Technical Research Notes
Infrastructure and methodology papers from the QuantEdge research program. No proprietary signals, entry logic, or execution parameters are disclosed. All work is published when experiments reach stable, repeatable conclusions.
How we eliminated compounding feed lag at SPY market open — from a growing 45-second backlog to stable sub-second bar sealing — through queue-first callback offloading, per-venue DOM isolation, event-time binning across six independent intervals, and a single source of truth for all technical indicators. Covers Python GIL contention, IBKR receiver thread architecture, and executor domain isolation. Infrastructure only; no signal logic disclosed.
A production-validated multiprocessing architecture that eliminates Python GIL contention by decomposing the execution engine into three independent OS processes: Data Ingestor (P1), Execution Engine (P2), and Telemetry Monitor (P3). Inter-process communication is handled by Redis using Apache Arrow serialization for DataFrame transfer and Redis Streams for event signaling. Achieves sustained feed lag below two seconds at market open, scheduler cycle times under 80 ms, and true CPU core isolation across a 28-core workstation. Implemented and validated in production on SPY equity and options markets via the Interactive Brokers TWS API.
An empirical study of the temporal gap between price-action structural events and MACD histogram confirmation signals across 30-second, 1-minute, and 5-minute SPY bar data. Using 10,815 one-second bars and 69,182 options contract observations, we quantify the average confirmation delay at each resolution tier and translate the resulting underlying price displacement into direct option premium cost. MACD confirmation trails structural entries by up to 140 minutes at 5-minute resolution, with option mid prices moving adversely in 89.5% of matched entry pairs. Annotated multi-panel candlestick charts with EMA overlays, MACD panels, and RSI panels illustrate each timeframe.