overnight returns vs. day returns

[jim leahy]

today in tg1, someone mentioned that most of the market returns occur overnight. i read this study
last year and decided to prove or disprove it myself. as with all market performance statistics, it depends
on when you start measuring. my conclusion is it's true and not true. i couldn't locate my spreadsheet
during the meeting but i'm including it here for those who want to slice and dice it further.

i used spy for the market returns and started my analysis jan 2000. i don't recall the time frame in the article,
but 2000 seemed like a good starting point to me. from 2000 till 2018 it's true. practically all the market
returns happened overnight. but things changed in 2009. starting then it's been somewhat mixed.

the below graphs separate ovenight performance and day only performance. the red line is overnight
and the blue is from open to close during the day the first graph starts in jan 2000, the second starts
in 2009.

 

[AndrewS]

Very nice job Jim. 2009 is around the time trading became dominated by computers and I wonder if that has something to do with the change in returns?

 

[jim leahy]

thanks, andrew.

good point, that is very well possible. i watch overnight quotes a lot more these days
but it doesn't help with options trading. you can trade spy, qqq, and a few other etfs
after hours as a possible hedge, though.

 

[jim leahy]

i saw another study related to this, now measuring the overnight gains vs. day gains during the pandemic.
the authors noted that since april, all of the gains have been outside regular u.s. trading hours.
i did my own analysis using spy as a proxy for the market, and again, it's true, but also it depends on where
you start measuring performance. i went back to jan 2020 and looked at the performance of the separate
time frames. if you look before april, all of the losses were in after-hours trading as well.