Skip to content
All library documents

Why One-Minute Market Data Has Uneven Sample Counts

Article Quant Q&A · Author: user2647513
Read the original

Summary

The document explains why one-minute price-history downloads for different symbols can contain different numbers of observations, even when requested with identical settings and an assumed regular trading window. The key clue is the timestamps: the returned records are not evenly spaced. A one-minute interval setting therefore does not guarantee a quote for every minute of the session; the count can reflect when observations are actually present in the source data.

The response points readers to inspect the dataframe’s date index to verify the timing pattern. It does not provide a detailed account of AlphaVantage’s collection or filtering rules, nor does it establish why particular symbols have more records than others. Researchers should check timestamp gaps and the provider’s data definition before treating the series as equally sampled or comparing raw counts.

Key ideas

  • An interval setting does not necessarily mean every interval has a recorded quote.
  • Different symbols can have different counts when their timestamp sequences contain gaps.
  • Inspecting the date index helps reveal whether observations are evenly spaced.
  • Raw sample counts should be interpreted in light of the source’s recording conventions.

Tags

EquitiesStatistics
Full text
# Why do 1min intraday price histories from AlphaVantage have different sample counts?


# Why do 1min intraday price histories from AlphaVantage have different sample counts?












I've downloaded intraday price histories from alphavantage for a dozen or so symbols, with the same parameters (1 minute interval) for each. Presumably this data represents prices in one minute intervals over a constant intraday window (9:30AM - 4:00PM ET). Why then do the number of quotes returned in the API responses vary for each symbol, from a low ~3000 to a high of ~10000?

## Answer by user2647513 (score 1)

https://quant.stackexchange.com/a/61887

I hadn't noticed when asking, but the response dataframe has a 'date' index, and the timestamps are indeed not evenly spaced!

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.