Validating Adjusted Close Prices Against Close Prices
Summary
The document raises a market-data quality problem: historical OHLC prices from low-cost providers may look plausible even when adjusted close values are erroneous. It asks how to screen adjusted close against close, citing an example of a negative adjusted close and considering whether a fixed ratio threshold could flag implausible values. The author recognizes that dividends and stock splits can legitimately create large differences, so a simple cap on the ratio could reject valid data.
A more complete validation approach would use dividend and split records to recompute adjusted prices, but the author describes practical drawbacks: multiple provider queries per ticker, cost, and the possibility that the providers’ corporate-action calendars are also wrong. The document therefore frames a trade-off between inexpensive plausibility checks and more involved reconciliation against corporate actions. It does not provide a tested threshold or a recommended rule, so it is useful as a statement of the validation challenge rather than a resolved method.
Key ideas
- Adjusted close can be erroneous even when open, high, low, and close values appear plausible.
- A fixed threshold for the adjusted-close-to-close ratio can flag suspicious values but may also reject valid split adjustments.
- Dividend and split histories can support independent recomputation of adjusted prices.
- Corporate-action data from the same provider may share errors with its price data, limiting validation confidence.
- The document poses the data-quality problem but does not establish a tested screening rule.
Tags
Full text
# Possible validation rule to check adjusted close values from close # Possible validation rule to check adjusted close values from close Since I can't afford Reuters or Bloomberg, I use two different free/cheap market data providers (yahoo finance and financialmodelingprep.com) to retrieve historical stock prices. Since I have noticed that the data is sometimes total crap, I have implemented different validation rules (e.g. Low<=Close<=High) and if one provider violates these rules, I go to the other one and if it also violates them, I exclude the ticker. Unfortunately, I have now noticed that there are also tickers where the normal OHLC prices look clean, but the Adj. Close is completely implausible (for the ticker 'NURAF', for example, yahoo finance even reports a negative Adj. Close). I would now like to extend the logic to checks on close vs. adj. close. Could anyone tell me some good rules that I can use to decide as reliably as possible whether the adj. close might be false? The only idea I can think of is something like adj. close / close <=10%, which certainly makes sense in the case of dividends, because no company pays out more than 10% dividends. However, in the case of a 10:1 split, for example, this obviously makes perfectly sense. The best way would of course be to query the dividend and split calendars as well and calculate the adj. close prices by myself. Unfortunately, this is programmaticaly very complicated and also I would need at least 6 api calls per ticker (2x price, 2x dividend calendar, 2x split calendar), which I also can't afford (I know I'm very poor :)) I also think this option is methodologically weak, because I can only query the calendars from the same two providers from where I get the prices, and I have thus no guarantee that the calendars themselves are correct. So I would only screen technical issues in the processing. Could anyone help me out this issue? Thanks in advance!
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.