Quote against state that exists.

Real-time book state, your fills with fees and rebates, and funding events: everything inventory management needs from one subscription.

Your problem

The parts you keep working around.

01

A crossed or stale book means quoting against state that doesn't exist.

02

Fill feeds without maker/taker role and fee/rebate detail make PnL attribution guesswork.

03

Funding payments arrive in your accounting a day late, not in your risk loop.

The proof

The book, one update per block.

Twelve levels a side, guaranteed never crossed. Sizes change; the layout never moves, the same discipline the live feed holds under load.

StreamL2Book · HYPE

mid 12.0000|spread 0.0010

live

NSizeBid
24415.4511.9995
134499.0011.9990
1110880.2211.9985
44341.9611.9980
48447.5611.9975
75051.4411.9970
1714054.5611.9965
59766.6011.9960
510613.9811.9955
86844.5011.9950
1614250.4511.9945
1220638.9311.9940
AskSizeN
12.00057113.8314
12.00102427.909
12.00153849.074
12.002010831.5915
12.00255091.917
12.00308405.3911
12.003511373.395
12.004015547.979
12.004518501.6710
12.00504976.302
12.00559380.3212
12.006018348.223

Simulated book in the StreamL2Book format: one update per block, never crossed. Full L4 carries every resting order when queue position matters.

Integration sketch

Depth and your own fills on one connection.

Filter fills to your trading wallets server-side. Every fill carries maker/taker role, the fee, and the rebate flag, so attribution happens on the wire, not in your accounting.

Read the docs
# Depth + your own fills on one connectionbook = stub.StreamL2Book(    orderbook_pb2.L2BookRequest(coin="HYPE", n_levels=20),    metadata=metadata,) fills = swap_stub.StreamSwaps(    swaps_pb2.SwapRequest(        market_type="ALL",        wallets=["0xYOUR_TRADING_WALLET"],  # server-side filter    ),    metadata=metadata,)
Where to start

The plan that fits.

Enterprise

CustomRecommended

Dedicated infrastructure, custom terms.

Serious quoting wants a dedicated node with private throughput and a 99.99% SLA. Pro is the right place to integrate and test.

Talk to engineering

Talk through a dedicated node.

Serious quoting wants private throughput and a 99.99% SLA. Integrate and test on Pro first if you prefer. The schema is identical.