Really high probability? (experiment)

I gonna try a little experiment, try to develope a technique that statistically shows that we have a high probability for the bar to close either higher or lower once the signal is given. My goal is a 80-90% probability, but the bar doesn't have to close higher than 1-2pips above the open.. for 5min bars, 1h or whatever.

The reason I want to do this, is that i'm interested in the statistical possibilities in the market.. to see how we can gain an edge with statistics and probabilities for reoccuring events in the market..

I'm not gonna trade this, at least not at the beginning 'cause the spread will eat my few winning pips.. more like to see if i really can develope a technique with theese statistics. There will have to be several "trades" each day..

So, the problem is how to do this.. my first idea is to compare the "closing distributions" with different values of the slope of an short avg.. and see if we can gain an edge. But i'm open to more suggestions, anyone else has any ideas?

If we all can help to build this together, i will share my statistics/research.. but i will need some extra ideas and "perspectives" on the result. Write in this tread or PM me if you think you can contribute..

/Johan

hi.
I used my excell to improve quiet nice method.
What i found is about gaining and crossing last high or low during next day [i use daily chart]

Since 2006 i found it quite constant [i use pairs AUDUSD and GBPUSD]

typical high for gbpusd is around 45 pips over last day high [if it gains last day high].
for audusd its around 22 pips
similar with lows.

My method is to open position, when last day high or low is gained, and establish TP on 20 pips for GBPUSD and 15 for AUDUSD.
method is supposed to give about 200+ pips a month for both pairs.

I suppose that it would work the best with trailing stop, but my broker doesnt provide it.
additionally i want to try same with 30 pips for gbpusd.

What do you look at exactly?

[BTW my end of a day is GMT-2]