I can easily prevent the crash by checking the divisor and to set the result to 0 when the divisor  = 0. However, this will make the formula useless. FSB will not be possible to distinguish the value of strategies with 0 trades and  with 1 profitable trade.

Ok. than I can make it 0 when there is no deals and 100 when the divisor is 0 and when there is at least one deal. It will solve the problem with the zero deals but will not be possible to distinguish strategies with 1, 2 or 100 winning deals.

If you want you can play with that. You can easily implement the formula and to show the results in the strategy output by modifying the "AccountStatistics.cs" file in the "C:\Program Files\Forex Strategy Builder Pro\User Files\Code" folder.

When you do it, switch on the Custom Account Statistics option in the Control panel.

sorry for not realizing that you have answered.

This is just an idea for your consideration to help to find better measure of risk, since 
although mathematically it is a risk to deviate from the expected values in both directions, we can live with it as long as it is happening on the winner side.
Consequently, we consider risk of losing only.

"The formula is incomplete and will crash if we have one or several winning deals producing higher return that the designated target return ("Division by zero")" -

I really doubt. A sum of squares of differences can hardly add up to be zero, since you are adding up positive numbers only. The only case that DIV/0 may happen that the data are running at a constant level and we were smart enough to foresee  this level, or we are always winning.   
Out of these cases it describes the perceived risk better than Sharpe ratio.

What I see is only a theoretical study without pointing the advantage of this method.

The formula is incomplete and will crash if we have one or several winning deals producing higher return that the designated target return ("Division by zero").