All donations go towards web site maintenance for all of Technical Watch,
keep it free of charge, and may be tax deductable as an investment expense.


PayPal Verified
Join our market chat sessions every Tuesday and Thursday at 4:00 pm Pacific time!
More information on subscriber services can be found at
http://www.technicalwatch.com/subs.htm




Support Technical Watch!
Buy from our Book Store!
Register Calendar Latest Topics
 
 
 


Reply
  Author   Comment  
mortiz

TW Patron++
Registered:
Posts: 1,054
Reply with quote  #1 
I know many of you peruse one of the busier message boards on the Web, and if you do, you may have seen a couple of posts about the price ramifications of small point changes in the NYSE ratio adjusted AD McClellan Oscillator (RAMCO).  The originator of the post(s) did a good job in defending his case that small point changes (4 points or less) in the NYSE RAMCO often result in significant price moves within two to four days following the small MCO point change.

Some of the responses in the threads were contentious, and some of the data presented attempting to detract the small point MCO change impact on prices over the following days, was suspect with respect to actuality.  Small point changes in the AD RAMCO often accompany the coiling or compression of price related indicators as well, so one would expect some sort of dramatic move following such compressions.

I looked at the NYSE AD RAMCO numbers from 1960 (arbitrary, but fits what could be called the "modern" era) to determine RAMCO small point changes' impact on changes in the SPX over the next two trading days.  In one of the threads on the other message board, Fib laid out the following criteria (doing this from memory, so weigh in Fib if I missed something):

1) NYSE AD RAMCO change of four or less points (up or down, doesn't matter)

2) Within two days, the probability favors a price move, one direction or the other, of 1% or more as measured by the S&P 500 (SPX)

3) Consecutive days of small point changes in the AD RAMCO often mute the effect on price, so if the AD RAMCO changed less than four points on two or more consecutive days, those small point changes were not included.

The results since 1960:
  • There were 1386 qualifying days, with 825 of those events resulting in a SPX move of 1% or more within two trading days.... many of the 1% moves came one day later, but a two day window was allowed.
  • The average percentage SPX change following all small point MCO changes on one day price advances was 0.72%
  • The average percentage SPX change following all small point MCO changes on one day price declines was -0.63%
  • The average percentage SPX change following all small point MCO changes on two day price advances was 0.64%
  • The average percentage SPX change following all small point MCO changes on two day price declines was -0.67%
Thus on average, the next trading day did result in a tradeable price move, with the trick being which direction to play the market..... maybe Fib will weigh in on his suggestion that the RAMCO posture itself can give a clue in what that direction might be.

If one were to measure the average random SPX price moves day over day in the past 47+ years, the days following a RAMCO small point change only nominally beat the performance of an average random SPX price move.  However, if one looks at the very dramatic SPX one-day price moves over history, the RAMCO point change the day prior to the multi-percent price change days is typically much higher than four points.

I didn't look at price changes over a four day period as stated by the contributor who initiated the RAMCO small point changes, so perhaps broadening the horizon in time would give different results than those generated by this quick and dirty exercise.

FWIW

Randy

0
doc

TW Patron++
Registered:
Posts: 497
Reply with quote  #2 
Randy, you probably won't believe me, but you are the first person I thought of that might have the database and initiative to do a longer term study to look at this and within hours, you have it posted! Kudos to you.

One important question whenever you get the chance: one of the replies on the other board suggested few instances of 1% moves after small point changes over the past 2 trading years. Based on your database, has the average effectiveness of this signal deteriorated from the 60% you found long term since 1960,  to a much smaller success rate over the past 2 years? If so, is it unique to the last 2 years or is it also much less effective over say the past 5 years as well? I wonder because sometimes, the more people know about an indicator, the less it seems to work sometimes...

Thanks, Doc
0
mortiz

TW Patron++
Registered:
Posts: 1,054
Reply with quote  #3 
Doc,

It would be interesting to look at the AD MCO small point change batting average in two year time windows over the past 80+ years, but don't have time for such an exercise right now.

Since the first of 2005, here are the stats I have for MCO small point changes resulting in 1% SPX changes over the two days following the small point changes:
  • There were 87 qualifying days, with of 33 those events resulting in a SPX move of 1% or more within two trading days (38% batting average)
  • The average percentage SPX change following all small point MCO changes on one day price advances was 0.59%
  • The average percentage SPX change following all small point MCO changes on one day price declines was -0.55%
  • The average percentage SPX change following all small point MCO changes on two day price advances was 0.43%
  • The average percentage SPX change following all small point MCO changes on two day price declines was -0.57%
So since 2005, the percentage of 1% moves following a small MCO point changes has decreased as well as the average SPX percentage price change following a small MCO point change.

From 2003 (through today):

  • There were 176 qualifying days, with 106 of those events resulting in a SPX move of 1% or more within two trading days (60% batting average)
  • The average percentage SPX change following all small point MCO changes on one day price advances was 0.87%
  • The average percentage SPX change following all small point MCO changes on one day price declines was -0.75%
  • The average percentage SPX change following all small point MCO changes on two day price advances was 0.64%
  • The average percentage SPX change following all small point MCO changes on two day price declines was -0.88%
So expanding the time window three additional years provides a whole new look, especially the average price changes.  To definitively conclude the MCO small point change indicator is no longer effective in forecasting nice price moves over the following two trading days (due to more folk being aware of the tendency), would require looking at all two to three year time windows over the past several decades. 

Has the price move forecasting capability of the MCO small point change indicator lost its polish over the past 30 or so months?  Before declaring the indicator a historic relic, additional statistical analysis would be needed to determine if other time windows over history have also produced reduced effectiveness of the indicator.

FWIW

Randy
0
doc

TW Patron++
Registered:
Posts: 497
Reply with quote  #4 
Randy,

The 2 year window was to confirm the other analysis. I believe that even a 40% chance of a big move in the next 1-2 days is useful info in terms of risk management and adjustment of any leverage if needed.

The 5 year window suggests that as I expected, any decline in the effectiveness is more likely due to variability than a true decline in its effectiveness as the tool has been out there for quite some time. If that is true, than we should see an increase in effectiveness going forward as we trend back to the average 60% success rate.

Only your mentioned detailed analysis might shed more light, but I don't think that is necessary. I'm happy just watching going forward.

Much appreciated.
Doc
0
chichi2

TW Patron
Registered:
Posts: 1
Reply with quote  #5 
mortiz, for the record, the orginal authors of the small chg MCO signals

matched the MCO small change to the Index
NYMO to NYA
NAMO to COMPQ

Note
SPX to its specific MCO for calculated for it.
Wilshire 5000 to its specific MCO for calculated for it.
OEX to its specific MCO for calculated for it.

NYMO and NAMO are readily available

The others are calculated on Decision Point thanks to Fib a long while ago.

When I was a heavy trader, I subscribed to Decision Point to give me access to additional data.  I no long do (age, and interests).

The importance is, A Specific Small chgs MCO apply to a given Index.

However, as we know, if SPX moves, NYSE tends to move, but maybe MDX does not move so much so NYSE is only a little mover.  thus using the wrong MCO may not give you the correct alert.

On the other hand, if you get some chg MCOs from SPX and NYSE, you are getting a more solid signal.

The basic mathematics of small chg MCO to forecast is beyond the scope of this post.  As is the basic mathematics of why the second Index in the same grouping with the first Index, increases its possibility of a sucessful forecast.

I love to watch all those doing the empirical trials.  They are fun, but the theorical answer is, the mathematics behind the action.  Provided the action is done correctly, same MCO as Index, will, in probability, work.

In general, i found the tendency to work within 2days, with chgs less than 4, particularily if 2 MCOs like indices are involved.  I always like one to be a grand one, like Wilshire or NYSE, and the other could any other.

I liked most, its alert idea, as i many times would get lost in some day trading and would forget the broad move coming, counter or inline.
0
mortiz

TW Patron++
Registered:
Posts: 1,054
Reply with quote  #6 
chichi,

Thanks for weighing in on the topic, and very good comments. I have breadth and volume data for many indices, but for the NYSE MCO study, I can easily plug in NYA for the price proxy. If complete rigor is the goal, then actually, NYSE common stock only AD numbers should be used with the NYA price proxy, since the NYA components are the commons.  The NYSE QCHA unweighted index would be best suited for NYMO if pure rigor is the requirement.  Since the commons comprise around 97% of the NYSE volume, UD volume MCOs would match up OK with NYA price.

Anyway, it was your post on TT that motivated my quick and dirty study, primarily to back your point, but aimed at the Technical Watch audience who are more interested in TA than the FA direction TT seems to be going.  I used Dave's criteria listed in one of the TT threads, thus the two day window.

Nice work, and good to see you posting on Technical Watch.

Randy

0
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.






Copyright 2000-2017 Technical Watch