What does it mean when the micro-price line sits near the ask?

It means the bid queue is much larger than the ask queue, so the imbalance-weighted fair value has shifted toward the top of the spread. Mechanically, the thin ask queue is the one likely to be consumed first — the book is leaning toward an uptick. The mid-price cannot show you this because it ignores the sizes entirely.

Is queue imbalance actually predictive?

At the one-tick horizon it carries real signal: on large-tick instruments, a classifier built on queue imbalance alone reaches an out-of-sample AUC of roughly 0.75–0.80, and when the queues are extremely lopsided the next mid-price move goes with the heavy side around 80–90% of the time. But that is a mechanical, one-tick effect: after queue position, latency and costs, it is not a tradeable edge on its own. Use it as a state read at your levels, not as an entry signal.

What is the difference between queue imbalance and OFI?

Queue imbalance is a snapshot: the balance of resting size at the best bid and best ask right now, from −1 to +1. OFI is a flow: how much the book has shifted — orders added, pulled, consumed — accumulated over recent updates, across multiple levels. QI tells you the current lean; OFI tells you the direction the book is being pushed.

Why does the micro-price line only cover the last few minutes?

The line is a trail, 180 seconds by default. The micro-price is a high-frequency object — it reacts to every book update and is meaningful at the scale of the current rotation, not the session. A longer trail is available in the settings (Trail seconds), but stretching it too far just draws noise over your candles.

Does the indicator work on every instrument?

It computes on anything with a live level-2 book, but the signals are strongest on large-tick instruments — markets where the spread sits at one tick almost all the time. ES is the canonical case, with queues hundreds deep; MNQ shares the pinned spread but with much shallower queues. On small-tick, fast-cycling books the queue signals carry much less information, which matches the published research.

The Microstructure Indicator: Micro-Price, Queue Imbalance & OFI Explained

One indicator, three reads of the same order book: the micro-price (where fair value actually sits inside the spread), the queue imbalance (which side of the book is heavier right now) and OFI (how the whole book is reshaping). This guide explains what each one measures, how to read the line and the gauge on the chart, and what every setting does.

Senzoukria · Learn · Updated July 2026

Most order-flow tools read the trades — the footprint, delta, CVD. The Microstructure indicator reads the book: the resting orders sitting on the DOM before anything trades. It compresses the book into three numbers, drawn as one line on the chart and one two-value gauge. This page goes through each of them, in the order you will use them.

Read 1 — the micro-price: fair value inside the spread

The mid-price — halfway between best bid and best ask — is what most platforms call “the price”. It is also a small lie, because it ignores the sizes. Take a concrete MNQ book: 40 contracts bid at 29150.00, 10 contracts offered at 29150.25. The mid says 29150.125, dead center. But the queues are 4-to-1: the thin ask is far more likely to be consumed than the thick bid — the next move is more likely up than down. The fair value is not in the center of the spread; it is near the ask.

The micro-price makes that exact correction. Compute the imbalance I = bid size / (bid size + ask size) — 0.8 in the example — then slide fair value that fraction of the way across the spread: micro = bid + I × spread, here 29150.20. This is the imbalance-weighted mid, the practical first-order form of the micro-price estimator studied by Stoikov. Two properties make it useful: it moves before the trade prints (queues shift first), and it lives inside the spread, so it is always finer-grained than the last traded price.

Reading the line on the chart

On the chart, the micro-price is drawn as a thin trail over the last three minutes (configurable). How to read it:

Line hugging the ask side of the spread — persistent bid-heavy book, upward lean. Hugging the bid side: the mirror.
Line crossing the mid repeatedly — balanced, two-sided book; queue signals carry little information at that moment.
Line leading price — the most useful pattern: micro-price drifting up while the last trade is still flat means the book is repositioning before the tape confirms. It resolves at the one-tick scale, so treat it as timing refinement at a level, not as a directional call on the next ten points.

Read 2 — queue imbalance: the gauge’s first number

Queue imbalance (QI) is the same information as the micro-price, expressed as a single signed number: QI = (bid size − ask size) / (bid size + ask size), from −1 (all size on the ask) to +1 (all size on the bid). In the example above, QI = (40−10)/(40+10) = +0.6.

QI is the one book signal with genuine published predictive content: on large-tick instruments, a classifier built on QI alone calls the direction of the next mid-price move with an out-of-sample AUC of roughly 0.75–0.80 (Gould & Bonart, 2016), and when the queues are extremely lopsided the next move goes with the heavy side around 80–90% of the time. The measured version of those numbers — what an AUC actually is, and why the effect is mostly mechanics — is in the companion field note: micro-price vs mid-price. Before they excite you, read the honest-limits section below — one tick of predictability is not an edge. What QI is, is a reliable read of which way the book is leaning right now.

Read 3 — OFI: the gauge’s second number

QI is a snapshot; OFI is the flow. It accumulates every change in the book — bids added or pulled, asks stepping in or backing away — across the top ten levels, into one signed pressure number, normalized by book depth so readings are comparable across sessions. It is the deepest of the three reads and has its own full guide: OFI explained. The one-line version: positive and climbing means the book is net-loading on the buy side; it explains the move happening now far better than trade volume does, and it predicts nothing.

The QI · OFI gauge

Below the chart’s info strip, the indicator shows both numbers side by side as QI · OFI. The coloring uses a ±0.15 dead zone: values above +0.15 tint green, below −0.15 tint red, and anything in between stays neutral — by design, so a balanced book reads as “no signal” rather than flickering between colors. The two numbers agreeing (both green or both red) is the higher-conviction state: the book is heavier on one side and being pushed further that way.

Every setting, explained

OFI levels (default 10) — how many book levels the OFI integrates. 10 is the research standard; 1 gives the classic best-level formulation. More levels read deeper intent but react slower.
Smoothing (default 20) — the displayed OFI is an EMA over roughly this many book events. Per-event OFI is flicker; 20 shows the direction of flow. Set 1 for the raw value.
Trail seconds (default 180) — how much micro-price history the line keeps on screen.
Line width / color — cosmetics of the micro-price trail.
Show line / Show gauge — each read can be toggled independently; gauge-only is a common minimal setup.

How to actually use it

The workflow is levels first, book second. Pick your levels — from the volume profile, an absorption zone, a GEX wall. When price arrives, the indicator answers one question: does the book agree? Price into your support with QI pinned positive, OFI climbing and the micro-price hugging the ask is the book backing your trade. The same level with QI negative and OFI bleeding is the book telling you the wall you are counting on is not being defended.

Honest limits

Horizon. The published QI result lives at the one-tick scale, and much of it is mechanical — the shorter queue is consumed first. It does not survive costs and latency as a standalone strategy, which is precisely why we ship it as a chart read and not a signal service.
OFI is contemporaneous. It explains the current move, not the next one — lagged OFI forecasting the next minute produces negative out-of-sample R² in the published research. Full discussion in the OFI guide.
Spoofing exists. All three reads trust the resting book, and resting orders can be pulled. A queue signal that appears and vanishes in seconds is noise; persistence is what makes it information.
Instrument matters. The signals are strongest on large-tick, deep-queue markets — ES above all; MNQ shares the pinned one-tick spread with shallower queues. On thin or small-tick books, expect much less.

Frequently asked questions

What does it mean when the micro-price line sits near the ask?: It means the bid queue is much larger than the ask queue, so the imbalance-weighted fair value has shifted toward the top of the spread. Mechanically, the thin ask queue is the one likely to be consumed first — the book is leaning toward an uptick. The mid-price cannot show you this because it ignores the sizes entirely.
Is queue imbalance actually predictive?: At the one-tick horizon it carries real signal: on large-tick instruments, a classifier built on queue imbalance alone reaches an out-of-sample AUC of roughly 0.75–0.80, and when the queues are extremely lopsided the next mid-price move goes with the heavy side around 80–90% of the time. But that is a mechanical, one-tick effect: after queue position, latency and costs, it is not a tradeable edge on its own. Use it as a state read at your levels, not as an entry signal.
What is the difference between queue imbalance and OFI?: Queue imbalance is a snapshot: the balance of resting size at the best bid and best ask right now, from −1 to +1. OFI is a flow: how much the book has shifted — orders added, pulled, consumed — accumulated over recent updates, across multiple levels. QI tells you the current lean; OFI tells you the direction the book is being pushed.
Why does the micro-price line only cover the last few minutes?: The line is a trail, 180 seconds by default. The micro-price is a high-frequency object — it reacts to every book update and is meaningful at the scale of the current rotation, not the session. A longer trail is available in the settings (Trail seconds), but stretching it too far just draws noise over your candles.
Does the indicator work on every instrument?: It computes on anything with a live level-2 book, but the signals are strongest on large-tick instruments — markets where the spread sits at one tick almost all the time. ES is the canonical case, with queues hundreds deep; MNQ shares the pinned spread but with much shallower queues. On small-tick, fast-cycling books the queue signals carry much less information, which matches the published research.