Market Data Levels (L1, L2, L3). How deep your view of the book goes.
Exchange market data comes in tiers of increasing detail. Level 1 gives you the best prices, Level 2 gives you the depth behind them, and Level 3 gives you individual orders. Each step up multiplies both the information and the engineering required to handle it.
Choosing the right level is a practical decision: most strategies need far less than people assume, and paying for detail you cannot process is a common and expensive mistake.
More detail is not automatically more edge
Level 3 data lets you reconstruct the exact queue of every order, which matters enormously if you are competing for queue position at microsecond latency. If your strategy holds positions for hours, that same data is a vastly heavier feed telling you nothing you will act on. The correct level is the least detailed one that supports your actual decisions, because every extra level adds bandwidth, storage and failure modes.
Market Data Levels (L1, L2, L3) in a strategy
Best bid, best ask and last trade. Enough for spread monitoring, signal generation and most retail automation.
Total resting size at each price level. What you need for slippage estimation and order book imbalance.
Every individual order with its own identifier, so queue position and order lifecycle are visible. Market making territory.
The shorter the holding period, the more the microstructure detail actually changes a decision.
Pick the lightest feed that answers your question
Not a trade. A design decision made once, which determines what your strategy can do and what it costs to run.
- IFIf the strategy only needs price triggers and spread checks, Level 1 is sufficient
- ANDIf it needs to size orders against available depth or read imbalance, Level 2 is required
- ANDIf it needs queue position or order lifecycle, only Level 3 will do
- ANDIn every case, confirm your infrastructure can process the feed without falling behind
The final condition is where this usually goes wrong. A Level 2 or Level 3 feed on a busy pair produces an enormous update rate, and a consumer that cannot keep up silently falls behind, at which point your view of the book is stale and every decision made from it is wrong. A strategy acting on stale depth is worse off than one that only ever had Level 1, because it is confidently mistaken rather than merely uninformed.
Level 1 and Level 2 are published by essentially every venue, usually as a snapshot plus incremental updates.
Full order-by-order data is offered by fewer exchanges, sometimes only to institutional clients or through colocated access.
A periodic depth snapshot is a lightweight substitute when you need occasional Level 2 detail rather than a live stream.
Data level cannot compensate for latency
Level 3 data seen 300 milliseconds late is not more useful than Level 1 seen instantly. If you are not colocated and running low-latency infrastructure, the strategies that genuinely require deep book data are ones you are structurally unable to compete in, and no data subscription changes that.
Trade the order book, not just the last price.
Your keys, your account, your rules. We only send the orders.
- Level 1: top of book
- Level 2: aggregated depth
- Level 3: full order detail