US stock end-of-day data with official open/close?

Discussion in 'Data Sets and Feeds' started by linear, Jul 12, 2018 at 1:09 AM.

  1. linear

    linear

    Does anyone know of a source for historical US stock end-of-day (OHLCV) data where the open/close prices are the official prices from the listing exchange (i.e., the prices from the opening/closing auction)?

    Ideally I'm looking for a source that includes delisted stocks and has a 20-30 year history.

    I'm currently using Norgate as a data provider and they fit the latter two criteria, but they use first/last regular hours consolidated trade as their opening/closing price. CSI seems to be the same.
     
  2. jharmon

    jharmon

    I also use Norgate for my end-of-day data on US stocks and asked them a similar question a while back.

    They said that the close is not actually the last regular hours trade, but is the official close as determined by the Consolidated Tape Association. They also said that it's the closing auction from the listing exchange except where there is no closing auction from that exchange, then it's the last consolidated price. The open is the first regular hours tape price and the CTA has no rules about the listing exchange.

    If a stock doesn't have a closing auction it's probably too thinly traded anyway and should be avoided.

    So, the outcome from all of that is it's only the open that is something to be concerned with.

    I also have tick data from NxCore that allows me to isolate exchanges and trade types so I can back-test on MOO orders (which only work on the listing exchange). I found that MOO orders on stocks that were thinly traded were quite different to the consolidated tape market orders, but that was just a function of the time difference between the actual market opening and when the listing exchange decided to open.

    The specialist/market maker driven exchanges (NYSE, AMEX or whatever they call it now) can delay their opening auctions for a few minutes until some dude (specialist or whatever they call it these days) thinks it's time to ring the opening bell on that stock - probably when they are already net positive for the day. Seems like a rort to me.

    My net outcome here was to trade liquid stocks (eg. S&P 1500). The difference between consolidated tape open and listing exchange open was just intra-day noise and didn't have any significant performance outcome on my daily trading systems.

    That's just my story though - but be wary of any system that dramatically changes as a result. I've developed enough "this system will beat the market by 150%" systems to realize that trading noise is is futile.
     
    d08 likes this.
  3. Yes, exactly. Most of the "high quality" US markets eod data feeds that contain unadjusted data, will provide exactly the same close print. As pointed above, that would be the official close. In general there is no divergence here.

    The problem will always be the opening price no matter what data feed provider you use.

    There are many ways of solving this. For example:

    1. If you have tick data, then you could filter for the primary exchange opening print and use that price.

    2. Or you could simply average the trade prices during a random opening interval weighted by the trading volume.

    3. Another way would be to subscribe to the official eod data from the relevant exchanges you are interested in, say NYSE, and use the open price they provide for their listed symbol.

    This is really not a trivial solution and the real open price ( read entry price ) you use is very much dependent on your own trading strategy, whether your signal is at market open or intraday. So in the end you will still need to adjust for slippage and the rest.

    And when you are finally done with this, then you will still need to decide what to do with the corporate actions adjustments. Do you only adjust for splits? Dividends? How about spin-offs? etc.

    So, in short, I very much doubt you will ever find a provider that will satisfy/guarantee the data quality you assume.

    Hope it helps.
     
  4. jharmon

    jharmon

    Yep - MOO orders are only available for the listed exchange. Due to the delays on open, I don't use MOO orders at all anyway.

    Corporate actions are interesting - my provider (Norgate) does adjust prior to open. Most others don't in my experience. I adjust my orders prior to open for splits and dividends. Others like CSI and Yahoo only provide this information after the close (way too late). For certain ETFS and ETNS that pay infrequent capital gains that I didn't pick up on, this has saved my bacon a few times.
     
  5. katie_h

    katie_h

    I don't know if I can help you but all sources I've tried have had some "unofficial" values to say the least. I guess it depends on what exactly you're using the data for but I wouldn't worry about it too much.
     
  6. Last edited: Jul 12, 2018 at 10:09 PM
  7. jharmon

    jharmon

    CSI is consolidated data only. Used them years ago, but has various other issues.
     
  8. linear

    linear

    Well, I really ought to have looked more carefully before I made claims about Norgate's data. I just went back and checked and their close prices do match the closing auction prices. It's only the open prices that are different.
     
  9. I don't know if this is true. The close may be the same as the auction most of the time (especially for Nasdaq primary stocks due to their precise timing) but I think it won't be the same in general. Did you check thousands of symbols/days or just a few?
     
    Last edited: Jul 13, 2018 at 2:17 AM
  10. linear

    linear

    I compared the prices from 8567 MOC trades I'd made since 2017 to the close prices from Norgate and found only 2 differences. The trades at mostly on Nasdaq, but about a quarter were on NYSE and for a fair number of those ones the auction didn't complete until a little while after 4pm.
     
    cruisecontrol likes this.