One approach I've had for a long time, that applies to this, support and resistance and other things in trading, is that you shouldn't do things arbitrarily. You could use stochastics with settings 8-3-3 or whatever. Unless you can say why 8-3-3 is better and is chosen rather than some other settings then what's the point?
Same applies to me for support and resistance. I don't draw a rough area based on a whim. I realise that it will in reality be a rough area. But I have rules for where that area should be, and specific lines, and the rules are clear, so that I am not in doubt of where my methodology suggests the S+R should be. There are a lot of skilful traders out there, and I don't believe they're just doing arbitrary things. Eventually a trader may be able to trade based solely on instinct, which may even be better, but I haven't reached that state, so can't comment. In the mean time, there is a reason for pivot points. Traders are aware of them,and they're not arbitrary. Everyone can see the same number. Big round numbers, everyone can see. Yesterday's high and low are important, because they're not arbitrary. Trendlines can be arbitrary, unless you define specific rules. Is your trendline seen by other market participants? Maybe if you have rules. If it's just a trendline based on gut, then probably not.
So with Fibs, it's easy to curve fit and see that after the fact which high and low worked. But I'd avoid that. If you want to check whether they work define a rule for where to draw them. A non-arbitrary rule. Example: if the weekly bar was up, then draw a fib from the high and low of the weekly bar, and see how that works out in the coming week. If you have 4 weekly up bars draw the high and low from the low of the first bar to the high of the last up bar and see how that works out. Take the all time high and the all time low as your fib-range, or the current day's high and low. Whether any of these work is for you to find out, but I emphasise again, it should be clear cut where the high and low is for you. If it's not, then forget it (imo).
