Info is inaccurate.

[FrostBid]

I feel like the data is somewhat inaccurate for certain years. The csv file listed AAL as a component in 1996, but according to Wikipedia, the company is only added to the index in 2015.

[fja05680]

Hi FrostBid,

1996 data is part of the original data from I got from "Trading Evolved". I have no good way to check if his data is 100% correct. In the description of this project, I suggested that one could use the data from 2001 onward because of the low symbol count for the first 5 years. I'll try to take a look at AAL this weekend. I could deal with this specific case, but generally, I have no way to verify the correctness of the original list. I assume it must be fairly accurate because this is what Andreas Clenow uses for his professional trading.

[bxclib2]

this AAL has been confusing me a lot. I even ask ChatGPT whether a stock symbol could be reused if a company is delisted.

Btw, may i know do you know the meaning of the numbers in his original CSV? such as ETS-200603, what 200603 means? I am so confused...

[fja05680]

I believe that indicts when the symbol gets removed from the index.

[bxclib2]

In fact this is same as what I initially thought but I ´ve checked for some symbols it is not correct so this is a wrong guess

[fja05680]

Ok, thanks for letting me know. I filter those out as you know. I would suggest just selling when assemble disappears that you have a long position in. If the symbol has been removed from the index, then the position needs to be closed. If it was a mere symbol change, and the company is still in the index, you should still sell. Your algorithm should immediately buy it with a new symbol name.

…

On Thu, Sep 21, 2023, 2:50 AM bxclib2 ***@***.***> wrote: In fact this is same as what I initially thought but I ´ve checked for some symbols it is not correct so this is a wrong guess — Reply to this email directly, view it on GitHub <#9 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACD3KHWKQ2IVBZWIXPP7LFLX3PPUVANCNFSM54RJJJ6A> . You are receiving this because you commented.Message ID: ***@***.***>

[swisstackle]

@fja05680 First of all, thanks for your work.
Did I understand it correctly that you filter out symbols that have weird names such as ETS-200603?

[JohnSparks2024]

The 200603 is the last date for which that symbol is valid for that company. It is how Norgate deals with symbol changes (whether due to merger, delist or simple ticker change). This is separate from whether or not the stock is still a member of the S&P 500. Norgate maintains that information in a separate table.

[fja05680]

@swisstackle Yes, they get filtered out. And John is correct.

[NinoBach]

Im looking at Reuters Refinitiv data and there is a similar problem. American Airlines has the RIC AAl, and before that, Aon Grp had the RIC AAl.
To be fair, they have different suffixes, but it's still confusing, before previously I thought I can derive the ticker from the RIC by deleting the suffix, but that won't work.

[ysaffar]

In 1996, AAL referred to Alaska Air Group existed from 1980, not American Airlines (ChatGPT 4o). Since then Alaska Air Group changed to ALK.

But this presents a problem when trying to match data to price date from delisted tickers because they'll show ALK price going back to 1980 (e.g. macrotrends). Not sure what to do about this. I feel like there shouldn't be too many examples like this so if we find them all that'll fix it.

[fja05680]

ysaffar, That's an interesting issue I had never considered. I'm not to sure what can be done abut this either. It might just need to be handled by at the application level. Like you said, there shouldn't be too many of these.