What's the evidence for the fact that swing points cluster around fib levels. Define fib level and swing point.
That's a very fair point, and you could waste an awful lot of time trying to get your head around that stuff
I would argue that you have to take a pragmatic approach. For me, the key point is applying whatever method youve selected consistantly
To grossly simplify the argument does it matter if you measure the length of an object in feet, inches, millimeters, centimeters, or as a multiple of the length of my cats tail, so long as you define the method and stick to it consistantly ?
It's easy enough to design a filter to identify market turning points, I grant you the sensitivity of that filter will be a function of the parameters used in its construction. I could write quite extensively on this issue, but will refrain from doing so, but there are pragmatic approaches that could be realistically justified.
Once you have a bunch of swing points in some timeframe (which again is quite arbitrary, but could be justified on the basis that traders use particular standard timeframes) then by definition, you have levels associated with those points.
Of course you have the classic data mining bias issues, and I'm sure it would be easy enough to optimize a method that would identify levels that where subsequently respected (again defined by another arbitrary rule)
Obviously there are additional rules regarding the persistence of those levels etc.
I hasten to add I wouldnt trade this nonsense, and I appreciate just how easily we can be fooled by randomness, but I wouldn't dismiss fib levels completely out of hand without a lot more research. But as you point out, TA based research is fundamentally flawed if you want to take a pure evidence based perspective.
The thing is though, if I tossed a coin 1000 times and it came up heads, I'd probably bet heads on the next throw, despite my academic understanding of the situation. So on a practical level, if I'm trading to keep my hand in, rather than making a living, I will use fibs if there's a confluence with per existing support and resistance. I suppose its a bit like avoiding walking under ladders