Patterns too big or too small to trade?

Hey guys!

I come here today asking for opinions on this one subject I've been stuck on.
Now I have been back testing Cypher patterns on the 15min time frame of AUDUSD and I found a lot of patterns however some were very big patterns and some were very small and other were "normal".
Now these big patterns were clearly visible on the 1h time frame and the small ones were clearly visible on the 5min time frame.

So I decided to put in a rule that says that If I can see it clearly on the 1h time frame it should be traded on the 1h time frame and if the 1st profit target is less than 15 pips I would not trade it on the 15 min time frame.

Now in my back testing I found 70 patterns in 2013-2014 41 of which were profitable which is 58% ish I think out of the top of my head.
Now before I removed the patterns according to the rules I had 131 patterns. I did not check the profitability on those.

Now I've started looking trough 2014-2015 data and I have gone a month without big enough patterns now.

Is limiting the size of the patterns you trade a good thing or a bad thing?
Am I crippling myself or is what I'm doing good?
Are patterns sometimes too big or too small to trade on the 15min time frame?


If any experienced pattern traders would kindly answer my question that would be nice!
Thanks
AntiVi

Depends of profit target vs stop loss size. If your entry system propagates SL sizes which are similar or even bigger than profit targets, than it's clear you will need significantly more than 50% success rate. So technically you will need to adjust your strategy according to that. (DUH!).
Overally, reducing the number of signals is good, so a trader can become proficient at recognizing those which have greater success rate, over time. Which unfortunately cannot be achieved via "backtesting".
If I understood correctly, if a pattern is "too big", it increases the stop loss size significantly, but it also increases profit target, therefore, trading those which have big stop loss size is simply adjusted with money management.
If a system is automated, adjusting money management seems impossible for me (probably because I'm not a fan of using automated trading)...Each pattern is unique from my perspective no matter how many times it repeats itself...
If you are proficient with backtesting, I suggest you test it on demo ofc, for as many patterns as possible in a given time frame + period. And then analyze the winners and losers and try to objectively approach each winner and loser as to why exactly did each win/lose - so you can potentially pick out those which you may find better signals than the others of the same kind. (Analyzing as to "why" did some trades win or lose is dangerous, I just want to point out that some patterns have higher reliability in certain market conditions, while those same patterns have totally different reliability in some other market codnitions.

For example, one of my most "likeable" daily candle patterns, Dark Cloud Cover (and it's downtrend counterpart Piercing Line) has around 73%+ reliability when they appear on top of trends (and combined with s/r lines, I trade those at every possible opportunity. BUT when those patterns appear in flat market, their reliability diminishes drastically, to the standard median of 50%. So when I see a double top on market for example, and a Dark Cloud Cover appears (has to be completed) on that double top, I'll increase my leverage since I find those very reliable (70%+ is huge). Over time I have completely expelled trading that pattern in a flat range. I need a prior uptrend in order for me to comprehend that as a reversal...

Sorry if I went out of topic for a moment,

Cheers.