Microprice — A Better Mid
The midpoint ignores how much size is sitting on each side. Microprice weights the book to track where price is actually going.
The problem with the midpoint
The standard midpoint $$ \text{mid} = \frac{P_{\text{bid}} + P_{\text{ask}}}{2} $$ assumes both sides of the book are equally authoritative. They're not. If the ask holds 100 contracts and the bid only 2, the next trade is far more likely to hit the bid — price is leaning down. Mid misses that entirely.
The microprice formula
A common form weights by book imbalance $I$, defined as:
$$ I = \frac{Q_{\text{bid}}}{Q_{\text{bid}} + Q_{\text{ask}}} $$
where $Q_{\text{bid}}$ and $Q_{\text{ask}}$ are the resting sizes at the best bid and ask. Microprice then shifts the midpoint toward the heavier side:
$$ \text{microprice} = P_{\text{bid}} + I \times \left(P_{\text{ask}} - P_{\text{bid}}\right) $$
When the bid is heavy ($I \to 1$), microprice moves up toward the ask. When the ask is heavy ($I \to 0$), it moves down toward the bid.
How I use it
- As the reference price for z-scores and mean-reversion signals — it's less jittery than last-print.
- To detect when mid is "fake" — a big gap between mid and microprice = a book about to reprice.
Caveats
Microprice reacts to spoofable resting size. On its own it's an estimate of fair value; pair it with actual trades before acting on it.