SoonTech CEX Institutional Market Data Feed: A Multi-Layer Data Infrastructure From Depth Snapshots to Microsecond Increments

ExchangeWhite Label Solution٢٨ يوليو ٢٠٢٦

Institutional traders' expectations for CEX market data have evolved from "5-level depth with 1s updates" to "full-book + microsecond increments + dedicated institutional links". Once HFT market makers, cross-market arbitrageurs, CTA strategies and index products co-live on a CEX, the exchange's market data infrastructure becomes the decisive factor for institutional retention. SoonTech's institutional market data feed covers tiered subscriptions, depth snapshots, incremental updates, trade streams, dedicated links and latency SLA — a product-level reference for white-label CEX operators.

1. Industry Background: From "One WebSocket" to Multi-Layer Distribution

Three generations:

1. Gen 1 (2018–2021) — single WebSocket push shared across all subscribers.

2. Gen 2 (2021–2024) — channelized push where users pick depth, trades and candles.

3. Gen 3 (2024–present) — dedicated institutional lines, UDP multicast, binary protocols, microsecond increments and explicit latency SLA.

Across SoonTech's white-label deployments, institutional willingness to pay for market data has risen sharply — clients pay separately for dedicated links + microsecond increments.

2. Market Pain Points: Latency, Jitter, Loss

Recurring themes:

· Unstable latency — WebSocket latency scales with connections; SLA cannot be promised.

· Jitter — network jitter and JSON serialization overhead create long-tail distributions.

· Packet loss — WebSocket disconnects require full resync; institutional strategies cannot tolerate that.

· Format not friendly — JSON is too heavy for HFT clients.

· No institutional line — institutions want isolation from retail subscribers to avoid "shared bandwidth" issues.

3. Data and Trends: Institutional Requirements for Market Data

From recent onboarding conversations:

DimensionInstitutional focusPlatform capabilityLatency

Median + P99

Microsecond increments + SLA

Jitter

Stable distribution

UDP multicast + time-stamping

Loss

Seamless resubscription

Sequence number + snapshot rebuild

Protocol

Binary vs JSON

Binary + JSON compatibility

Line

Isolated from retail

Institutional line + dedicated VLAN

Billing

Per channel / depth / segment

Tiered billing + API

Institutional market data is not "the push feeding the UI" — it is a bottom-layer product co-equal with the matching engine.

4. Case Analysis: Onboarding a HFT Market Maker

Anonymized scenario: a HFT market maker wants to deploy on a SoonTech-white-labeled CEX with:

· Median latency < 200μs from match engine to client.

· P99 latency < 800μs.

· Seamless resub within 50ms after packet loss.

· Binary protocol + UDP multicast.

· Dedicated line, not shared with retail.

On SoonTech's infrastructure the MM:

1. Subscribes to the "institutional line" channel.

2. Receives full-book increments via UDP multicast.

3. Uses sequence gap detection + snapshot rebuild for resubs.

4. Uses binary protocol to cut serialization overhead.

5. Uses latency SLA reports for monthly reconciliation.

Interim takeaway

Institutional market data is not about "more fields" — it is about making latency, jitter and loss committable, measurable and auditable.

5. SoonTech CEX Institutional Market Data Capabilities

Six modules:

5.1 Tiered subscriptions

· Retail WebSocket / semi-pro WebSocket / institutional UDP.

· Subscription by channel, depth and segment.

· Custom channel combinations.

5.2 Depth snapshots and increments

· Snapshots at 1s / 100ms / event-driven cadence.

· Increments with sequence number + timestamp.

· Gap detection triggers snapshot rebuild.

5.3 Trade stream

· Tick-by-tick trades with match timestamp, side and position markers.

· Filter by symbol or segment.

· Shares sequence numbers with depth increments.

5.4 Institutional line and multicast

· Dedicated VLAN and network path.

· UDP multicast reduces fanout cost.

· Colocation direct connect supported.

5.5 Binary protocol

· Fixed header + variable field binary format.

· JSON-compatible mapping.

· SDKs in C++ / Rust / Go / Java.

5.6 Latency SLA and reconciliation

· Median / P95 / P99 / P999 reports.

· Monthly SLA reconciliation and compensation.

· Time-stamping supports client replay.

6. Enterprise Implementation Suggestions

1. Segment your client tiers — retail, semi-pro and institutional.

2. Define latency SLA — commit to both median and P99.

3. Buy or build binary protocol + UDP multicast for HFT clients.

4. Ship a loss-resub mechanism — sequence number + snapshot rebuild is table stakes.

5. Design tiered billing — institutions will pay for isolation.

6. Vet vendors on tiered subscriptions, institutional lines and latency reconciliation.

Vendor Selection Checklist

· Native tiered subscription support.

· UDP multicast and institutional line.

· Binary protocol with multi-language SDKs.

· Sequence + snapshot resubscription.

· Median + P99 dual-dimension SLA.

· At least one live institutional market data reference.

7. Future Outlook: From "Data Push" to "Data Product"

For 2026–2028:

1. Market data as a standalone product — decoupled from matching fees, standalone revenue line.

2. Latency SLA standardizes — median + P99 + P999 becomes default institutional onboarding.

3. Historical replay as a service — exchanges expose replay APIs for backtesting.

Market data becomes an independent product line for institutions, strategies and reconciliation.

FAQ

Q1: What is the core difference between WebSocket and UDP multicast?

A1: WebSocket is ordered, reliable, unicast. UDP multicast is unordered, unreliable, multicast. The former suits retail UIs; the latter suits HFT. SoonTech offers both, chosen by subscription.

Q2: How is latency measured?

A2: From matching kernel timestamp to client receive timestamp. SoonTech inserts time-stamps at key network hops, so clients can replay the whole latency distribution.

Q3: How fast can loss-resub complete?

A3: Via sequence gap detection + snapshot cache, SoonTech typically completes resubs in under 50ms.

Q4: Do institutional lines require Colocation on the client side?

A4: Not mandatory. SoonTech offers Colocation direct connect, regional POPs and cross-border dedicated lines. Clients choose based on latency needs.

Q5: How is market data priced?

A5: SoonTech bills by channel, depth and tier. Institutional lines are priced separately from matching fees.

Conclusion

Institutional market data is the key infrastructure that takes a CEX from "retail platform" to "institutional platform". SoonTech's market data feed turns latency, jitter, loss, protocol, dedicated lines and SLA into a productized deliverable — helping white-label CEX operators onboard HFT MMs, cross-market arbitrage, CTA and index strategies in the next institutional wave.

🌐 Build secure and scalable Web3 platforms with SoonTech.

Explore our solutions for White Label Crypto Exchanges, Prediction Markets, MPC Wallets, Matching Engines, Liquidity Integration, and Compliance.

ابدأ رحلة blockchain الخاصة بك

سيقدم لك الفريق المحترف استشارة مجانية حول الحلول

اتصل بنا