Flipping stock prices.

Is someone using flipped price level data for doubling amount of historical data?
Did it prove to be beneficial in out of sample results or is there some fundamental reason data would be wrong if inverted?

 

If your method plays longs and shorts easily then surely same result?

 

I noticed worsening of results with all algos for some reason.

 

I've no idea what "flipping level" means. The historical DAILY data I have, same as what you get for free on YahooFinance, lists open, close, low, high and adjusted (for splits and dividends). I use close-to-close in my calculations. Have experimented with close-to-open and then open-to-close to double the amount of data but the figures didn't seem to change (for better or worse) so I reverted to the simpler approach.

 

Few years ago i collected all availible yahoo data and tested.
Its relatively poor quality.
Lot of strange edges there, but lots of data.

By flipping i mean reverting price level data.
100,104,105 to 100,96,95...
Thought it may balance out positive long term bias of markets.

 

Strange thing to do, I don't see a point to it.