Ojas Sapkota

I’m a data analyst focused on statistical modeling, machine learning, and large-scale data acquisition, especially for financial market research. I’ve built pipelines and frameworks to turn raw transaction data into usable datasets and testable signals—grounded in statistical rigor and reproducible workflows.

Currently, I’m developing a framework to identify potential cross-trading and pump-and-dump behaviour from trade-level NEPSE floorsheet data. I built a concurrent Scrapy pipeline to scrape over 70 million transaction records since 2020, using PostgreSQL for high-throughput ingestion, and I’ve automated daily scraping with a GitHub Actions workflow to store data in a Hugging Face dataset. I also evaluated NEPSE SMA, EMA, and MACD strategies and used stationary bootstrap methodology for inference, finding statistically significant predictive ability across all three.

Earlier, I supervised contract enumeration work for the National Economic Census—extracting geographic coordinates, converting them into OpenStreetMap-compatible polygons, validating over 3,700 questionnaires, and supervising eight enumerators across 79 enumeration areas. Before that, as a Research Intern at Nepal Rastra Bank, I conducted ARDL analysis of interbank rate transmission to lending and deposit rates and developed a Scrapy web scraper to collect over 14,000 real estate listings for housing market research.