I would like to implement this strategy but I have no idea about coding or how would I even become doing all those things, I’d love to get some feedback on how can I start comprehending what’s even going on here!
Hi Charles! I just sent a note to Vincent, the maintainer of Sharadar Core US Equities Bundle (the dataset I'm using and recommend; check it at https://data.nasdaq.com/databases/SFA), so we can collaborate to write a post about the dataset! Thanks for the suggestion! Cheers
Yes, I've heard great things about Norgate data. Unfortunately, as far as I understand, it's Windows-only, and/or you cannot extract the data from their system to manipulate directly... (which makes it a no-go for Linux/Mac users and users who want to do more custom manipulations, machine-learning, etc...)
Thanks for the tip, Scott!! Based on your successful experience extracting the data out of the system, I will give it a try! I actually contacted their support and asked specifically about it, and their support said it was not possible... but I will give it a try... thanks again!
Great article, thanks for sharing. I found what seems like a legit article suggesting trading the whole S+P 500 would incur an average spread of 4.5bp. It's cap weighted, so will be skewed to tighter NBBO's associated with larger caps. In the real world if we had to hold 3 positions randomly qualified from the universe of 500, this suggests a bare minimum NBBO of 5bp. What would a 5bp spread do to your CAGR? Thanks!
1. Are you rebalancing whenever you close out one ore more active positions (for profit or loss)? If yes, when one position drops out do you just replace it with the highest ranked (measured by volatility) equity that meets the defined criteria?
2. If the strategy rebalances one position at a time did you happen to analyze the win rate for all trades that became available in the S&P 500 universe?
I would like to implement this strategy but I have no idea about coding or how would I even become doing all those things, I’d love to get some feedback on how can I start comprehending what’s even going on here!
All best
Thanks a ton for sharing all this hard work. I'm so glad I found your substack 🙌
I'll be trying to recreate some of these in my research environment (starting with this one here) and will share my progress.
Nice work; regime structure likely gating the alpha waves
Thanks for the great article.
I was wondering how you maintain a survivorship bias-free dataset. If you can, could you do a post on this?
Thanks!
Hi Charles! I just sent a note to Vincent, the maintainer of Sharadar Core US Equities Bundle (the dataset I'm using and recommend; check it at https://data.nasdaq.com/databases/SFA), so we can collaborate to write a post about the dataset! Thanks for the suggestion! Cheers
Norgate data is best in class solution
Yes, I've heard great things about Norgate data. Unfortunately, as far as I understand, it's Windows-only, and/or you cannot extract the data from their system to manipulate directly... (which makes it a no-go for Linux/Mac users and users who want to do more custom manipulations, machine-learning, etc...)
I have extracted the historical index database into other platforms, which requires custom work, but typically I use it within AmiBroker
I have also used the Sharadar database, which is OK although not as elegant.
Thanks for the tip, Scott!! Based on your successful experience extracting the data out of the system, I will give it a try! I actually contacted their support and asked specifically about it, and their support said it was not possible... but I will give it a try... thanks again!
If you want to chat live, I can describe how to do it.
I like that painting.
I love it too! :)
https://www.nasdaq.com/articles/sampling-sp-500-minimize-spreads
Great article, thanks for sharing. I found what seems like a legit article suggesting trading the whole S+P 500 would incur an average spread of 4.5bp. It's cap weighted, so will be skewed to tighter NBBO's associated with larger caps. In the real world if we had to hold 3 positions randomly qualified from the universe of 500, this suggests a bare minimum NBBO of 5bp. What would a 5bp spread do to your CAGR? Thanks!
Thanks! And thanks for sharing this article!
Considering 5bps of slippage instead of 1bp in each trade reduces the annual CAGR to 17.4%.
Cheers!
Great write up! A couple of questions:
1. Are you rebalancing whenever you close out one ore more active positions (for profit or loss)? If yes, when one position drops out do you just replace it with the highest ranked (measured by volatility) equity that meets the defined criteria?
2. If the strategy rebalances one position at a time did you happen to analyze the win rate for all trades that became available in the S&P 500 universe?
Thank you.
you trade stocks instead of spx, right?
yes
What time period do you use when computing the Normalized ATR?
TA-Lib's default value (14 if I'm not mistaken)