Level 2 Order Book Data: When You Need It and When It Is Over-Procurement

"Do we need Level 2?" is one of the most common debates in market data procurement. It is usually not a cost question but a capability anxiety — the fear of missing something. Yet Level 2 carries far higher data volume, storage cost, and engineering complexity than Level 1, so getting the direction wrong is expensive.

This article provides a decision framework: what Level 2 actually solves, and when you do not need it.

The essential difference between Level 1 and Level 2

The distinction reduces to one line: Level 1 tells you what the price is; Level 2 tells you how much size that price can absorb.

What Level 2 genuinely solves

Only three classes of question require Level 2:

Question one: will this order move the price materially?

Level 1 shows the size resting at the best price. If the size you need to trade exceeds that, the price walks to the next level — but Level 1 cannot tell you how far, or what the total cost will be. This is Level 2's least substitutable use case.

Question two: is the book real?

Displayed liquidity can be fleeting — orders posted and cancelled rapidly. The best bid alone cannot tell you whether a quote is firm; the distribution of size across levels offers more signal.

Question three: microstructure research

Order flow imbalance, book pressure, and spread dynamics all require multiple levels. Research of this kind demands Level 2 by definition.

Three situations where you do not need Level 2

Conversely, buying Level 2 in these cases is usually waste:

1. Display and dashboards. Users need prices and change percentages, not a five-level ladder. The vast majority of consumer-facing market interfaces are well served by Level 1.
2. Low and medium-frequency strategies. For daily or hourly decisions, the best price carries enough information; book depth will not change the signal.
3. Valuation and risk snapshots. Portfolio valuation and exposure calculations work off the last traded price and do not need book structure.

What these share: the decision depends on "at what price", not "how much size". If that holds, Level 2 is over-procurement.

The cost is not just the subscription

The real cost of Level 2 has four parts, and the last three are routinely ignored:

Engineering complexity is the most underestimated. A multi-level book is not "a few more fields" — it is stateful. You must apply incremental updates in order to reconstruct the book, handling out-of-order delivery, gaps, and snapshot alignment. Building and maintaining that logic often costs more than the subscription itself.

Decision checklist

Answer these questions. If every answer is no, you do not need Level 2:

1. Does your strategy place orders larger than the size resting at the best price?
2. Do you need to assess "how much can trade at this price" rather than "what is the price"?
3. Do you need to judge whether displayed quotes are genuine?
4. Are you doing order flow or microstructure research?
5. Do you have a team able to maintain book reconstruction and malformed-book filtering?

Question five is the blind spot for many teams: being able to consume Level 2 data is not the same as being able to use it correctly. Without someone maintaining book reconstruction, multi-level data becomes decoration.

Python: validate the need with Level 1 data first

Before upgrading, run a rough validation on Level 1 data. If your typical order size routinely exceeds the size at the best level, Level 2 becomes meaningful.

import httpx

API_KEY = "your_api_key"
BASE = "https://api.tiqex.io/api/v1/market"

# Your typical order size, in shares or contracts
TARGET_ORDER_SIZE = 5_000


def assess_depth_need(codes: list[str], order_size: int) -> None:
    """Screening pass using Level 1 bid/ask size to decide whether depth is needed."""
    resp = httpx.get(
        f"{BASE}/quote",
        params={"codes": ",".join(codes)},      # up to 50 codes per call
        headers={"X-API-Key": API_KEY},
        timeout=10,
    )
    rows = resp.json().get("data", [])

    over = 0
    for r in rows:
        if r.get("unavailable"):
            print(f"{r.get('code'):<12} unavailable, skipping")
            continue

        # Level 1 exposes only the top-of-book size; take the thinner side as conservative
        sizes = [v for v in (r.get("bidSize"), r.get("askSize")) if v is not None]
        top_size = min(sizes) if sizes else None

        if top_size is None:
            print(f"{r.get('code'):<12} no book size field; confirm manually")
            continue

        if order_size > top_size:
            over += 1
            print(f"{r.get('code'):<12} order {order_size} > top size {top_size}  -> depth needed")
        else:
            print(f"{r.get('code'):<12} order {order_size} <= top size {top_size}  -> Level 1 suffices")

    print(f"\n{over}/{len(rows)} symbols would breach the top level")
    print("Conclusion: Level 2 is not required at the current order size" if over == 0
          else "Conclusion: slippage risk exists; Level 2 is worth evaluating")


assess_depth_need(["AAPL", "MSFT", "NVDA"], TARGET_ORDER_SIZE)

The logic is to use "does my order exceed the size at the best level" as the first screening pass. If the best level dwarfs your order size across every symbol, the extra information Level 2 provides will not affect your decisions.

One caveat: the availability of book size fields depends on plan entitlements and market. If the fields come back empty, your current tier does not expose them, and you should confirm the actual response structure before relying on this screen.

The pragmatic conclusion

1. Validate with Level 1 first. If order size never exceeds the best level, Level 2 does not solve anything.
2. Define an explicit trigger. Make "order size exceeds the top level" or "liquidity assessment is required" the condition for upgrading, rather than a gut feeling.
3. Assess engineering capacity first. Book reconstruction, level alignment, and anomaly filtering usually cost more than the subscription.
4. Go hybrid. Enable Level 2 for the handful of symbols that need depth and keep Level 1 for the rest.

FAQ

Is Level 2 always better than Level 1?

No. Level 2 provides more information at higher cost and greater engineering complexity. If the decision does not depend on "how much size", Level 1's best-price information is sufficient.

Does high-frequency trading require Level 2?

High-frequency strategies usually do need book depth — but only if the team can handle book reconstruction, incremental updates, and anomaly filtering. Without that engineering capability, having the data does not mean you can use it.

Can Level 2 book data be used as-is?

No. A multi-level book is stateful data: you must apply incremental updates in sequence to reconstruct it, handling out-of-order delivery, gaps, and snapshot alignment. That is the largest source of Level 2 engineering cost.

How do I tell whether my order size genuinely needs depth?

Screen with Level 1 best-level size. If your typical order routinely exceeds the size resting at the best price, slippage risk is material and Level 2 is worth evaluating.

Can I enable Level 2 for only some symbols?

Yes, and you should. Enabling depth for the few illiquid or large-order symbols while keeping Level 1 elsewhere is a sensible cost-to-benefit balance.

Next steps

Answer the five checklist questions before deciding to upgrade. The conceptual difference between Level 1 and Level 2 is covered in Level 1 versus Level 2; market entitlements and depth data availability by tier are on the pricing page. If you are working on liquidity assessment or order slicing, see the trading solution, and book-related field definitions are in the multi-market quote docs.