[zdo]

Originally Posted by Kiwi »

More seriously, for the OP, the poster who sarcastically suggested that this is applying elementary school ideas to the market is somewhat correct. Most try an acceleration measure (second derivative of distance wrt time or price wrt time in this case) at some point and discard it. Extremes (absolute or relative) of velocity and position seem to be more effective in the end and good trading is rarely complicated.

Yes - re second derivative
re velocity I have basically same issues with velocity as I have with stock acceleration measures. Central tendencies may move without flitting about, but price doesn't...
So a twist on the question - does anyone have algo's for 'vibratory rate' that are applicable to raw price?

[BlowFish]

You could look at Ehlers work. A lot of his stuff is about splitting the cyclical component from the non stationary. He seems to use the same technique for that in most (all?) of his work. The Hillbert transform. You should be able to find code in TS to do this.

[BlowFish]

Worth re-iterating the problem is not that it is an elementary school technique but that velocity will swing wildly (as much as the extents of its limits) tick to tick.

[zdo]

Thanks, Blowfish.
I do use both Ehler's modified and Hilbert - but only as indications to lengthen or shorten cycle duration projections (a la JMHurst type variations on nominality, etc ).
But, so far for the 'acceleration of vibratory rate' questions of this thread, neither of those is sufficiently granular or 'instantaneious' at any timeframe .

[zdo]

Originally Posted by BlowFish »

Worth re-iterating the problem is not that it is an elementary school technique but that velocity will swing wildly (as much as the extents of its limits) tick to tick.

If you have time, could you please elaborate a bit. Thanks.

[BlowFish]

Long time since I read Ehlers (first) book but I seem to recall he puts arbitrary smoothing into the mix in a couple of place. This may well be why.

Volatility ,even at the tick level, (some might say especially at the tick level) changes very rapidly. I think it will always be a trade off between pre and post processing to smooth this a little with loosing what you want to see. (though I am not completely sure exactly what you want to see or perhaps more importantly why you want to see it).

This kind of leads to the 'velocity will change rapidly' statement I made if you get several seconds between a couple of ticks the velocity will be pretty low in it's potential range. Actually if they are at the same price I think the velocity will be zero, right? Then you get a tick very close maybe a few milliseconds ticking up. You are dividing by a very small number ,ms, (twice). The velocity will shoot up to the top of its range in fact you will probably want to handle a divide by zero there (2 ticks the same time stamp) which would give infinite velocity!!

A good title for a paper Zero to Infinite Velocity in no Time Flat!