity Problem

To understand why market makers exist, you first need to understand a problem that markets face without them. Imagine you own 500 shares of a mid-sized company and want to sell immediately. In a world without market makers, you would need to find a willing buyer at exactly the moment you want to sell, for exactly the quantity you want to sell, at a price you both agree on. This is called a "double coincidence of wants" — the same problem that made barter economies inefficient before currency was invented. In illiquid markets, this problem is very real. Bid and ask prices diverge wildly. Transactions become slow, expensive, and uncertain.

Market makers solve this problem by standing ready to trade at all times. They post a bid price — what they will pay to buy — and an ask price — what they will accept to sell — simultaneously, continuously, in every market they operate in. The difference between those two prices is the spread, and it is how they are compensated for providing that service. Think of them as a currency exchange booth at an airport: always open, always quoting both sides, earning the margin between what they pay and what they charge. The traveler gets immediate liquidity. The booth earns the spread. Nobody thinks about it. The system works.

The spread also compensates market makers for a risk called inventory risk. When a market maker buys your shares, those shares sit on their books until they can sell them to someone else. During that time, the price might move against them. If they buy 10,000 shares of a stock at $50 and the price drops to $49.80 before they can sell, they have lost $2,000 on that position. Their models must constantly estimate how likely that is, how long they are likely to hold inventory, and how wide the spread needs to be to compensate for that exposure across millions of simultaneous positions. The accuracy of those models — their ability to price inventory risk correctly in real time — is the heart of what separates a great market maker from a failed one.

From the Trading Floor to the Algorithm: A Brief History

The role of market maker is not new. What is new is who fills it and how they operate. For most of the twentieth century, market making in US equities was the province of "specialists" — human beings assigned to specific stocks on the floor of the New York Stock Exchange, whose job was to maintain orderly markets in those securities. A specialist in IBM was responsible for ensuring that there was always a bid and ask for IBM, stepping in with their own capital when buy and sell orders were imbalanced. It was a privileged, profitable position — specialists earned spreads and had advance knowledge of pending order flow that modern regulators would not permit.

The shift began in the 1990s with the rise of electronic communication networks (ECNs) and the SEC's push to open markets to competition. The defining regulatory moment was decimalization in 2001, when US equities moved from fractional pricing — stocks quoted in eighths of a dollar, meaning minimum spreads of 12.5 cents — to decimal pricing, where spreads could narrow to a single penny.¹ This destroyed the specialist model overnight. A 12.5-cent spread on a million shares is $125,000 in daily revenue for a single stock. A one-cent spread on the same volume is $10,000. Human specialists could not survive on that margin. Algorithms could.

The firms that emerged — Citadel Securities, Virtu Financial, Jane Street, Hudson River Trading, Two Sigma Securities, IMC, and others — were built from the ground up around computers, mathematicians, and engineers rather than floor traders and relationships. They did not inherit their position in the market. They engineered it. And in doing so, they created a new kind of financial institution: one that makes money not by predicting the future but by processing the present, faster and more accurately than anyone else.

The Spread: A Penny, Repeated a Billion Times

The economics of market making are counterintuitive to most people because they operate at a scale that is difficult to visualize. The individual transaction — earning $0.002 on a 100-share order — sounds trivial. The aggregate — repeating that transaction across billions of shares per day across hundreds of securities and dozens of markets — is what makes it extraordinary. Let the math speak plainly.

The comparison that makes this intuitive is a toll road. A toll road does not care whether you are driving to a funeral or a vacation. It does not have a view on whether your trip will be worthwhile. It charges a small fee for every vehicle that passes through, operates continuously, and collects that fee regardless of which direction the car is heading. A market maker is a toll road built on the flow of capital — earning a few cents per transaction whether the market is rising or falling, whether the buyer is right or wrong, whether the news is good or bad. The revenue is not a function of market direction. It is a function of volume.

The Speed Arms Race: Fiber, Microwaves, and Colocation

If the spread is the revenue, speed is the moat. In market making, the firm that knows the current price most accurately and can update its quotes fastest holds a structural advantage over every competitor. An outdated quote is a liability — it invites other market participants to trade against you at a stale price, a form of loss known as being "picked off." The firm that quotes last is the firm that loses most.

The resulting arms race in market infrastructure is one of the most extraordinary capital expenditure stories in modern business. To connect the Chicago Mercantile Exchange and the New York Stock Exchange — a distance of roughly 700 miles — firms initially used the fastest fiber optic cable available, achieving round-trip transmission times of approximately 13 milliseconds. Then they switched to microwave towers, which transmit data at the speed of light through the air rather than through glass fiber, cutting the latency to approximately 8 milliseconds. Then laser links. Then millimeter-wave radio towers spaced more precisely to optimize line-of-sight transmission. Each incremental improvement — each shaved millisecond — cost tens of millions of dollars in infrastructure. Each was worth the investment because the competitive edge it provided was measured in billions of dollars in annual advantage over slightly slower rivals.²

Exchange colocation takes the same logic inside the data center. Major exchanges — the NYSE, Nasdaq, CBOE, and others — rent space in their server rooms to market-making firms, allowing those firms to place their trading servers physically adjacent to the exchange matching engines. Instead of a signal traveling from a firm's office across the city, it travels a few hundred feet of cable. The time savings is measured in microseconds — millionths of a second. The cost of colocation space at a major exchange can run hundreds of thousands of dollars per year per rack. The demand for that space has never declined.

How the Algorithm Actually Works

The image most people carry of algorithmic trading involves a single automated program placing buy and sell orders. The reality of a modern market-making algorithm is considerably more complex — and considerably more impressive. A market maker operating in US equities is simultaneously quoting prices in thousands of securities, updating those quotes thousands of times per second, managing an inventory that can swing by millions of dollars in seconds, and doing all of this while continuously re-estimating the fair value of each position based on data arriving from dozens of sources simultaneously.

The core function is a real-time fair value model. For any given stock at any given moment, the algorithm maintains an estimate of what that stock is worth, derived from the last traded price, the order book (the visible queue of pending buy and sell orders), recent transaction data, futures prices, correlated securities, index movements, and a variety of other signals. That estimate is updated continuously — sometimes thousands of times per second in volatile conditions. The bid and ask prices the firm posts are functions of that fair value estimate, adjusted for the current inventory position and the estimated risk of holding it.

If a firm has accumulated a large long position in a stock — it has bought more than it has sold — the algorithm shifts its quotes slightly lower to encourage more sellers and fewer buyers, reducing the inventory imbalance. If it has a short position, the opposite. The quotes are not static prices. They are dynamic outputs of a continuous risk management process, re-evaluated millions of times per day by systems that hold no human emotion, no attachment to a position, and no delay between observation and response.

Payment for Order Flow: Why Your Retail Trades Are Valuable

One of the most debated practices in modern market structure is Payment for Order Flow — PFOF. The mechanics are simple. When you place a trade through Robinhood, Schwab, Fidelity, or most retail brokers, your order does not go directly to a public exchange. Instead, the broker sends it to a market maker — Citadel Securities, Virtu, or a similar firm — which executes the trade directly against its own inventory. In exchange for receiving that flow of retail orders, the market maker pays the broker a small fee: fractions of a cent per share.

The arrangement sounds suspicious on the surface. Your broker is getting paid to send your trades to a specific firm. How could that possibly be in your interest? The answer lies in understanding why retail order flow is valuable to market makers — and what they have to offer in exchange.

Retail traders, as a group, are not systematically better-informed than the market. They are not trading on insider information or sophisticated signals about short-term price direction. This makes their order flow relatively safe for a market maker to handle — the risk of being "picked off" by a more informed counterparty is lower than with institutional flow. Because the market maker faces less adverse selection risk, it can offer retail traders a price that is better than the publicly quoted bid or ask — what the industry calls "price improvement." Studies have consistently found that retail investors executing through PFOF internalization receive prices that are, on average, better than the national best bid and offer that they would have received on a public exchange.³

The critique of PFOF is not that retail investors receive worse prices — they generally do not. The critique is that the system directs retail order flow away from public exchanges, reducing price discovery, and that brokers' incentives may not be perfectly aligned with getting the absolute best possible execution. It is a legitimate structural debate. But it is a different debate than the one most retail investors think they are having.

The Hidden Risks: What Can Actually Go Wrong

The consistency of market-making profits — Virtu Financial's public filing showing 1,237 profitable trading days out of 1,238 — creates an impression of near-certain profitability. That impression is misleading. Market making carries genuine, asymmetric risk that can materialize suddenly and devastatingly. The profitability record exists not because the risks are small, but because the risk management is exceptional.

Volatility spikes and inventory risk. During periods of extreme volatility — a flash crash, a major earnings surprise, a geopolitical shock — bid-ask spreads widen dramatically because uncertainty about fair value increases. A market maker holding inventory when a stock drops 20% in seconds faces catastrophic losses if its models have not already hedged or reduced the position. The 2010 Flash Crash, when the Dow Jones Industrial Average fell nearly 1,000 points in minutes, temporarily disrupted the automated quoting of multiple market makers who pulled their bids rather than provide liquidity into a falling market — a rational self-protective response that paradoxically amplified the crash.⁴

Technology failures. An algorithm that quotes prices without working correctly is an open invitation to losses. Knight Capital's August 2012 failure — in which a software deployment error caused its trading system to flood markets with erroneous orders, losing $440 million in 45 minutes — remains the definitive case study in what happens when market-making infrastructure fails catastrophically.⁵ The firm, then one of the largest market makers in US equities, was forced to seek emergency capital and eventually was acquired. Four decades of business destroyed by 45 minutes of broken code.

Adverse selection. When an institutional investor with superior information — a fund that has just analyzed private satellite data showing deteriorating sales at a retailer, for example — trades against a market maker's quotes, the market maker is on the losing side of an informed trade. Over millions of transactions, the statistical edge belongs to the market maker. In any individual high-stakes trade, the edge can belong to the counterparty. The art of market making includes learning, in real time, to distinguish informed flow from uninformed flow and adjust quotes accordingly.

Beyond Equities: The Global Expansion of the Penny Empire

The firms that began as equity market makers have not stayed in equities. The logic of the business — quantitative models, low-latency infrastructure, risk management at scale — is portable across any liquid market where a spread can be earned. And the firms have followed that logic aggressively into every corner of global finance.

Jane Street is perhaps the most striking example. Founded in 2000 as an equity market maker, the firm has become one of the dominant liquidity providers in exchange-traded funds globally — a market where the ability to simultaneously hedge a basket of underlying securities while quoting a price for the fund itself requires exactly the kind of multi-asset, real-time risk management that market makers have spent decades perfecting. Jane Street's ETF business is so significant that it is estimated to account for a substantial share of global ETF trading volume on any given day. The firm has expanded into options, fixed income, foreign exchange, commodities, and cryptocurrencies. Its 2023 revenue of $21 billion was larger than that of Goldman Sachs's trading division.⁶

Citadel Securities, majority-owned by Ken Griffin, operates across equities, options, rates, and foreign exchange across more than 50 countries. It is the largest market maker in US equity options and one of the largest in US Treasuries. Hudson River Trading, founded in 2002 by a group of PhDs and mathematicians, now operates in asset classes and geographies that would have been unimaginable for its original mandate. The pattern is consistent: build the infrastructure, prove the model, expand the mandate.

The expansion into cryptocurrency markets has been particularly notable. Digital assets, with their 24-hour trading cycles, fragmented liquidity across dozens of exchanges, high volatility, and persistent inefficiencies, represent an environment where quantitative market makers have a structural edge over less sophisticated participants. The same firms that dominate US equity markets have quietly become among the largest liquidity providers in Bitcoin and Ethereum markets as well.

Artificial Intelligence: The Next Competitive Frontier

Market making has always been an engineering discipline. The transition from floor specialists to algorithmic systems was a technology transformation. The emerging transition — from rule-based algorithms to machine learning systems — is the next one, and it is already underway.

Traditional market-making algorithms are explicitly programmed: if inventory exceeds threshold X, adjust quotes by Y. Machine learning models, particularly reinforcement learning systems, are trained on historical data to discover optimal quoting strategies without being told what those strategies should look like. They learn from billions of past trading scenarios which quote adjustments produced the best outcomes, and they apply those learned patterns to current conditions in ways that no human programmer would have explicitly prescribed.

The potential advantages are substantial. ML models may identify non-obvious correlations between market conditions and inventory risk that deterministic algorithms miss. They may adapt more quickly to changing market regimes. They may find edges in the relationship between different asset classes that hand-coded systems cannot exploit. The leading market makers are investing heavily in machine learning research — Jane Street, Citadel Securities, and their peers compete aggressively with technology companies for machine learning talent, offering compensation packages that routinely exceed $500,000 for quantitative researchers. The arms race that began with microwave towers is entering a new phase in the neural network.

The Poker Table: Who Is Actually Sitting Across From You

There is a saying in poker that has been attributed to various players over the decades but is universally accepted as wisdom: if you cannot identify the sucker at the table after thirty minutes, you are the sucker. The observation is not cruel. It is instructive. It is a reminder that poker is not a game played against the house — it is a game played against other people, and those other people have varying levels of skill, information, and experience. The way to win is not to be the best player in the world. It is to be better than the people sitting across from you.

Every financial market transaction has a counterparty. When you buy, someone is selling to you. When you sell, someone is buying from you. In the short term — the minutes and hours that define day trading — the question of who that counterparty is, and what they know, and what their infrastructure looks like, matters enormously.

Consider what you are sitting across from when you day trade a liquid stock. Your counterparty has, with high probability, spent years and billions of dollars building a system that processes more data per second than you will read in a lifetime. It has observed billions of past transactions and learned statistical patterns from each one. Its quote is not a guess about fair value — it is a continuously updated, multi-variable estimate refined across every piece of relevant market data available. It is faster than you by a margin that is literally incomprehensible in human terms. It is more diversified across scenarios than any individual investor could achieve. And it does not make decisions based on emotion, conviction, or fear.

This is not an argument that markets are rigged. It is an argument that different market participants are playing fundamentally different games — and that some of those games are designed for opponents whose strengths you do not share and cannot replicate. The day trader competing with Jane Street over the next five minutes is, in a very real sense, sitting at a poker table they did not choose against players who have been studying the hands for decades.

But here is what changes entirely when you extend your horizon from minutes to years: the nature of the competition. A market maker's edge operates in the seconds and milliseconds around a transaction. It is indifferent to whether the underlying business is a great one or a mediocre one, whether the CEO is extraordinary or incompetent, whether the product is a durable competitive advantage or a commoditized offering. Those are not the variables the algorithm is optimizing for. They are not the signals it reads. They are entirely outside the game it is playing.

The long-term investor is playing a different game altogether — one where the counterparty's microsecond advantage is irrelevant, where the competitive edge available to any patient individual is not speed or sophisticated models but the willingness to hold a productive asset through periods of volatility that would force a short-term participant to liquidate. The greatest investors of the modern era — Buffett, Munger, Lynch — did not beat the market by being faster. They beat it by being more patient, more disciplined, and more willing to wait for the logic of business value to assert itself over the noise of short-term price movements.

The Last Unfair Advantage

Perhaps the greatest lesson of market making is not how impressive the firms have become. It is the clarity it provides about where average investors actually have an edge — and where they do not.

You cannot outrun a microwave tower. You cannot outcompute a system running on custom chips in a colocation rack twenty feet from the matching engine. You cannot outmodel a team of PhDs who have trained their systems on decades of tick data from every exchange on the planet. In the game of microseconds and pennies, the infrastructure gap is not a challenge to overcome. It is a structural reality.

But that game is not the only game available to you. Every time you trade on impulse — reacting to a headline, chasing a move, closing a position because the market made you anxious — you are electing to play a game designed for someone else's strengths. You are stepping into the casino as a recreational player when the house has a mathematically defined edge, and you are paying the spread each time you enter or exit.

By extending your horizon from minutes to years, you change the rules entirely. The market transforms from a poker table — a zero-sum competition against well-armed opponents — into a partnership with the productive businesses you own. The expected value of that partnership, compounded over time, has historically been positive. The S&P 500 has returned approximately 10% annually over the last century, through wars, depressions, panics, and crises. Not because the market makers got it right on any given day, but because the underlying businesses they were quoting prices on were, in aggregate, creating value over time.

The greatest edge available to most investors is not an algorithm. It is not a data feed. It is not colocation or microwave infrastructure. It is the capacity to hold a well-reasoned conviction through periods of noise that would force a short-term trader to fold. In a world obsessed with milliseconds, time itself remains the last unfair advantage available to everyone.

The penny empire is real. It is enormous. It is quietly embedded in every transaction in every market in the world. And the wisest response to understanding it is not to try to beat it on its own terms — but to recognize the game it is playing, step off that board, and remember that the most powerful force in investing has always been compound interest and the patience to let it work.