Hi everyone. I\u2019ve been building a historical research dataset for low-float\/small-cap stocks and I\u2019m running into a much bigger data-coverage problem than I expected.<\/p>\n
The general universe I\u2019m looking at is:<\/p>\n
- \n
- Stocks with fewer than roughly 20M shares in float.<\/li>\n
- Lower-market-cap names.<\/li>\n
- At least 1M shares of daily volume.<\/li>\n
- Dates from 2023 through 2025.<\/li>\n
- Premarket, regular-session, and ideally postmarket one-minute data.<\/li>\n
- Delisted, renamed, merged, and inactive tickers included.<\/li>\n<\/ul>\n
So far, I\u2019ve been able to build a fairly large list of eligible ticker-days and recover point-in-time information for a lot of old and delisted names. I\u2019ve tested data from Alpaca SIP and Massive\/Polygon, but the historical minute data is much more incomplete than I expected.<\/p>\n
The issue is not just whether a ticker exists in the database. I need enough complete data to reconstruct:<\/p>\n
- \n
- Premarket high and volume.<\/li>\n
- Regular-session volume and VWAP.<\/li>\n
- Intraday highs and lows.<\/li>\n
- Technical setups around PMH breaks and pullbacks.<\/li>\n
- Whether a theoretical entry reached a stop or 2R target first.<\/li>\n
- Missing trading minutes versus minutes where no trades actually occurred.<\/li>\n<\/ul>\n
In my broad initial universe, only around 1,600 out of roughly 293,000 eligible ticker-days had completely stored paths under the original strict definition. I\u2019ve improved coverage using multiple providers, but there are still thousands of older, renamed, or archive-risk ticker-days where the data is partial or completely missing.<\/p>\n
I\u2019ve also found that mixing feeds can create problems. For example, bars from a direct-feed composite may look valid on their own but have very different volume from consolidated SIP data because off-exchange and FINRA TRF prints may be missing. Since my study uses volume, VWAP, liquidity, and dollar-volume features, I can\u2019t just patch missing days with a different feed unless the data is reasonably comparable.<\/p>\n
Has anyone built something similar and found a provider that offers:<\/p>\n
- \n
- Historical one-minute bars going back several years.<\/li>\n
- Full premarket and postmarket coverage.<\/li>\n
- Raw\/unadjusted price and volume.<\/li>\n
- Consolidated SIP or TAQ-derived data, including FINRA\/TRF volume.<\/li>\n
- Delisted and inactive securities.<\/li>\n
- Historical ticker changes and point-in-time symbol mapping.<\/li>\n
- A flat-file or bulk-download option so I can save the data once and reuse it.<\/li>\n
- Coverage where an empty response can be distinguished from a genuinely inactive\/no-trade day.<\/li>\n<\/ul>\n
I\u2019m open to a paid provider or one-time historical download. I\u2019ve looked at Alpaca, Massive\/Polygon, Databento, and IBKR so far. Databento\u2019s direct feeds are interesting, but the feeds visible to my account were missing several exchanges\/reporting facilities, so the volume was not comparable to SIP. IBKR also appears to filter some historical trades and may not retain data for securities that are no longer trading.<\/p>\n
I\u2019d especially appreciate feedback from anyone who has used AlgoSeek, Tick Data, NYSE TAQ, LSEG Tick History, or another provider for old low-float and delisted stocks.<\/p>\n
I\u2019m not looking for real-time alerts or a trading platform. I\u2019m trying to build one clean historical research dataset that I can download, validate, and keep locally.<\/p>\n<\/div>\n
submitted by \/u\/metalmania7778 <\/a>
[link]<\/a><\/span> [comments]<\/a><\/span><\/p>
