01.TRADING INFRASTRUCTURE// mev · bots · arbitrage
MEV / TRADING

The route from signal
to landed trade.

Four stages decide every trade: see it, understand it, land it, scale it. We run the infrastructure for each stage — on every chain we cover.

[ get started ]explore_services → 48h trial · no card
0.4 ms
shreds (FRA) · pre-conf
3.8 ms
gRPC p99
98.7 %
landed · SWQoS + Jito
99.9 %
uptime 30d
02.THE ROUTE// signal → context → execute → scale
SIGNAL

See it before the crowd.

Order books, accounts, and blocks pushed in real time — no polling lag. On Solana the serious edge is raw shreds before block confirmation.

data streaming →SOL gRPC·ShredsHL WSMON gRPC
stream — live● pushed
pool event +0.4 ms
account diff decoded
window open
CONTEXT

Know what you're looking at.

Counterparty history, market structure, backtest data — indexed tables you query like a database instead of replaying the chain block by block.

context.sql// instant history
SELECT * FROM swaps
WHERE wallet = counterparty
→ rows · ~15s fresh
EXECUTE

Land it first.

Reads are half the trade — the order has to arrive. Fast writes on every chain; on Solana a stake-weighted priority path with a Jito fallback.

fast tx sending →SOL Sender 98.7%RPC writes — all chains
delivery// priority path
await sender.send(tx)
path: SWQoS + Jito
→ landed · same window
SCALE

Outgrow the caps, not the provider.

Shared plans publish their limits so bursts are plannable. When the strategy gets serious, a bare-metal dedicated node takes over — your region, your tuning, no rate limits. Any chain, built per order.

dedicated →ANY CHAINcustom quote
spike.log// dedicated
09:30 volatility ↑
09:30 p99 flat · no caps
→ trades through it
03.THE SWITCH// generic rpc → trading stack
data pathHTTP polling, seconds behindstreams pushed (0.4 / 3.8 ms on Solana)
submissionpublic queue with everyonepriority delivery, 98.7% landed
volatilitysilent throttling at the worst timepublished caps · none on dedicated
coverageone chain per providersol · hl · mon · bnb — one chat
supportticket queueengineers in Telegram
04.TRADING INFRASTRUCTURE FAQ// 8 questions
Which chains does Supanode support for trading?

Solana carries the full stack: JSON-RPC, Yellowstone gRPC, ShredStream, the TPU Sender, and a DEX indexer. Hyperliquid adds live WebSocket orderbook streaming and a nanosecond-precise fills indexer, Monad runs its native gRPC feed, BNB Chain runs as dedicated nodes, and a Polymarket indexer covers prediction-market history. Ethereum, Base, Arbitrum, and other ecosystems are available as dedicated nodes on request.

What latency can I expect for trading on Solana?

ShredStream delivers raw shreds at 0.4 ms p99 and Yellowstone gRPC streams live state at 3.8 ms p99, both from the Frankfurt edge; on the write side, the TPU Sender runs a 1.2 ms submit-to-leader p50. Nodes sit adjacent to major validator clusters in top-tier datacenters, so in-region round trips stay in single-digit milliseconds.

What is the lowest-latency data path Supanode offers for trading?

ShredStream — raw Solana shreds delivered over UDP multicast at 0.4 ms p99 from the Frankfurt receiver. Reading shreds at the source avoids the latency of RPC indexing or gRPC re-emission. It is rented per IP, with a 7-day minimum and 90-day maximum subscription.

How does Supanode get transactions to the slot leader?

The TPU Sender fans every transaction out over two parallel paths: stake-weighted QoS direct to the current slot leader, and the Jito Bundle Engine. Whichever lands first wins. It tracks the leader schedule in real time and lands 98.7% of transactions at a 1.2 ms submit-to-leader p50, with a 1,000,000-lamport (0.001 SOL) minimum tip.

Is streaming throughput metered?

No. Yellowstone gRPC throughput is unlimited and unmetered — there is no per-MB or per-credit billing. The bundle tier sets concurrent connections and per-stream filter width; data volume on top of that is not billed, so a volatility spike does not create a surprise invoice.

Are there rate limits for high-frequency trading?

Shared Bundle plans publish their caps openly — RPS scales 15 to 500 and TPS 5 to 100 by tier — so the hardware is never oversold. The TPU Sender is open at 5 TPS per client. For unlimited RPS and TPS, dedicated nodes have no rate limits: the hardware is reserved entirely for your traffic.

How do I migrate from my current RPC provider?

Supanode runs standard interfaces, so the change is the endpoint URL: JSON-RPC stays JSON-RPC, the gRPC feed is Triton/Yellowstone-compatible with the same protobuf and subscribe API, and the TPU Sender keeps your sendTransaction calls as they are. Provisioning is hands-on over Telegram — share your IP and workload, and an engineer sets up the endpoint and IP allowlist.

How is trading infrastructure priced?

Streaming and RPC products are flat monthly subscriptions, crypto-only, with no per-request metering — a busy month costs the same as a quiet one. The TPU Sender is pay-per-use via the tip on each transaction (minimum 0.001 SOL) with no monthly fee, and dedicated nodes are quoted per deployment. Most products carry an up-to-48-hour free trial, no card; exact rates live on the pricing pages.

Ready to trade faster?

Pick your chain, walk the route, start executing. Provisioning over Telegram with an engineer.