[SHARE] Simple RSI BULL/BEAR strategy

[richard]

Here's an article with some ideas on optimization:

http://mechanicalforex.com/2010/10/optim...ation.html

Notice he suggests using lots of data ... preferably 9 to 11 years worth!  So we're at a bit of a disadvantage.

Some sort of stop loss seems necessary.  Losses should never get quite that large!  Even if it damages your magical (absurd?) back testing score.
Remember, back testing is not real.  It's just a guess.  The real world will throw you curves every time.  New price action never seen before, etc.

So testing and using an accumulation of trades as your primary score is a very bad idea.  That will lead to curve fitting your test data.

Number of wins vs. number of losses.  Average win vs. average loss.  Largest loss.  Greatest number of losing trades in a row.  These are important.

You're building a weapon to use against an enemy you've only heard about and never seen before.  Keep that in mind.  Optimizing to fractional perfection is wishful thinking.

Thanks.