Back to Community
Widespread splits related data integrity issues

There seems to be widespread splits related data integrity issues with some fundamental indicators that are computed on the per share basis. In particular, I was focusing on morningstar.valuation_ratios.forward_earning_yield and morningstar.valuation_ratios.pb_ratio. I have inspected several stocks around split dates and discovered that in all but one cases, there were forward_earning_yield and/or pb_ratio jumps a couple of days after the split. Abnormal values are consistent with the split ratio (i.e. normal values can be obtained by forward_earning_yield/split_ratio and pb_ratio*split_ratio). I hope that Q folks can fix this.

Loading notebook preview...
Notebook previews are currently unavailable.
6 responses

I've seen the same issues. It really creates havoc with an algorithm if one is trading on these fields.

Can the data be cleaned up?

Nobody else cares about huge abnormal jumps in key fundamental metrics?

U should send this to support directly so that they put it in their list of bugs

Unfortunately, Quantopian requires you to share ALL notebooks with support in this case which is something I wasn't ready to do. Actually, I am quite surprised that nobody cared about it. I had another post showing that other key fundamental metrics are also way off. Got no response as well.

I believe adding tags such as "Show Q", "using data" and "seeking help" to the post might increase the chances of a Q reply.

I would imagine if more people used the fundamental data there would be greater outcry; believe most of the algos revolve around mean reversions and technical analysis of sorts.
You could try sending a direct email to [email protected] I have sent emails there before and received a prompt response. The problems I observed were with incorrect stock pricing of a relisted ticker symbol after having been delisted. Problem stocks include: PVAC, MXIM, ITWO, HERO and GGP. So I have an exclusion list for those stocks because of this and others that throw up errors when using fundamental analysis.
Have a good day.