Rahul Sharma
200 posts

Rahul Sharma
@siamese_cat98
Either my ambition will consume me, or comparison will.


Project - HFT orderbook trading simulator benchmarked in Python and C++ Versions benchmarked:- ->V1 (Python baseline): Dict-based order book; rebuilds top-5 by sorting the full book on every snapshot correct reference, slow snapshots. ->V2 (Python top-5 cache): Same dict book as V1, but keeps an incremental top 5 cache so snapshots are fast without sorting the whole book. ->V3A (Python direct array, scan): Orders and price levels stored in arrays (FPGA-style direct indexing); snapshots scan the full price range fast updates, slow snapshots. ->V3B (Python direct array + cache): Same array book as V3A, plus a maintained top 5 cache combines direct indexing with fast snapshots (best Python version in our full pipeline). ->V4 (C++ map baseline): Same logic as V1 in C++ (unordered_map + std::map) validates correctness and gives much lower native-code latency. Github repo in comments







JEFFERIES SAYS INDIA’S SIP BOOM GAVE FPIS A MASSIVE EXIT WINDOW, PUTTING HEAVY PRESSURE ON THE RUPEE.



















