Over here Glen!.........
(This discussion was forwarded from: Subject 18538
post 231
Hi Glen, I'm not a big fan of indicators that flail back and forth between their limits because a person can read almost anything into these. I'm really not even a big fan of using them in conjunction with other indicators simply because standing on their own they give too many false signals. Sort of a "boy who cried wolf" indicator. Having said that, Lane's Stochastic oscillator consists of a fast (unsmoothed) stochastic, denoted by k, and a slow (smoothed) stochastic, denoted by %d.
The fast stochastic is defined as:
Close-lowest low over window divided by the max range over the window. In Metastock parlance, that is:
(( C - LLV( L,5 ) ) /( HHV( H,5 ) - LLV( L,5 ) )) * 100 .
5 is the "recommended" setting for the look-back window size.
Further, there is a %k slowing factor which is an n-period sum of the numerator and of the denominator before it does the division. 3 is the recommended setting.
%d is a smoothed version of this, and uses a 3 period simple MA.
The stochastic oscillator ranges from 0 to 1 and is multiplied by 100 to scale it from 0 to 100.
Basic properties:
--Can range from 0 to 100
--When it's zero, it means the closing price today was also the lowest price over the observation window.
--When it's 100, it means the closing price today was also the highest price over the observation window. (An observation window is the group of "n" consecutive day's data that is analyzed to give today's stochastic value)
--Anything in between shows the ratio of the current close to the high to low range of the observation window.
--The oscillator is pegged at zero when we're making consecutive new lows (relative to the observatio window), and the low=the close; vise-versa for highs (except it's pegged at 1).
I won't go into why I don't like this for now, but I won't criticize without giving an alternative. First, as a general rule, "noisy" indicators are generally more useful if they are changed into an accumulated indicator, and scaled appropriately.
A Substitute for Lane's Stochastic:
I like the following: Cum((C-L)/(H-L))/C. This is the accumulation of each day's price "balance" (i.e. where the close is within the days range), normalized to today's closing price. It's sort of a twisted version of a 1-day stochastic. Let's call it the Cumulative Price Balance Indicator, or CPBI for lack of a better name. Then plot a CPBI signal line: use the DEMA (double exponential moving average), 89 periods. (89 is in the Fibonacci sequence-always using Fibonacci set members for time periods in indicators is an interesting topic for another discussion). Compare that for entry conditions versus Lane's stochastic oscillator and I think you'll like it, but let me know. Now this only gives us the entry.
For exit, try: Cum(If(C-O>0, V*(H-L),-V*(H-L)). This is a volume accumulation indicator that adds all volume to the cumulative total on white candle days, and subtracts all volume from the running sum on black candle days (slight twist on Granville's On-Balance-Volume, but he used close to close). Let's call this the Cumulative Range-Polarized Volume (CRPV) (what the heck, it sounded good!) The signal line for this is: Apply a 21-period DEMA (3 Fibonacci ranks below the 89 used for the entry) to the CRPV plot.
When BOTH of these are simultaneously below their signal lines by some amount, Exit...stage left!
It's also useful to monitor the opening gap tendency, since the above indicators concentrate on each day's intraday battle. A cumulative gap monitor is created as: Cum(o-ref(c,-1))/Ref(c,-1). This is a running tally of the daily opening gaps, scaled by the previous close. Up gaps add, down gaps subtract. A string of consecutive gaps will produce a noticeable slope to the plot. On average it is relatively flat. Check these out and let me know what you think or if you have any questions. If you don't have a TA package, let me know and I'll post these indicators on a stock or index of your choice to my website just so you can get an idea of what they look like.
Hope this helped. As far as I know, I developed these from scratch, but they're pretty simple so someone, somewhere, has probably done them before. If you've seen any of these before, let me know.
dh |
