Level 3 market data is the full order book shown order-by-order — every individual buy and sell order, each with its own ID, price, size, side and a stream of add/modify/cancel/trade events. It is the most granular public market data there is. The catch: the term means three different things. To data vendors and quants it means market-by-order (MBO) data such as Nasdaq TotalView-ITCH. In the original NASDAQ sense it means the interactive terminal that market makers used to enter quotes. In crypto it means the order-by-order API feed. This guide separates all three and shows what Level 3 actually contains, who can get it, and what it costs in 2026.
What is Level 3 market data?
Level 3 market data is a feed that exposes every individual order in the book, not just aggregated prices. Where Level 2 tells you “there are 500 shares bid at $10.00,” Level 3 tells you those 500 shares are actually ten separate orders of 50 each, in a specific queue order, every one carrying a unique order ID you can track from the moment it is placed to the moment it is filled or cancelled. In modern usage this is the same thing as market-by-order (MBO) data, also called the “full order book.”
If you already understand Level 2 market data and the order book, the simplest way to frame Level 3 is this: Level 2 is the summary of the book by price; Level 3 is the raw book, every order kept separate. That single difference — keeping orders individually addressable instead of summing them — is what makes Level 3 both far more powerful and considerably harder to work with.
Level 1 vs Level 2 vs Level 3: the full ladder
The three “levels” describe increasing resolution on the same underlying market. Each level contains everything below it and adds more.
| Level | What you see | Granularity | Typical user |
|---|---|---|---|
| Level 1 | Best bid & best offer (the inside market) + sizes; last trade | Top of book only | Almost every investor; default broker quote |
| Level 2 | Multiple price levels of depth, quantity aggregated per price | Market-by-price (MBP), usually ~10 levels | Active and day traders |
| Level 3 | Every individual order, with order ID, size, side and event history | Market-by-order (MBO), full depth | Quants, market makers, microstructure researchers |
The jump from Level 1 to Level 2 adds depth. The jump from Level 2 to Level 3 adds identity — the ability to follow each order as a distinct object. That is a smaller-sounding change with much larger consequences, because identity is what unlocks queue position and accurate fill simulation, which we get to below.
The terminology trap: “Level 3” means three different things
This is where most explanations go wrong, and it is the single most important thing to get right. “Level 3” is used in three incompatible ways, and people argue past each other because of it.
1. The classic NASDAQ meaning — a market-maker terminal
Historically — and still in the FINRA Series 7 curriculum — Level 3 was not a data feed at all. It was the interactive NASDAQ Workstation service available only to registered market makers. A Level 3 quote looked like a Level 2 quote, but it was two-way: it let the market maker enter, change and remove their own quotes, execute orders, and send trade confirmations. Retail traders never had it and never could. By this definition, “Level 3” is a permission, not a dataset.
2. The modern data-vendor meaning — the full order book (MBO)
Today most vendors, exchanges and quants have repurposed “Level 3” to mean full order-book granularity: every resting order shown individually. This is the meaning behind feeds like Nasdaq TotalView-ITCH, CME’s MDP 3.0 market-by-order, and IEX’s order-level feeds. When a data company sells you “Level 3,” this is almost always what they mean. It has nothing to do with quote-entry permissions.
3. The crypto meaning — order-by-order API data
Crypto exchanges adopted the same numbering but applied it cleanly: L1 = top of book, L2 = aggregated by price, L3 = every order individually with its own ID. Because crypto venues are newer and API-first, many of them expose true L3 order-by-order data directly through their public or paid APIs — something traditional equity markets guard far more tightly.
If a broker or course says “Level 3” and means the market-maker terminal, and a data vendor says “Level 3” and means MBO, they are describing two unrelated things. Always ask: do you mean the permission to enter quotes, or the order-by-order data feed? In 2026, when people search “Level 3 market data” they almost always want the second.
Market by order (MBO) vs market by price (MBP)
The technical heart of Level 3 is the difference between MBO and MBP. It is best shown with a single price level.
Suppose ten traders each place a buy order for 50 contracts at 4500.00. In a market-by-price (Level 2 / MBP) feed, you see one row: 4500.00 — 500 — 10 orders. The quantity is consolidated; the individual orders are invisible. CME’s MBP feed, for example, restricts updates to a maximum of 10 price levels and collapses everything at each price into a single total. In a market-by-order (Level 3 / MBO) feed, you see ten separate records, each with its own order ID, its 50-lot size, and — crucially — its place in the queue.
That last line is the whole point. With MBP you can estimate your queue position only by guessing; with MBO you can compute it. For anyone trying to model whether a passive limit order will actually get filled, that distinction is the difference between a rough heuristic and an exact answer.
What is inside a Level 3 order record?
A Level 3 / MBO feed is a stream of events, not a static snapshot. Each event references a single order. The core fields are:
- Order ID (order reference number) — the unique handle that lets you track one order across its whole life.
- Price and size — where the order sits and how big it is.
- Side — buy or sell.
- Event type — add (new order), modify/replace (size or price change), cancel (order removed), trade (a fill), plus venue-specific events.
- Sequencing — a sequence number and/or timestamp so you can apply events in the exact order they happened.
Implementations differ in subtle but important ways. On queue position, Nasdaq TotalView relies on implicit timestamps, while CME’s MDP 3.0 publishes an explicit priority field (tag 37707, MDOrderPriority). On size modifications, most feeds assume an order keeps its queue position when size decreases but loses it when size increases; IEX’s DEEP+ feed makes that behaviour explicit rather than implied. On trade publishing, Nasdaq TotalView shows only the passive (resting) side of a fill, whereas CME MDP 3.0 shows both the aggressor size and the individual fills. None of this matters at Level 2; all of it matters at Level 3.
Why Level 3 is harder than Level 2
Level 2 usually arrives as a ready-made depth table you can read directly. Level 3 does not. With an MBO feed you have to build and maintain the order book yourself from the raw event stream — adding orders, applying modifications, removing cancellations, matching trades — and you have to do it per venue, respecting each venue’s quirks above. Get the rules slightly wrong and your reconstructed book silently diverges from reality.
There is also far more of it. Because an MBO feed has to publish an event for every order at every price level — not a periodic snapshot of ten aggregated levels — the bandwidth and message rates are dramatically higher. That is one reason, in Databento’s words, that “much fewer vendors provide L3 data than L1 data due to the additional complexity and bandwidth requirement.”
# Reconstructing a Level 3 (MBO) order book from an event stream.
# Each event references one order by its unique order_id.
from collections import defaultdict
book = {} # order_id -> {"side","price","size"}
levels = defaultdict(int) # (side, price) -> total resting size
def apply(event):
oid = event["order_id"]
etype = event["type"] # "add" | "modify" | "cancel" | "trade"
if etype == "add":
book[oid] = {"side": event["side"], "price": event["price"], "size": event["size"]}
levels[(event["side"], event["price"])] += event["size"]
elif etype == "modify":
o = book[oid]
levels[(o["side"], o["price"])] -= o["size"] # remove old
o["price"], o["size"] = event["price"], event["size"]
levels[(o["side"], o["price"])] += o["size"] # add new
elif etype in ("cancel", "trade"):
o = book.pop(oid, None)
if o:
filled = event.get("size", o["size"]) # trade may be partial
levels[(o["side"], o["price"])] -= filled
if etype == "trade" and filled < o["size"]: # partial fill: order stays
o["size"] -= filled
book[oid] = o
levels[(o["side"], o["price"])] += o["size"]
# Level 2 (MBP) is just this MBO book aggregated by price.
# That is the one-way relationship: you can derive L2 from L3, never the reverse.
Who can actually get Level 3 data — and what it costs in 2026
This is where reality bites. In US equities and options the consolidated public tapes give you almost nothing at this resolution:
- The equity SIPs (the CTA and UTP consolidated feeds) provide only Level 1. For full depth you must license proprietary direct feeds from each exchange.
- For options, the OPRA consolidated feed is also Level 1 only; Level 3 again means direct proprietary feeds.
- The best-known equity full-depth product is Nasdaq TotalView, which Nasdaq says shows roughly 20× the liquidity of Nasdaq Level 2. Its display price is modest — about $15/month for non-professional users (TotalView $14 + OpenView $1) versus $76/month for professionals — but that is the display entitlement; the raw TotalView-ITCH order-by-order feed for systematic use is a different, heavier licensing tier.
- In futures, CME offers market-by-order via MDP 3.0 for supported products.
- Specialist vendors such as Databento normalise MBO across venues so you do not have to integrate each exchange feed yourself.
- In crypto, many exchanges expose order-by-order data directly through their APIs, often free or cheap — the most accessible true Level 3 on the market.
“Cheap Level 3” usually means a display tool like Nasdaq’s BookViewer. “Real Level 3” — a normalised, machine-readable MBO feed you can run strategies and backtests on — is a professional data product with professional pricing and real engineering work attached.
What is Level 3 market data used for?
The use cases all flow from one feature Level 2 cannot provide: order identity.
- Queue position. Knowing exactly how many orders sit ahead of yours at a price is impossible with aggregated MBP and exact with MBO. This drives passive execution decisions.
- Fill simulation and backtesting precision. Because you can track each order, you can simulate whether your passive order would have been filled at a given moment, instead of guessing. Databento notes this gives a “better backtesting engine,” and that even traders who are not latency-sensitive benefit from the accuracy.
- Microstructure signals. With order identity you can measure exactly how long each order has rested on the book, building staleness- and order-flow-based signals that simply do not exist at Level 1 or Level 2.
- Market making and HFT — the original audience — but, importantly, not the only one.
Do retail traders need Level 3 data?
For almost all retail traders, the honest answer is no. Level 2 already shows more depth than most discretionary strategies can act on, and Level 3’s advantages — queue position, exact fill simulation, microstructure signals — are only realised by systematic strategies with the engineering to consume and reconstruct an MBO feed. Paying for raw Level 3 to trade by eye is buying a telescope to read a street sign.
There is also a structural point that catches a lot of newer traders, and I learned it the practical way trading CFDs on Plus500. When you trade CFDs you are not even looking at the exchange’s order book — you are trading against your broker’s synthetic price. Level 3 of the underlying market is invisible to you by design. If genuine order-book depth matters to your strategy, that is an argument for direct market access (DMA) equity or futures trading through a serious platform, not for adding a data subscription on top of a CFD account. If you are choosing where to trade, our guides to the best online brokers and AI trading platforms are the better starting point.
Level 3 = the full order book, order-by-order, with IDs and queue position (MBO). It is a professional-grade dataset, not a retail upgrade button — and “Level 3” in a market-maker context means something else entirely.
FAQ
Is Level 3 market data the same as the full order book?
In modern usage, yes. Level 3 means market-by-order (MBO) data — every individual order shown separately with its own ID, rather than aggregated by price. That is exactly what “full order book” refers to. The only confusion comes from the original NASDAQ meaning of “Level 3,” which was a quote-entry terminal for market makers, not a dataset.
What is the difference between Level 2 and Level 3 market data?
Level 2 is market-by-price: it shows several price levels with the quantity at each price summed together (e.g., “500 at 4500.00”). Level 3 is market-by-order: it shows each order at each price individually, with a unique order ID, so you can see exactly how many orders make up that 500 and where yours sits in the queue. Level 3 contains everything in Level 2 plus order identity.
Can retail traders get Level 3 market data?
Sometimes, but rarely usefully. Display products like Nasdaq’s BookViewer (powered by TotalView) are available to non-professionals for around $15/month, and many crypto exchanges expose order-by-order data through their APIs. But a normalised, machine-readable MBO feed for systematic trading or backtesting is a professional data product with professional pricing, and consuming it requires building the order book yourself.
What does an order ID do in Level 3 data?
The order ID (or order reference number) is the unique handle that lets you follow a single order through its entire life — placement, any modifications, and its eventual fill or cancellation. It is the defining feature of Level 3: without per-order identity, you only have aggregated Level 2.
Why is Level 3 data more expensive and harder to use than Level 2?
Two reasons. First, message volume: an MBO feed publishes an event for every order at every price level, so bandwidth and message rates are far higher than a Level 2 depth snapshot. Second, you must reconstruct and maintain the order book yourself from the raw event stream, respecting venue-specific rules for queue position, size changes and trade reporting. Fewer vendors offer it as a result.
What is the original NASDAQ meaning of Level 3?
Historically, Level 3 was the interactive NASDAQ Workstation service available only to registered market makers. It let them enter, change and remove their own quotes, execute orders, and confirm trades — a permission, not a data feed. This is still the definition used in the FINRA Series 7 curriculum, which is why “Level 3” is so ambiguous today.
Does Level 3 data apply to crypto?
Yes, and crypto uses the cleanest version of the terminology: L1 is top of book, L2 is aggregated by price, and L3 is every order shown individually with its own ID. Because crypto exchanges are API-first, many expose true L3 order-by-order data directly — often more accessibly than traditional equity markets, where full depth requires proprietary exchange feeds.
Bibliography (10 sources)
Sources prioritise official exchange documentation and primary data-vendor specifications. Pricing and feed details reflect publicly listed figures and may change; verify current entitlements with the venue. Links accessed June 2026.
- Databento — What is Level 3 (L3) market data?. Primary technical reference for MBO definition, venue differences and use cases.
- Databento — What is market by order (MBO) data?. MBO schema and order-record fields.
- CME Group — Market by Order (MBO) FAQ. Official MBO vs MBP definition for futures.
- CME Group — MDP 3.0 book management messages (MBO). Source for explicit queue priority (tag 37707, MDOrderPriority).
- Nasdaq — Nasdaq TotalView. Full-depth equity feed; ~20× the liquidity of Nasdaq Level 2.
- Nasdaq — Nasdaq BookViewer. Non-professional display tool powered by TotalView.
- IEX Exchange — IEX market data feeds (DEEP+). Explicit handling of size-modification queue behaviour.
- U.S. Securities and Exchange Commission — Market data and the consolidated tape (SIP). Why consolidated feeds provide Level 1 only.
- FINRA — Series 7 exam content (NASDAQ quote levels). Classic Level 1/2/3 market-maker definition.
- CoinAPI — Level 1 vs Level 2 vs Level 3 market data (crypto order book). Crypto L3 order-by-order definition.
