Market efficiency is not binary.

01 · The mechanism

What happens to the opportunity?

In practice, 102 is an estimate, not a certain future price. Imagine that only a small number of participants initially discover that, under certain conditions, 100 has a tendency to become 102. They buy at 100 and their orders move the price to 100.40. Other participants recognise the same relationship and begin buying earlier too; the price moves to 101, then more capital arrives.

Part of tomorrow’s expected move has already happened. A later buyer may pay 101.80 instead of 100. Keeping the same expected sale price of 102 reduces expected gross profit from 2 to 0.20 per unit. The buyer must deduct costs and separately assess uncertainty in that estimate and the loss they could incur. The future price remains unknown.

02 · Uneven absorption

An efficient market is not an all-knowing market.

It is tempting to imagine “the market” as a single intelligence that sees everything and immediately calculates the correct price. No such mind exists. There are participants with different information, objectives, constraints, capital and time horizons. Some must trade while others can wait; some are reducing risk while others are adding it.

The market price is what emerges when those decisions meet. That process can make markets extraordinarily efficient without making them perfectly efficient. Information does not become price simply because it exists: it has to be observed, interpreted and acted upon.

Recognising an opportunity is not enough to use it. Someone buying a very large quantity may exhaust offers at the initial price and pay more for the rest. Another investor may already have committed their money elsewhere or have rules prohibiting that purchase. A third may decline because they cannot tolerate the possible losses. Understanding the outcome therefore also requires quantities, available funds and accepted risk.

03 · Time and capacity

Not everything happens at once.

Participants may immediately change the prices at which they offer to buy or sell, even before a trade occurs. Other adjustments take longer: selling a position to free cash, buying another asset, or reducing a risk. This illustration does not prescribe the same sequence for every announcement or market.

Asking whether a market is efficient without specifying a horizon is therefore incomplete. A market may be fiercely competitive over one interval while retaining patterns that unfold more slowly over another. An opportunity that disappears in milliseconds is not the same problem as a relationship that develops over several days. Efficiency depends not only on what the market knows, but on how quickly the consequences of that knowledge become price.

04 · Evidence

Edge is probability, not prophecy.

A strategy wins more trades than it loses. To find out whether a profit remains, also check the amounts won and lost and the cost of completing each trade. Consider one hundred invented trades, all for the same amount: 55 gain 1 invented monetary unit and 45 lose 1 unit. The gross total is 55 − 45 = 10. If each trade costs 0.15, total costs are 15 and the net result is −5. More winning than losing trades before costs is not enough to produce a net profit. You also need the amounts gained, amounts lost and trading costs. No predictions are supplied here from which to measure directional accuracy.

An outcome distribution describes observed gains and losses and their frequencies. A researcher can ask whether situations recognisable before a decision have different outcomes from other cases. That does not mean knowing a fact causes those outcomes. The relationship must be checked on later data, including whether trades remain favourable after costs and what losses occur. The hundred invented cases do not establish a real relationship.

Historical patterns are easy to find once events are known. Given enough data, variables and attempts, relationships can look extraordinary and mean nothing. The hard part is discovering whether they persist outside the period in which they were found, remain coherent when assumptions change and continue to exist when decisions are made prospectively rather than reconstructed with hindsight.

05 · Competition and friction

Why would any edge survive?

Competition is powerful, but it is not frictionless. Some advantages are easy to copy and disappear quickly. Others are costly to identify, difficult to execute, dependent on risk tolerance, constrained by capacity or embedded in relationships that keep changing. They may weaken, move, reverse, become crowded or return when competitors leave.

Market efficiency is therefore more useful as a process than as a permanent state. Participants discover opportunities, capital moves towards them, the opportunities shrink and participants adapt. New relationships emerge, and the process begins again.

This creates a useful paradox. If nobody searched for mispricing or predictive structure, markets would have less reason to become efficient. Yet if markets were already perfectly efficient, there would be little incentive to spend resources acquiring information that could never produce an advantage. The search for edge attempts to benefit from inefficiency and, in doing so, helps remove it. That is not a contradiction in market efficiency; it is part of the mechanism that creates it.

07 · A relative advantage

Where the edge goes.

An opportunity is never simply “an edge”. It is an edge for a particular participant, at a particular time and scale, using particular information, with particular costs and constraints. A relationship can be valuable to one participant and useless to another; exploitable with modest capital and impossible with enormous capital; relevant over one horizon and absent over another.

Return to the example: buying at 100 and selling at 102 would earn 2 per unit before costs. If bid and offer prices move so that entry costs 101.80, the same exit price would leave 0.20. The possible gain at those prices has fallen; we have not established that the future price is less predictable. Selling at 102 remains an assumption.

The interesting question is not whether markets are efficient. It is where, how quickly and relative to what information they become efficient.