The Stock Market Works by Day, but It Loves the Night

I downloaded historical SPY from Yahoo Finance going all the way back to 1993. Backtested buying one share at close and selling back that share at open the next day. After 25 years, I netted 0% profit, not counting commissions and slippages.

To be exact:

Total cost of purchase: $788,966.24,

Total receipt of sale: $788,983.23.

Total profits: $16.99.

What am I missing? Where is the free lunch?
One share is your problem. What software are you using?
 
I use index futures contracts. There is no free lunch. Anything worth while takes much effort. If trading where easy everyone would be doing it. Stock index futures rise overnight with great regularity. You need to figure out the rest.
 
There is some potential here. Big moves are made overnight.

I would look at when the model hits big up or down and see what that looks like.

I know you are saying you can not repeat the outcome. You would have to look at the study and determine how they selected their data. It shouldn't make a big difference but getting the opening price is not straight forward.
 
Maybe its not so cut and dry as the article states. Lets try this, for the rest of the month I will post the close and open. So lets get started, 20 min till cash close, start building your position.

Fridays close scale in 7-12-19.jpg
 
I downloaded historical SPY from Yahoo Finance going all the way back to 1993. Backtested buying one share at close and selling back that share at open the next day. After 25 years, I netted 0% profit, not counting commissions and slippages.

To be exact:

Total cost of purchase: $788,966.24,

Total receipt of sale: $788,983.23.

Total profits: $16.99.

What am I missing? Where is the free lunch?

Using yahoo finance SPY prices adjusted for splits and dividends buying a single share at the close (starting Jan 29, 1993), selling at the next day's open (last sell July 11, 2019) without accounting for slippage or commissions, some statistics were:
numValues 6659
sum 273.537898999999
min -9.54807600000001
max 5.866806
mean 0.0410779244631325
sampleStdDev 0.696582829813863
median 0.0391690000000011
medianAbsDev 0.253011000000001
skewness -0.884340196556304
excessKurtosis 16.3349982402817
>Thresh_0_Pct 56.21


It backtested profitably, but an average profit of 4.1 cents per share might have been consumed by slippage and commissions.

Buying a single share at the close on Jan 29, 1993 and selling at the open on July 11, 2019 using the adjusted prices would have resulted in a profit of 272.613250 or a compounded annual growth of 9.57 percent.
 
Using yahoo finance SPY prices adjusted for splits and dividends buying a single share at the close (starting Jan 29, 1993), selling at the next day's open (last sell July 11, 2019) without accounting for slippage or commissions, some statistics were:
numValues 6659
sum 273.537898999999
min -9.54807600000001
max 5.866806
mean 0.0410779244631325
sampleStdDev 0.696582829813863
median 0.0391690000000011
medianAbsDev 0.253011000000001
skewness -0.884340196556304
excessKurtosis 16.3349982402817
>Thresh_0_Pct 56.21


It backtested profitably, but an average profit of 4.1 cents per share might have been consumed by slippage and commissions.

Buying a single share at the close on Jan 29, 1993 and selling at the open on July 11, 2019 using the adjusted prices would have resulted in a profit of 272.613250 or a compounded annual growth of 9.57 percent.
Mine started on 1/29/1993 and stopped on 5/21/2019. I think our results agreed.

IMHO, it is much more profitable just buy and hold SPY - the real free lunch without effort.
 
Back
Top