AChartist's Blog

Welcome Guest ( Log In | Register )

Preperation for next week

Posted by AChartist, Jun 15 2008, 09:13 AM

Buy QID Monday open ( hoping for some opening pop in the indexes )
Sell QID Tues Open.

Here are some stocks I've located for Tues long.
If there is a general sell down, many of these will be priced right.
DYOD, I am so confident in the surge to July 2 I am getting the
cheaper OTM calls on some of these. I have my orders loaded
and will execute if it happens that way with a shock down early next week,
monday ideally, possibly tues.

Some of these configurations to look for, are 60 min turned up but
overbought, that is the indicators overran the price. Look for 60 min
macd kissbacks if the selloff comes, that sets the daily chart.

These will be a variety of conditions, with some in continuation, some
holding up the relative strength, some needing
$2-3 drops to 40 day lower BB. Check each one out, I'll only take
some if they get the lower 40 day BB tag on Tues.

AMD, AMD, AMD the top pick, the 40 day BB's are up
FLEX, .50-.60 risk from here, consolidating in top half of bands
BAX, rev head and shoulder in continuation with relative strength
WFR, 60 min buy take a pullback
EMC, on lower 40 BB and chart gap ~ $15
VZ, $1.50 risk from here, take it Tues
NVDA, they might crush it to $19 on Tues,
AMZN, if they hit $73 on tues
MSFT, consolidated on lows
AMAT, consolidated
STLD, continuation
CSCO, continuation
HOLX, staying with this one with 1/2 position, to add tues no matter what
HOG, a little 5th wave to lower 40 day BB will be perfect
WFMI, bullish divergence, minor 5th wave low needed.
CIEN, should be interesting in a general swoon


Urgent stock selection

Posted by AChartist, May 28 2008, 10:22 PM

05/28/08

From the description below, filtering for PMO value > 3, on a standing PMO buy signal, with a
-4.2% 15 day price decline, located a favorable chart, AMD. Looks urgent in the 60 min, it
has pulled back to moving averages from the top daily band, where a tag of the lower 60 min
band with rising macd signal is immediate. Bullish div in all 60 min setups.

Also appears energy will get a bounce, not necessarily immediately.
My interest there is to find the next entry in DUG, lost $40 on the last attempt.


Stock Selection, statistical use of DP data

Posted by AChartist, May 25 2008, 10:47 AM

The available data in Decision Points SPX Plus database of 564 stocks
was analyzed for statistical significance. These are the terms that are proven
significant P value < .05.

15 Day Price Change. A negative 15 day price change is favorable for the next 15 day price gain.
Excessive + 15 P gain > 10% is clearly dangerous. Price gains < 10% are generally negative or mixed
for the next 15 days.

The keypoint here, is a negative 15 price decline is favorable for the next 15 days. 45% R-squared.



The next components are PMO Rank, PMO Value, Trend Value, and PMO signal Value.
Trend Value is +1 if PMO is rising, -1 if PMO is falling. PMO signal value of +/-1 is DP PMO Buy/Sell Signal.

From my casual observation here is how rank works. When viewing the DP stock chart
pages, items are sorted from highest Rank 1, to lowest Rank 564. My observation is that highest Rank 1-20
eventually have large corrections. These are top momentum items that can have large short term gains, but also are
the next items to drop off. Most former #1 rank go all the way down into 400's.
I like items in the top 50-200 that may be rising as opposed to 1-20 that may be topping.

Items 100-250 are mixed with stocks rising out of >200 and former leaders dropping down.
These are good long candidates if rising items can be distinquised from the falling items.

Items 240-400 may be most unreliable. Items 400+ are most likely to decline.

Locate rising items from 50-250 Rank is my preferred strategy.





Clearly you want higher PMO Rank, Higher PMO Value, +1 PMO Buy Signal, and +1 Rising PMO Trend.

Selecting items that meet these criteria, with a negative 15 day price at the same time,
are the cinches.

I have done it numerically with multiple regression, it can also be done by sorting in a spreadheet.

Numerically, this regression formula locates the higher Rank Items with a negative 15 day price,
but rising 1 and 5 day price. ( PMO rank and PMO value are collinear so the formula has rejected one
of them )

Signal Rank = - 19.2 P 15 Day + 14.0 P 5 Day - 0.0158 PMO sig V
+ 0.00172 PMO 15 Day - 0.000872 Rank - 0.00430 PMO 5 Day
+ 10.9 P %Change

The terms of the formula like a negative 15 day price, a positive 5 day price change, a negative
PMO signal value ( is on ST sell ), a positive 15 day PMO change, a lower PMO rank, a negative PMO 5 day change,
and a positive 1 day Price change.

You can feel what this formula is measuring, a 15 day price decline in a higher relative strength item, with
positive 15 day PMO rank change, positive 1 and 5 day price change. It makes sense, buy 15 day pullbacks.

( small PMO rank 1 = highest relative strength )
( PMO signal value is +/-1 Buy/Sell signal level times number of days )

The negative 15 day price carries the highest r-sq.

This formula will produce a numerical +/- signal strength.

This week a candidate located is THC, and AXP, possibly WFMI from the washout category.

Financials are well represented but most in the 300+ PMO rank category.
Last week was HOLX from the washout category.

This week is concentrated in lowest relative strength items, 300+ RANK so I expect
this week is a rotation into lower price sectors. There are not many positive selections
this week in the 50-200 PMO Rank category.

Higher PMO ranked items, <200 Rank, with only moderate positive price return projected are,
BIG, JNY, JNPR, BIDU, GOOG, TEL

I expect it will change this week when some positive 1 and 5 day price changes are booked.
For now the selections are in financial, probably BIG JNY JNPR BIDU GOOG TEL
and more will surface in a day or so.

I'll be working on it more and will be revising it, it will not change much. I think I want
to use PMO value instead of PMO rank, is the next step ( cant use both together because
of collinearity ).

Stay tuned, I expect we'll have more buy candidates when some positive 1 and 5 day price change
is registered.

For Monday I like JNPR, WFMI, THC, AXP









Decision Point Fidelity Select Funds, Sector Tool and Signals

Posted by AChartist, Apr 25 2008, 07:57 PM

This is huge, combining snapshot databases from 12/3/7, 1/31/8, 3/17/8, and 4/24/8
captures a variety of market conditions.

First the Conclusion or Trading Rules, then the explainations
1) Dont trade against a long term signal ( will explain frontrunning exceptions later )
2) Short term buy signals are reliable from 0 to 32 days
3) Short term buy signals within long term sell signals are reasonably reliable
14%, 8 of 57 signals fail for and average 3% loss.
4) Short term buy signals within long term buy signals are powerfully reliable
5) Short term sell signals are reliable from 2 to 20 days.
6) Short term sell signals within long term buy signals are unreliable
7) Short term sell signals with long term sell signals are powerfully reliable

8) Why mess with the rest, only take short term buy within long term buy
and short term sell within long term sell signals. New long term buys should
be taken immediately with max 1% drawdown likely.

9) But, do not sell a new long term sell immediately, wait up to 14 days for the
oversold snapback to retrace up to 5%. Or short sell the sector at about 14 days.

10) Special case of frontrunning. See item 3 above, taking short term buy within long term
sell signals. This signal will fail < 8 days or last up to 30 days and switch to long term
buy ( where the tech funds are sitting right now on 4/24/8). This is the highest reward
trade with average 3% risk. The reliability of this trade is enhanced by chart interpretation.


How did I do this year? Not following this mechanically but in knowing it, my Fidelity 401k is up
9.5% vs SP500 is down 1.2%, Nasdaq is down 2.8%. And I made an embarrasing amount
of mistakes ! I sold funds the day before they made 3% moves, I took drawdown frontrunning,
What an idiot, I should be up 15% by now!

The top table is the Anova of the short term PMO signals and Signal Days Elapsed
vs %profit. The short term signal days duration is 50% correlated to % profit.
The signal condition buy or sell has no effect at P=.976, the signals buy or sell are equal.
The signal duration, days elapsed, is 100% predictive of % profit at P=.000

Conclusion : trading short term signal without knowledge of the long term signal is 50% chance. No good.

The second table is the Anova of the long term PMM signals and Signal Days Elapsed
vs % profit. The long term signal days duration is 86.1% correlated to % profit.
The signal condition buy or sell has no effect at P=.941, the signals buy or sell are equally reliable.

In summary # days elapsed of the Long Term Signal PMM explains 86.1% of % profit and
buy or sell signal are equally profitable.

The # days elapsed of the Short Term PMO signal explains 50% of the % profit and buy or sell
signal are equally profitable.

Obviously we cant short a fidelity select fund but it is useful for sector selection when short selling.





How to maximize % profit with this knowledge?

At first brush, with long term model explaining 86% of profit, never trade against a long term signal.
On second pass, within the direction of a long term signal, use of the short term signal may
increase returns 50% !

Lets's drill another layer deeper.

This is the test for equal variances with conclusions noted. Days duration is very important.
Variances occur around cycle dates with 20 and 30 nominal days important for short term
signals, and 20,24,90 days important on long term buys, 13, 28, and 64 days important on long
term sells. These are the days when signals can become unprofitable ( changing signal ).



So we've isolated we want short term buy signals on long term buy signals to be long in
the 401k. Already knew that?

Ok, it is that easy, take long term PMM buy with a PMO buy immediately,
none unsuccessful even in bear mkt conditions. They may be weak for 20 days
of the new long term buy. Wait 20 days before giving up on a new long term buy.




But, Long term sell signals tell a different story, not so easy.
A long term sell signal is wrong until about 14 days. So if caught long in a new long term
sell, sell near 14 days to recover up to 5% losses.




Special Cases on Front Running. We have a lot of data on short term PMO buy signals within
long term PMM sell signals. I did a lot of frontrunning in the lows.

Only front run by buying a short term buy signal while under long term sell signal.

Short term sell signals in long term signals are statistically dangerous.

This is fascinating stuff, the short term buy signals last either <8 or ~30 days.
What happens to them? The ones that die <8 days switch back to short term sell,
snapback rallies. The ones at ~30 days may switch to long term buys and drop off
this long term sell database. This is presently the case with some technology funds just switching to
long term buy. And as we saw above, long term buys are good immediately with only
small 1% risk and relatively weak for 20 days.

Tech is here to Stay after this top then a 20 day consolidation period.



So this case of front running short term buy signals with a long term sell signal is what
we like to do, pick the bottom, this has highest reward potential, it's sexy.

This can be enhanced by chart interpretation.

The one we all know, bullish divergence. It is so shrill on the boards and hard to do, why?




The harder one, so easy in the rear view mirror.








Decision Point Signals

Posted by AChartist, Dec 4 2007, 09:46 PM

Hi, working on this analysis, here's what I have, still a work in progress.
I am planning to start trading this system next week, no promises.

Considering all PPM signals, the signal direction buy or sell is insignificant, P=.633.
A PPM buy or sell is equally reliable. The days duration of the signal is significant
explaining 77% of the model.

[Image Not Displayed]

In short term PMO signals, the signal direction buy or sell is not insignificant,
P value = .12 is significant for me in trading and hand grenades.
We'll see below to only take short term PMO signals in the direction of the standing long term
PPM signal.

[Image Not Displayed]



The conclusion below is, don't take a short term PMO sell signal while under a long term PPM
buy signal:

[Image Not Displayed]


The PPM long term sells, have a sweat spot to open new shorts within 20-25 days
of the new signal. This appears to be that the new long term sell signal required
significant weakness such that the item recovers during 20 days of a new long term sell signal.
The item price should be up 20-25 days after a new PPM sell signal, for best short sale entry.

[Image Not Displayed]


A more complete view of the PPM long term sells, explains do not take PMO short term
buy signals during a PPM sell signal. PMO short term sell signals are very useful while
under a PPM sell signal :

[Image Not Displayed]

It appears to take PPM long term sell signals around 20-25 days after the new signal,
and to take PPM long term buys signals anytime, but less than 50 days of the new PPM buy
should maximize profit and timeliness. A short term sell during a standing long term sell
is very useful.

This data only represents the last 2 years under bull mkt conditions.