If the long and the short positions are highly correlated and very similar (eg SPY and VOO), the net result will be in the vicinity of zero, regardless of what the market does. What one leg makes, the other leg loses.
What you are suggesting requires perfect timing, closing the winning position with the expectation that the losing position recovers, offering the possibility of recovering your paper losses. If the losing leg doesn't reverse then you have booked a gain while accruing an equivalent paper loss and now you are going to accrue additional losses on the losing side.
There are situations where a short position makes sense. For example, like yesterday when the market was tanking (DJIA down 1,800) and you want to either hedge your existing portfolio short term (intraday or for a few days) or if you are just trading the momentum of price (no portfolio).
Another possibility is pairs trading in correlated issues. This market independent and you are trading the spread between the two securities, in the expectation that the spread between the two will contract. For a simple example, two gold stocks where XYZ has run up quickly and you expect (hope for?) a reversal. Buy ABC and short XYZ. The pair profits if ABC catches up or XYZ retraces, narrowing the spread. If achieved, close both positions. A variation of this is if the pair makes a decent move up or down, replace the winning leg with an appropriate over (or under) valued third gold stock that hasn't made a similar move. This type of trading is more viable in volatile periods (like March) when the components percolate much more intraday and daily.