Order Books & Market Microstructure

Everything you have learned so far in this institute treated the price as a single number that simply arrives. One closing price per day, one return per bar. In this course we pull that number apart, because at the timescale where execution actually happens, there is no such thing as "the price." There is a queue of people willing to buy and a queue of people willing to sell, and the gap between them is where all the interesting evidence lives.

That structure is the limit order book. On one side sit the bids — resting orders to buy, stacked from the highest price downward. On the other side sit the asks — resting orders to sell, stacked from the lowest price upward. The highest bid and the lowest ask are the best bid and best ask, and the distance between them is the bid-ask spread. Nothing trades inside that gap until someone crosses it. When an impatient buyer sends a market order, it walks up the ask side, consuming resting sell orders until it is filled. The book is, quite literally, the supply-and-demand curve of the asset rendered in real time.

Start with the two numbers everyone quotes. The mid-price is just the average of the best bid and the best ask — add them and divide by two. It is convenient and it is also slightly dishonest, because it pretends the two sides are balanced when they rarely are. Suppose the best bid is one hundred with five hundred contracts resting behind it, and the best ask is one hundred and one with only twenty contracts resting. The mid says one-hundred-point-five. But there is an ocean of buyers and a puddle of sellers. The next trade is far more likely to lift that thin ask than to hit the deep bid, so the true center of gravity sits much closer to one hundred and one than the naive midpoint suggests.

That intuition has a name: the microprice. Instead of weighting the two quotes equally, you weight each side by the size resting on the opposite side. The best bid gets multiplied by the ask size, the best ask gets multiplied by the bid size, and you divide by the total. Why the crossing? Because heavy size on the bid is buying pressure that pushes the fair value up toward the ask, not down. In our example the microprice lands near one hundred and ninety-six hundredths of the way to the ask — it tells you the book is leaning hard to the upside, something the mid-price flatly concealed. The microprice is the first lesson of this course in miniature: the same event, read with structure, becomes evidence; read without it, becomes noise.

Now let us turn that imbalance into a proper signal. Define order-flow imbalance as the bid size minus the ask size, divided by their sum. It runs from minus one, when the book is all sellers, to plus one, when the book is all buyers, and sits at zero when the two sides match. This single ratio is one of the most robustly documented short-horizon predictors in all of finance. When imbalance is strongly positive, the next mid-price move is, on average, up — not always, but often enough and consistently enough to trade on. You are no longer guessing at sentiment from a chart. You are reading, directly, which side of the queue is exhausting first.

But predicting direction is only half of execution. You also need to know cost — how far the price moves when you yourself trade. That is the deepest result in this chapter, and it comes from Kyle. In the Kyle model, the price the market makers quote moves linearly in the net order flow, and the slope of that line is a single coefficient, written as the Greek letter lambda. Price change equals lambda times net order flow. Lambda is price impact per unit of size — the steepness of the book's response. A small lambda means a liquid, resilient market where you can push size through and barely disturb the price. A large lambda means a thin, fragile market where even modest orders move the quote violently.

Lambda is also, and this is the part that ties back to the throughline of this entire institute, an information measure. Kyle derived it from the presence of informed traders: the market maker widens lambda precisely when he suspects the flow hitting his book is informed rather than random. So a price move accompanied by a low lambda — a move that happened on a deep, resilient book that barely flinched — is weak evidence. A price move that came with a high lambda, where small flow shifted the quote sharply, is the market itself confessing that someone with information is in the order flow. When you judge whether a number is genuine evidence an asset will move, lambda is the lens that separates a real informational push from a few contracts rattling around an empty book.