I've found a problem recently that I couldn't find a thread on but was curious as to what was causing it. With nearly every stock I'm backtesting, if I stay below a 3 Minute Time Frame the close price for the last bar will be different than if I do a 3 Minute time frame or greater.
Eugene: copied from a preceding thread:
I'm getting some weird results when I use a one minute time frame. If I look at the data for the SPY plotted on a chart, for example, it shows the following information for the 4PM bar of 1/19/2016
Open: 188.11
High: 188.19
Low: 187.87
Close: 185.13
I'm not understanding how it can have a close below the low. I have to take the time frame to 3 minutes before it appears to be corrected. I see it on a lot of the close bars.
Size:
Color:
QUOTE:
if I stay below a 3 Minute Time Frame the close price for the last bar will be different than if I do a 3 Minute time frame or greater.
Different from what exactly?
QUOTE:
following information for the 4PM bar of 1/19/2016
If you have a problem with Fidelity data quality, then please contact them directly. As we're a 3rd party resource, data and login issues is something beyond our capability.
Size:
Color:
Size:
Color:
I'm not sure if it's a problem with the data quality or not, that's what I'm trying to determine. You'll note the above was an example I found. I've attached a pair of screenshots. The first is what the end of the day looks like with a one-minute interval. The second is with a 3 minute interval.
If it's an issue with the data then I understand and I will switch data sources. I just wanted to make sure it wasn't something I was doing.
Size:
Color:
In this case the 1-minute OHLC bar indeed looks wrong. If "Reload chart history" (right-click on 1-min SPY chart) doesn't fix it, you'd want to contact Fidelity.
Size:
Color:
Thanks Eugene! Much appreciated.
Size:
Color:
It's a curious case because I don't see that closing value in my Fidelity chart for 1/19/2016. Data are subject to change and corrections, and updating intraday will check for corrections for a minimum of the last day already in the cache (at least 3 days for daily data).
Anyway, you have two options:
1. Reload the data (takes several minutes for a 1-minute chart
2. Double click the bar and correct it yourself.
Also, make sure to run the "Bad History Data Check [Rev A]" script before doing serious backtesting. I'm not sure why, but Fidelity data for some symbols tends to become "corrupted" over time (especially missing several days or weeks). Follow the instructions in the script's Description.
Size:
Color: