What I can interpret from this data (If I am wrong correct me please), is that the trades that have an adverse execution higher of 2,25 pts are most likely to end in a loser trade, therefore the initial Stop Loss should be located at a distance of 2,25 or 2,5 points.
In addition is important to highlight that the 87% of the all profit was done in trades that had a MAE below of 2,25 pts, moreover the 70% of the losers trades have had a MAE above the 2 pts.
You're not wrong, but it is necessary to control the "trade" variable. IOW, your conclusions are based on the way in which the data is presented. The data is as presented based on a particular entry. If the entry is altered, that changes the data.
For example, if one is focusing on retracements, he must define a retracement.
1. What bar interval is being used?
2. From the point at which price turns back on itself (we'll use a long here to save time), i.e., fails to make a higher high, how far can it go before signalling a reversal?
3. Is there any significance to the number of bars in the retracement?
4. Is there any significance to the lengths of the bars?
5. When deciding where to enter the ret, does one measure from the bottom of the bar? the right tick? the top of the bar?
6. If, for example, one is using the top of the bar, at how many ticks above that bar does one enter?
7. Will the MAE be measured from the point at which price begins to turn back, or will it be measured from the entry point? If the latter, there are advantages to measuring it from the former, particularly with regard to trailing stops -- mental or hard -- and with reversal signals. This saves having to do all of this all over again for each task.
8. How will this change, if it does, according to whether the market is ranging or trending? For example, after all of the above is defined, was the MAE different during that six-week ranging period from the beginning of March this year to the middle of April than it was during the trending periods? Has it been different during the change in stride from June '13 than it was beforehand, when the pace was more leisurely? This may be of more interest to someone who's trying to code this than the discretionary trader as he will more likely understand the need for and be able to make slight adjustments from month to month and even week to week. This is part of the self-correcting and self-adjusting aspect of PA trading. Nonetheless, it should be considered if for no other reason than to avoid the puzzlement when one finds himself being regularly "stopped out" of trades that had been perfectly safe for weeks, or months.
This sounds like much more work than it really is. If one were a data geek, he could also look at how the MAEs vary according to whether the entry is made off an extreme, a DT or DB, a HL or LH, a median, a line break, etc. But if one has rid himself of fear, making these decisions based on price behavior rather than a complex decision-making matrix should not present a problem.