Flagship project

OnChainIQ

A multi-chain, real-time on-chain data engineering and quant research platform: six chains streamed into one core, wallets ranked by realized PnL, and every claimed edge put through a backtesting gauntlet.

PythongRPCZMQJSON-RPCSQLiteSolanaBitcoinEVMBaseSuiBNB
1.13B
on-chain swaps analyzed
500k
wallets ranked by realized PnL
220k
tokens covered
6
chains streamed in real time
Overview

What it is

A real-time on-chain data system that streams launch, swap and migration events across six chains and unifies them under one chain-agnostic core, sharing roughly 70% of its code across every chain it touches.

Build

What I built

Ingestion

One core, six chains

Each chain speaks a different transport. The core normalises them behind a single interface.

Solana
Yellowstone gRPC
Geyser stream, sub-slot latency
Bitcoin
Core ZMQ + JSON-RPC
block and tx notifications, RPC enrichment
EVM · Base · BNB
WSS eth_subscribe
log subscriptions over websockets
Sui
SSE
server-sent event stream
Findings

What held up

Skill is real. Top-quartile wallets by realized PnL stayed 86% net-positive out-of-sample against 42% for the bottom quartile. A clean monotonic gradient. ✓
The edge persists. Copy-edge reached +78% after costs and passed walk-forward validation, rather than evaporating once fees and latency were charged against it. ✓
Rigor built in. Out-of-sample splits, walk-forward, cost and latency haircuts, and artifact rejection. That last one caught a bonding-curve "dust-price" mirage faking +22x returns.
Social media intelligence. On-chain across six chains plus social-signal collectors collecting (Telegram, Twitter, Twitch & Discord) feeding the same analytics layer to be smart and robust.
Stack

Built with

Languages & storage

PythonSQLite

Streaming transports

Yellowstone gRPCBitcoin Core ZMQJSON-RPCWSS eth_subscribeSSE

Chains

Solana (Anchor IDL)Bitcoin CoreEVM / Base / BNB (ABI / web3)Sui (Move)

Execution & research

Jito bundlesMEV-aware microstructureRealized-PnL analyticsWalk-forward backtesting

Want the detail?

The source is on GitHub. If this is the kind of system your team needs, I would like to hear about it.