The uninformed trader loses, whichever way they trade

Market microstructure is the field that studies how prices form and how orders get executed. Its standard reference, by Larry Harris, closes its treatment of spreads and market making with a conclusion that admits no softening: uninformed traders lose whichever way they trade; if they want to avoid losing, they must avoid trading.

The term deserves defining before anyone reacts to it. Uninformed does not mean ignorant, careless or inexperienced. In its technical sense it means holding no information about the value of an asset that the rest of the market has not already priced in. Someone can have read the entire quarterly filing and still be uninformed, because thousands of people read it first and the price already reflects it.

An uninformed trader who believes they are informed is any professional desk's favourite counterparty. Not out of foolishness — out of arithmetic. They are the one who pays the toll again and again, convinced they are on the way to a gain.

The proof is short because there are only two ways to trade. You supply liquidity, posting a limit order and waiting; or you take liquidity, crossing the spread with a market order. There is no third. Both lose, for different reasons.

That sounds like an advantage, and in pure cost terms it is. The problem appears when you ask who decides whether the order fills. Not the person who posted it: whoever arrives next. A limit order is, technically, a free option handed to the market — the right, not the obligation, to trade against you at the price you set.

Anyone holding information will exercise it only when it suits them. That leaves exactly two branches, and both are bad:

• The price moves away and the order never fills. You miss a move that did happen. No accounting loss, but the opportunity cost is real.
• The order fills instantly. That looks like the good outcome until you ask why: it filled because someone who knew more wanted to trade against it at that price. And there you would rather not have traded.

This is called adverse selection: your orders execute preferentially at the worst possible moment, because the person executing them chooses when. It is not repeated bad luck. It is the expected result of giving somebody else the right to choose.

The apparent way out is to cross the spread: immediate execution, no more letting someone else choose for you. And it is true that this avoids direct adverse selection. But you pay the bid-ask spread, and that spread is not what it appears to be.

The spread has two halves, and only one of them comes back

The second component is worth reading twice. A market maker knows it will systematically lose to informed traders, and cannot prevent it, because it cannot tell which of the orders arriving comes from someone who knows. Its defence is to widen the spread for everyone. Which means: it recovers from the uninformed exactly what it loses to the informed.

Without anaesthetic: part of what you pay when you cross the spread is a fee charged to you for risk that other people create. You lose to informed traders even if you never trade with them directly, through the intermediation of whoever sets the price.

One detail gives confidence that the model describes something real rather than being an elegant construction. That component can be computed along two independent routes: an informational one, asking how much the market maker revises its value estimate depending on whether the next arrival buys or sells; and an accounting one, asking how much it must charge the uninformed to offset what it expects to lose. Both routes give exactly the same number. That result is known as the Glosten-Milgrom theorem.

The immediate objection is reasonable: in the most liquid instruments the spread is one cent. How is a cent going to ruin anyone?

Because the argument does not depend on the size of the spread but on which side of the information you are. A narrow spread does not change the sign of the mathematical expectation: it changes its speed. You lose more slowly.

And there is a consequence that runs against almost everyone's intuition. If trading without an edge carries negative expectation, then every round trip applies that negative expectation once more. Trading faster does not accelerate gains: it accelerates the outcome, which is the opposite one. Very short-term trading is, mathematically, the most expensive way to discover you had no edge.

Bid-ask bounce also explains something that confuses a lot of people: on a one-minute chart, part of what looks like movement is simply the price bouncing between the two sides of the spread, with nothing about value having changed. You can be trading microstructure noise while convinced you are reading a signal.

Harris's sentence sounds like a verdict, but it never says never trade. It says something far more precise: if you are uninformed, trading has negative mathematical expectation. From that, exactly three paths follow, and there is no fourth.

1. Do not trade, and accept the market's return. Buy a broad index and sit still. It beats most independent traders, precisely because it does not pay spreads and adverse selection over and over again.
2. Trade for a reason other than beating the market. Someone moving money into the future, or hedging a real risk, accepts losing a little in exchange for a different benefit. That is not irrational: it is a conscious trade-off. The problem was never losing a little; it is losing while believing you are going to win.
3. Stop being uninformed — which does not mean guessing better. It means holding a structural edge that is measurable and explainable in words. It is the hardest of the three and the only one that justifies the toll.

Before every trade: “I have an edge here because ______. The other side loses or accepts losing because ______. My round-trip cost is ______ and my target beats it because ______.” If any blank stays empty, there is no trade — there is a bet. Thirty seconds, and it filters more than any indicator.

The third path is not walked with more conviction: it is walked with better instruments. If the cost of trading is real and the edge has to be demonstrated, then a useful tool is one that helps you measure both — not one that helps you feel certain.

• It shows the formula as well as the number. A figure with no method behind it is not a calculation: it is an opinion with decimals.
• It subtracts realistic costs from any historical result, spread included. Many strategies are profitable before costs and losing after them.
• It distinguishes the transitory from the permanent when estimating probabilities, instead of treating every move as noise that will revert.
• It leaves a visible gap when the data is missing, instead of filling it with a comfortable estimate. A gap gets investigated; an invented number propagates.
• It issues no buy or sell verdicts. The one who has to complete that sentence is you, not the software.

It is only fair to declare our own position, since that is exactly what this report demands of everyone else. Quantika Research sells software by subscription. It does not hold client funds, does not execute orders, is never anyone's counterparty and does not charge by traded volume. None of our revenue improves if you trade more. That structure is why we can publish a text whose first practical conclusion is that many people should trade less.