How do you pick a statistical edge?

[autofx]

That's pretty easy.

You pick some event that always precedes any worthwhile
move. Note: this is not the same as saying a big move
always follows the event. All black Corvettes are cars but not
all cars are black Corvettes get it?

There are many such events but one such an event would be a
MACD or moving average (SMA EMA WMA) crossover.

Now before you trounce this is no good for using as the basis
for a trading strategy consider this.

The fatal flaw in many crossover systems is that they always close
the open position reenter in the direction of the new
crossover. This looks simple tempting to many traders
because there are times when that works wonderfully.

The problem lies in the fact that in choppy action crossovers
happen frequently can act as buzzsaws that cause strings
of losses.

I have seen some good systems that use crossovers as entry
triggers but use other criteria for exits allowing for multiple
crossovers to occur while the trade is on -- in other words not
using the next crossover as an exit signal.

There are other events that always occur before big moves.

Pick one then spend most of your effort on position
management. Loss limiting tactics like halting trading after so
many stopouts in a day or a week or a month are important.
I don't think though that it's statistically advantageous to halt
trading when a certain profit goal is hit in a week. If you want
to quit after 100 pips in a week get that by Tuesday you
might regret quitting for the week because you might have had
300 pips by Friday. It's also true you could lose the 100 pips
you gained through Tuesday by Friday but in general it's smart
to limit losses but not put a halt to trading as long as you keep
winning.

Other's thoughts on choices of edges why they picked them?