Why Your Limit Order Might Not Fill: Reading Market Depth

Imagine you're at a busy railway ticket counter. You've picked the right window — the price is good. But there are eleven people ahead of you, and the train leaves in four minutes. Your ticket choice was sound. Your position in the queue was the problem.

That is, roughly, what happens to a limit order in a modern electronic market. And it is one of the most under-appreciated ideas in trading education: the price you post is only half the story. The other half is where you sit in line, and how much volume is standing between you and the next person who wants to trade.

The book, and why it looks like a ladder

Most exchanges today run on something called a limit order book (LOB). Think of it as a live ladder of buy and sell intentions. On the buy side, orders sit at various prices below the current market; on the sell side, they sit above it. Each rung of the ladder is a price level, and each rung carries a certain volume — the total size of orders resting there.

The best bid is the highest price anyone is currently willing to buy at. The best ask is the lowest price anyone is willing to sell at. The gap between them is the spread, and it's a common shorthand for how liquid a market feels right now.

There are three main kinds of order in this system. A limit order posts a price away from the current best and waits — it provides liquidity, and there is no guarantee it ever fills. A market order trades immediately at whatever the best available price is, consuming liquidity. If its size exceeds what's available at the best level, the remainder spills into deeper price levels, further from the top. And a cancellation withdraws or shrinks a resting limit order, usually at no cost.

First in, first out — and why it matters

Execution priority on most modern exchanges follows a first-in, first-out (FIFO) rule. Within a single price level, orders are matched in the order they arrived. This is the queue. Your queue position is simply how much volume is ahead of you at your price.

So when you post a limit buy order, you're not just competing on price. You're joining a line. And when a market sell order arrives, it eats through that line in sequence — oldest first. If the incoming order is smaller than the volume ahead of you, you get nothing. You keep waiting.

This is the slightly humbling arithmetic of passive trading: a perfectly reasonable price can still yield no fill, simply because the queue in front of you was longer than the appetite of whoever showed up to trade.

What "fill probability" actually means

Fill probability is the estimated likelihood that your limit order gets executed within some timeframe you care about. It's not a fixed property of your order — it's a moving estimate that depends on a whole bundle of conditions.

Among the factors that shape it: how far your price sits from the current market, your order size relative to the depth resting at that level, recent trading volume, the time of day, market volatility, and the ongoing churn of the order book itself. A limit buy placed well below the market will generally have a lower chance of filling than one placed close to the best ask. That part is intuitive.

Less intuitive is the trade-off underneath. Traders who want high certainty of execution tend to price more aggressively — meaning they accept a worse price. Traders who want price improvement tend to post passively and accept lower fill probability. You can't maximise both at once. Fill probability models exist mainly to make that tension visible rather than accidental.

The queue is not a fixed thing

Here's where it gets interesting. The order book is not static. Market orders arrive and consume resting volume. Limit orders arrive and extend the queue. Cancellations remove volume — sometimes from ahead of you, which quietly improves your position, and sometimes from anywhere else, which changes the character of the level.

Research on limit order books models this as a stochastic process: order arrivals and cancellations that depend on the current state of the book. In plain terms, the rate at which new orders show up changes depending on what the book already looks like. If there's a large imbalance between bid and ask volume, the flow of incoming orders may shift accordingly. The queue isn't a waiting room where nothing happens. It's more like a bus queue where people keep leaving, arriving, and occasionally wandering off to a different stop.

This is why traditional shortcuts have limits. Simple models have often assumed fill probability decays exponentially with distance from the market price — neat, but blind to shifts in liquidity or volatility. Statistical survival models capture the time dimension better but struggle with the randomness of order flow. Machine learning approaches can handle complexity, though they can behave like black boxes and typically demand large amounts of high-frequency data. More recent state-dependent frameworks aim to capture the evolving nature of the queue directly, using the current state of the book to inform the estimate.

None of these are crystal balls. Filling a limit order remains a probabilistic affair, and any model is an approximation of a system that is genuinely noisy.

FX, crypto, and the same underlying idea

The mechanics described here come largely from equity and futures research, but the concept travels. In foreign exchange, trading is fragmented across multiple venues and liquidity providers rather than concentrated on a single exchange. That changes the practical picture: there may be several order books, different fee structures, and different priority rules. Your "queue position" is not one number but a collection of them, and the visible depth on any one screen may not represent the whole market.

Crypto markets add their own wrinkles. Venues vary widely in how they handle order priority, and some operate differently from the FIFO standard. Depth can thin out quickly during volatile periods, and the depth you see displayed is not always depth you can rely on. There is also the matter of where you are — rules, market structure and venue practices differ by jurisdiction, and this article describes general mechanics rather than any specific market's setup.

One thing worth flagging plainly: liquidity conditions change constantly, and past patterns in order book behaviour do not guarantee future ones. A level that absorbed a large order calmly an hour ago may behave quite differently now. That's not a flaw in the analysis — it's the nature of the thing being analysed.

What the research keeps pointing at

Several findings recur across the academic work in this area. Studies of high-frequency order book forecasting find that a stock's microstructural characteristics — its tick size, its liquidity, how much activity happens at the best levels — shape how predictable its short-term price movements are. Predictability varies by asset, and strong statistical forecasting power does not automatically translate into a usable trading signal.

There's also a consistent observation about evaluation: traditional machine learning metrics don't fully capture what matters in this setting. A model that predicts direction accurately may still be impractical once you account for whether complete transactions actually occur. Some researchers have proposed assessing performance by the probability of correctly forecasting whole transactions instead.

And from the market microstructure literature more broadly, one theme stands out. The cost of trading — both the visible price impact and the subtler frictions — is shaped by the structure of liquidity provision itself. When few participants are willing to provide liquidity, or when holding inventory becomes costly for them, the market's ability to absorb orders changes. Those are structural forces, not daily noise, and they sit underneath every individual fill decision.

What to carry with you

The practical upshot is not a strategy. It's a habit of looking. When you place a limit order, the price is the visible decision. The invisible one is how much volume stands between you and the next trade, and how quickly that volume is likely to change. Watching depth is not the same as watching price — and for anyone who trades passively, it may be the more informative of the two.

The queue will always be a queue. The useful skill is knowing roughly how long the line is, and being honest with yourself that it might not move at all before you change your mind.