All market gains since 1993 have occurred after hours

[cruisecontrol]

Known about this for almost a decade now. There was a guy that ran a service based on it called Overnight Edges

https://web.archive.org/web/*/http://overnightedges.com/

It is a legit edge but I don't personally trade it as the sharpe is too low.

 

[sle]

Mkay, I am bored stiff (stuck at home with a sprained ankle) so I just ran it on SPYs:

Sharpe 0.82
Daily Vol 0.62%
Median 0.05%
Mean 0.03%
Min -8.69%
Max 5.89%
% Up 54.55%
Total 210.37%
Worst DD -42.71%

if you only run it over the weekend, there is some improvement in PnL/trade but your totals get worse and stuff

Sharpe 1.07
Daily Vol 0.68%
Median 0.06%
Mean 0.05%
Min -8.58%
Max 5.89%
% Up 55.84%
Total 56.18%
Worst DD -18.84%

It's in log space, so don't be alarmed that it does not add up to your "600%"

 

[sle]

Known about this for almost a decade now. There was a guy that ran a service based on it called Overnight Edges

https://web.archive.org/web/*/http://overnightedges.com/

It is a legit edge but I don't personally trade it as the sharpe is too low.

It's less about Sharpe and more about PnL/trade, IMHO. I'd gladly add a 0.5 Sharpe strategy that has a t-stat of like 6.5, but it pisses away most of the theoretical alpha to execute

 

[cruisecontrol]

It's less about Sharpe and more about PnL/trade, IMHO. I'd gladly add a 0.5 Sharpe strategy that has a t-stat of like 6.5, but it pisses away most of the theoretical alpha to execute

What is the sharpe or t-stat for buy and hold? Once you factor in commissions, taxes, execution risk, I doubt this strategy really comes out ahead enough to justify it. Unless you have other predictive edges on the market that you are filtering it with (like the overnight edges guy), then it makes more sense to me.

EDIT: would be interesting to compare overnight trading to buy and hold spending an equal amount on protective puts as all the commissions for the daily entries/exits.

 

[TheBigShort]

Red is normal Log return.
Blue is Log Intraday return
Green is Log Gap Return
Data since 2007

I attached non-cummulative log returns if anyone wants to do further analysis. But It looks like a clear winner!!

 

[krugman25]

Red is normal Log return.
Blue is Log Intraday return
Green is Log Gap Return
Data since 2007

I attached non-cummulative log returns if anyone wants to do further analysis. But It looks like a clear winner!!

View attachment 200747

This is a great addition to the conversation, thanks!

 

[sle]

What is the sharpe or t-stat for buy and hold?

Sharpe 0.5, so this thing is somewhat better. It's correlated to buy and hold, so that would make it less exciting, even if PnL/trade was OK.

 

[TheBigShort]

This is a great addition to the conversation, thanks!

np, I'd calculate the risk metrics for you, but I am a little busy right now. I'll do so tomorrow, if no one else does it.

 

[Kevin Schmit]

It's in log space, so don't be alarmed that it does not add up to your "600%"

Still doesn't match. exp(2.1037) = 8.2, so 720%, which is a long way from from 600%.

...it pisses away most of the theoretical alpha to execute

Execute on Robin Hood! I don't think Robin Hood traders' trades affect the opening and closing auction price, so zero execution costs.

 

[TheBigShort]

This strat looks kind of interesting. Leveraged 2x overnight (long), short 1x intraday.
red = SPY
blue = Strategy