Quant / Systems · Independent research system
SKURRA
Event-driven Solana quantitative research engine
Built a real-time Solana research system that tracks whale wallets and analyzes their trades as they happen. It connects directly to on-chain activity, identifies swaps across major protocols, and keeps track of each wallet's positions, cost basis, and PnL. I also designed scoring layers for wallet quality, token risk, liquidity, developer behavior, and momentum.
Monitored five whale wallets at once using WebSockets, so new activity could be processed as on-chain events arrived instead of waiting for periodic refreshes.
Parsed swaps across Pump.fun, Raydium, Orca, and Jupiter even though each protocol represents trades differently.
Converted trades into one common format containing the wallet, tokens, amounts, timestamp, signature, route, and price context.
Tracked each wallet's position size, average cost, and realized/unrealized PnL across buys, partial sells, and full exits.
Designed statistical features such as win rate, drawdown, holding time, PnL volatility, entry timing, and volume momentum.
Different Solana subscription methods did not always surface swap activity the same way, so transaction detection and lookup had to be tested across providers.
Some provider responses were missing useful token metadata, which required a separate enrichment step.
PnL tracking depends on earlier trades, so a live event stream alone was not enough; the system needed persistent position state.
Built the working real-time ingestion, swap parsing, enrichment, and position-tracking core; designed the database and scoring layers for the next stage.