DislikedI don't think taking all the signals is good. If I ride a trend up to where it starts to retrace and exit, I don't enter right away on the next peak or valley. Many times after a strong trend there is a period of consolidation. I try and make sure the market is not consolidating before I get back in, or at least in a large enough range to be able to range trade successfully. I then enter on a breakout of the range to catch the next trend.Ignored
One thing I noticed in that ranging period is once it was clear the up trend was over, there was an extreme high and low around the start of Sept and the next 2 months couldn't get out of either end so waiting for the break would be one approach.
The other thing I have been working on after staring at daily charts all day, we all know that in a trend, there are mostly new highs and lows in the direction of the trend, with the occasional travel beyond the stair step ideal situation (triggering a stop I guess) but then the trend may resume, so the way that may make sense to detect a possible consolidation vs a temporary stall in the trend that will resume - This needs more work because I only watched a few months but....
Say the trend was up, then suddenly you see lower highs/lows. If there are two of each formed successively, lower high and lower low, then put things on hold and wait. Don't consider it a reversal yet - consider it a range/consolidation pending investigation.
I haven't worked it out completely but one way to resume the up trend, or to fully accept a reversal and start taking a down trend, wait for another set of two successive higher highs and lows, or lower highs and lows. Then take the next high or low break out in the trend direction. So in the case of the up trend that suddenly ended up seeing two lower highs and lower lows in a row, if the next move went lower than that second low, I'd consider the trend now short and take a short trade on that break. Or if things just got messy in true consolidation fashion and there were highs and lows all over the place sideways, then nothing is traded until 2 consecutive in the same direction form and are broken.
So this might keep you out of ranging periods and during a trend you will occasionally get stopped out on those few out of bound peaks or valleys but you get back in when they are broken again as usual so you just miss out on a few pips without getting tossed back and forth (don't stop and reverse, just stop and wait to see if it re-breaks or forms consolidation).
I told you there would be a 20 page essay coming soon. Remember my fib days in late November, I didn't sleep until I was satisfied. I'm not feeling very tired tonight....
Edit: I just progressed onward in the chart and around July and August 07 there were those 2 big sudden down moves, very large moves with no warning and I wonder how I could have captured that if I didn't ease into a trend reversal, and didn't stop and reverse during the up trend. So I'm wondering about having a shortcut trend reversal if something drastic happens like a single counter trend move takes out two peaks or valleys in a row - use the first violation as a stop in the trend as usual and if the next one in line is taken out, take the counter trade. That was just a rough thought, I know it would trigger a lot of reversals falsely but maybe the stop outs would be so small on the false trades, it wouldn't matter. Or better yet, instead of a fixed number of highs/lows, find a recent significant looking support or resistance area, a sudden spike outside all the recent action, and if that is taken out against the trend, reverse. Or a reasonable trend line following the obvious trend in play, if that's taken out...Doesn't really matter how it's done. One good way might be to look at the chart and visually find the price at which the trend would be arc'd in the other direction and take that level as an emergency trend reversal to catch an unexpected move. That seems to look good in the mid-late 2007 E/J examples.
I don't mean to go curve fitting a fixed data set, just making notes in what I see, then when I'm done I'll review the observations and see if something can be summarized, then go test it on a long sample set of the pair, if it looks good, consider using it, and testing it on other pairs. If it only seems to work custom to that one pair it was developed on, nothing wrong with that. I don't mind coming up with a specific set of rules for each pair if I need to, as long as I make money.
It's art. And technical. Perfect balance of a career.
"Support and Resistance either holds or it doesn't." -Anonymous hillbilly