I am using daily data from Yahoo for ETFs. I was wondering why is there only the adjusted close price and not adjusted open, high and low? I am asking this because if I make use of the other prices not adjusted, the results of the backtest will be wrong.
I use Yahoo data for ETFs and do not use their adjusted close. I account for dividends separately from the closing price. Also, in the case of splits you will need to adjust previous prices. For example, if there is a 2 for 1 split, you need to take all of the pre-split OHLCV data and divide by 2. fan27
fan27 that's a lot of work to do across the entire market to keep it accurate.... and it's not just dividends and splits - there are lots of other events that occur too (many of which are poorly done or missing on Yahoo). I tried your approach years ago and was overwhelmed by the amount of events and also all the ones that are missed in the free sources. Your volume adjustment is wrong too - that needs to be multiplied by 2 instead since you've halved the price data. If you value your time there are data vendors out there that do all of this for you for less than the cost of a coffee a day and if you don't understand what you're seeing they will provide you support (well, at least the good ones will do so).
Ok so it's known that even daily prices are not good. Where do you suggest to buy good quality daily data for american ETF?
@jharmon, thanks for the correction on the volume adjustment. And yes, spending money to get these issues handled for you is likely the best approach for most. fan27
I've used a few data vendors over the years.... For daily data I have settled on using Norgate's data with Amibroker since it also provides delisted stocks. Their new version is due out this year - i'm testing it right now and it's solid. For live data, I use NxCore but that's pretty complex (and not cheap).
Norgate's data looks interesting. Regarding NxCore: does it make a difference to have such a low latency system compared to in for an intraday/interday strategy?
With NxCore I can be assured of nothing missing. It's not just about latency but I'm not HFT though. I don't have any confidence in any of the others I've used (they miss various things in various different ways, or are too clunky to allow accurate integration). I've tried/paid lots for daily data... various databases on Quandl, eoddata, Worden's TC2000, CSI, Reuters Datalink and lots of other "free" sources, before settling on Norgate. For live, I've used: IB's feed, eSignal, IQFeed, Money.net, Reuters 3000xtra (now called Eikon I think). Never did get around to bloomberg due to their crappy multi-year term (2 yrs and 60 prior days notice).