{
 "title": "StockQL blog \u2014 practical market-data and backtesting guides",
 "description": "Practical guides to backtesting, market-data quality, stock screening, and research workflows with StockQL.",
 "url": "https://stockql.ai/blog.html",
 "slug": "blog",
 "published": null,
 "site": "StockQL",
 "contact": "hello@stockql.ai",
 "body_text": "StockQL blog\n\nPractical market-data research\n\nRunnable guides for building screens, testing strategies, and extending market data with tools you control.\n\nOctober 22, 2026 \u00b7 Research methods\n\nHow to backtest earnings-call signals with StockQL\n\nSee how an audited transcript split, fixed point-in-time panel, post-reaction returns, and issuer-aware uncertainty produced three bounded nulls and one rejected proxy\u2014and left a reusable method for the next hypothesis.\n\nRead the guide \u2192\n\nSeptember 13, 2026 \u00b7 Provider tradeoffs\n\nWhen is a stock really delisted?\n\nSeparate missing prices, provider availability, ticker reuse, and formal delisting evidence. The comparison shows why zombie prints and premature removals require different safeguards.\n\nRead the article \u2192\n\nSeptember 13, 2026 \u00b7 Provider tradeoffs\n\nDead ETFs are part of the backtest\n\nSee how erased fund histories create survivorship bias even when a provider handles current delistings well, and preserve the universe that existed at each simulated date.\n\nRead the article \u2192\n\nSeptember 13, 2026 \u00b7 Provider tradeoffs\n\nSplit-adjusted prices can change after you ingest them\n\nMap raw and adjusted field semantics, detect stale corporate-action bases, and refresh the full history before a false return enters a screen or backtest.\n\nRead the article \u2192\n\nSeptember 13, 2026 \u00b7 Provider tradeoffs\n\nMarket-data coverage is not one number\n\nCompare providers by instrument, venue, listing freshness, retained history, and usable fields instead of collapsing unlike research domains into one percentage.\n\nRead the article \u2192\n\nSeptember 13, 2026 \u00b7 Provider tradeoffs\n\nTwo matching fundamentals feeds can still be wrong\n\nSeparate freshness, accounting semantics, and source independence. A filing-level check explains why identical commercial rows are not always verification.\n\nRead the article \u2192\n\nSeptember 13, 2026 \u00b7 Provider tradeoffs\n\nToday\u2019s bulk market-data file may not be final\n\nDesign daily ingestion for mutable snapshots, late real rows, placeholder padding, and explicit-date rechecks instead of treating one successful download as complete.\n\nRead the article \u2192\n\nSeptember 13, 2026 \u00b7 Provider tradeoffs\n\nA ticker is not an earnings transcript\u2019s identity\n\nVerify issuer lineage across acquisitions, ticker reuse, fiscal-label differences, and shared transcript upstreams before text enters a research corpus.\n\nRead the article \u2192\n\nSeptember 12, 2026 \u00b7 LLM research\n\nHow to categorize earnings call transcripts with an LLM\n\nTurn recent calls into evidence-grounded demand, pricing, margin, and capital-spending labels. The guide includes a downloadable classifier, explicit cost and quality controls, and a PostgreSQL screen over the results.\n\nRead the guide \u2192\n\nSeptember 12, 2026 \u00b7 Stock screening\n\nHow to screen Japanese net-net stocks with SQL\n\nBuild an auditable JPX screen for below-NCAV, Graham-style, double-net, and triple-net candidates. The guide covers date-aware financial statements, current market capitalization, liquidity, cross-source checks, and results from the exact query on a live StockQL database.\n\nRead the guide \u2192\n\nSeptember 11, 2026 \u00b7 Backtesting\n\nHow to backtest a trading strategy with StockQL\n\nBuild a 200-trading-day trend strategy directly in PostgreSQL. The walkthrough covers signal timing, transaction costs, drawdown, annual returns, CSV export, and results from the exact queries on a live StockQL database.\n\nRead the guide \u2192"
}