2/26/2007

Episodic Pivots and "Idea Pickle"

Studying episodic pivots on daily basis can be profitable strategy. It is not a purely mechanical strategy. The starting point of the idea was my study to find what leads a stock to move rapidly in short time period. Or what triggers a 100% plus move in a stock in a year. Or 50% move in a month or 25% plus in a quarter.

Majority of the stock in the list had one day 10% or plus move within first 20 to 25% of their 100% or more growth in 260 days. So I framed the question as "Can a one day move signal something about future returns". Initially I tried to look at it as purely mechanical thing, but it was not very useful.There were too many stocks. So I abandoned it.

Most of the time when I abandon a concept, I put it in "Idea Pickle" jar. This is a long term practice I have used for several years, where I put half baked or promising ideas in a jar to pickle. Revisiting them after some days or months sometimes gives you a completely new idea. I have several of such "Idea Pickle" jars to ensure I never run out of ideas. Not all are related to trading , in fact most are related to new business ideas or other stuff.

Some months after I stopped working on the idea, I read a book:Swing Trading:Power Strategies to Cut Risk and Boost Profits by Jon D.Markman. There is not much useful information in that book except for one system. It describes a trading system used by George Fontanills. It is a system based on price and volume.

Now if you know Fontanills, he sells a 3000 dollar option course which basically tells you if you want to master these strategies then you need to buy another $5000 platinum course and so on. So when I read it at first I was very skeptical of it. But then I remmbered my "idea Pickle". So I investigated it further.

The system in short as per the book:

" Basically it comes down to finding stocks with greatest short term momentum that will turn in to long term momentum."
The Scans:
1 up or down 30% in a single day on at least 300000 shares traded or
2 up or down 20% in a single day on at least 1 million shares traded or
3 up or down 5$ in a single day on volume of at least one million shares and
4 still trading above $7 and
5 has options
6 has a long term catalyst like earnings or accounting fraud

He buys options on it after few days , typically such stocks after a pause tend to move up or down.

Now I played around it a bit but could not find good results. So I put the idea aside again in the "idea Pickle" jar. Last year I revisited this concept when I saw Mark Minervini site where he put up a list of his past trades. After close xamination I found that most of them were triggered by news pivots and the entries were post such high volume spikes. So that set off a bell in my head.

I revisited that Fontanills concept and my original price volume spike idea and converted it in to hybrid strategy of using a discretionary filter to eliminate some Episodic Moves.

The basic idea is to study large one day move to find stock likely to go up.

I changed the Fontanills scan a bit and added some more scans to it. So here are the scans in TC2000 for it
Bullish
1 ( 100 * (C - C1) / C1) >= 20 AND V > 10000 AND C >= 5
2 ( 100 * (C - C1) / C1) >= 30 AND V > 3000 AND C >= 5
3 (C - C1) >= 5 AND V > 10000 AND C >= 5
4 ( 100 * (C - C1) / C1) >= 10 AND V > 1000 AND C >= 5
5 C > C1 AND V > 5 * AVGV50.1 AND V > 3000 AND C > 5
These five seperate scans give the bulish pivot list
Bearish
1 ( 100 * (C - C1) / C1) <= ( - 20) AND V > 10000 AND C >= 5
2 ( 100 * (C - C1) / C1) <= ( - 30) AND V > 3000 AND C >= 5
3 (C - C1) <= ( - 5) AND V > 10000 AND C >= 5
4 ( 100 * (C - C1) / C1) <= ( - 10) AND V > 1000 AND C >= 5
These give bearish pivot candidates.

Together the scans give you less than 20 to 25 candidates to work with daily. After that I analyse each for why it is up or down to isolate the catalyst. Most of the time it is earnings, biotech news break, insider buying, major sector breakout or some news.

I enter the 1-2 with clear catalyst and where the move is just starting. Most of them I research during the day itself and enter,some next day.Many I enter later. I have also created same scans for stocks priced less than 5.

After a trial and error and building a database of such episodic pivots over some time now I have got a list of high probability events and low probability events. Earnings, new product, analyst earnings change, increased earnings guidance, sector moves etc have high probability to trigger multi month rallies.Like this I have around 12 Episodic Triggers which have high probability of starting off a 100% plus move. James Crammer, Barron's, WSJ, and other publication mentions, analyst upgrade etc. have a low probability of follow through.

After such experimentation the hybrid concept as of now has climbed to the best performing strategies amongst my pool of 20 odd strategies, beating even my earning lead breakouts strategy(which was the best strategy in my pool of strategies so far).Obviously most of the earnings lead breakouts also figure in the Episodic Pivot list.

Sometime revisiting an abandoned concept works and sometime a semi mechanical approach is better than purely mechanical approach.The Episodic Pivots approach gives you really really top quality ideas to focus on. But it is not purely mechanical.

Other Related Posts:
Episodic Pivots on stocks up 50% plus in a month- Part1
Episodic Pivots on stocks up 50% plus in a month- Part2
Episodic Pivots on stocks up 50% plus in a month- Part3
Up stocks and down stocks

20 comments:

Art said...

Paradeep, after making all 5 of PCF's of your bull scans (ie "These five seperate scans give the bullish pivot list"), then entering them into an EasyScan (after updating the pcf's) on AllStocks, I ended up with zero stocks. Any idea what went wrong?

Pradeep Bonde said...

They are seperate pcf's. Not to be combined. Each finds different kind of stocks. Some stocks will have overlap.

Art said...

OF course. Because I don't read code I couldn't see it. Thank you paradeep

Art said...

Paradeep, because you follow Thursday's IBD200 closely, do you run these scans on the IBD200 only, or on AllStocks ?

Pradeep Bonde said...

All stocks.

gosu said...

It looks like, since "Stocks up 50% or more in a month=19", the number was upwards of 20 last week, we are in the process of getting a correction.
If Nasdaq closes below 2500, it could test 2450.

gosu said...

Man! I wish I had half the brain you have.
I would be a millionaire by now.
Or maybe, I just need to use the right software.

Chanman said...

Do you use a time-based stop with this strategy? How long do you wait for a move before giving up?

Pradeep Bonde said...

Justin
Yes 5 days from entry. I re-enter later if the stock offers better entry later. I monitor stocks on this scan for 6 -1 year (especialy for those with 20 to 30% plus breakouts). Episodic Pivots often kickstart a multi year rally.

ybn said...

Pardeep,

I just got around to look at your program for Episodic move. Number 4 doe not make sense to me. Number 1 looks for stocks that are up at least 20% and a volume over 1,000,000. Number 2 is for over 30% and volume over 300,000. Number 4 looks for over 10% only on a volume exceeding only 100,000. That's look like a step backward or is that a typo?

I don't use TC2000 but the old Super Charts by Omega. It is easy to change the parameters and program it in SC. You can combine all five scans into one pass by putting each scan in brackets and using the "or" command between each scan.

Thanks for all the good information.

ybn said...

Pardeep,

Another comment on Episodic move. I noticed that scan number 3 can be misleading. It may generate a false positive results almost daily on stocks such as GOOG and CME.

Pradeep Bonde said...

YBN
The fourth scan I added later. It is not a typo. In fact I also have 5 th scan
C > C1 AND V > 5 * AVGV50.1 AND V > 3000 AND C > 5

The central idea behind the Episodic Pivot is a big price or volume move plus some episode. The Episode is basically some new information which leads to market participants acting in a way which leads to the price volume action. The news element is as critical as the price/volume action.

What I did is to analyse episodic pivots on individual stocks on their entire trading history and find what happens after that. In that data analysis I also varied the % price move. My analysis showed what % price move was more important in kicking off a enduring rally. So I played around with various % move from 2% onwards. Based on that analysis and ease of use I decided on 10% move. The fourth scan basically gives more candidates to work with. But even if one sticks to just first 3 scans, one finds extremely good results.

As to false positive on high priced stocks, yes that is a problem, but it is manageable. In the software which I use ( my own custom designed software), I vary that 5 dollar amount on high price stocks. Making it 10 dollar up to 200, 15 for 300 and 20 for 300 plus, to partly solve the problem.

Even though for illustration purpose I used TC2000 software in that post on Episodic Pivots (EP), in actual practice, my scans are on my custom designed software which also looks at things like days between EPs, number of EPs in stocks history, past price move past EPs, last 2 year price before EP,float, and so on. EPs is my continuing project and eventually 90% of the decision making on it will become automated, but at this stage the hybrid thing with all its limitations works wonderfully. The discretion adds significant value. It is not a mechanical strategy.

ybn said...

Pradeep,

Thanks for the explanation. In effect scan number 4 trumps scans 1 and 2 and makes them unnecessary. They are automatically included in number 4. I like this concept of Episodic move. I used to ignore stocks jumping over 20% in one day thinking they are already overpriced. Now, I'll pay more attention to them and analyze the reason behind the jump.

Thanks again

Pradeep Bonde said...

The reason I kept them seperate is because the first 3 scans I monitor for a year for subsequent entries. The 4 th and 5 th I don't.
In the overall scheme of things the 1 and 2 are more important signal than 4.

Alok said...

Pradeep, thank you for sharing these ideas with us. Love reading your blog.

I have a few basic questions about the EP method. In all these scans you've identify stocks after they cross the 5 dollar mark. What about stocks below $5, will this concept not work for stocks when they are in their infancy i.e. between 0-5.

In the 5 bullish scans, 1-4 look like they are independent of each other and can be run on their own. But the 5th scan has me scratching my head. Is the 5th scan supposed to be run at the end of the process. For example take the result set from 1 and pass it through the 5th scan to get a better result set.

Also can you tell us what your normal holding pattern is after you enter these stocks, or let me rephrase my question, can you tell us what might be a good exit strategy (3 mts, 6 mts, or do you set price targets for exits).

Sorry for the long questions.

Pradeep Bonde said...

1 For 0-5 the problem is noise to signal ratio is high. A 10% plus move can happen in low priced stock without any reason and may or may not trigger a long rally.
2 All scans are independent. Some stocks have high one day volume but no corresponding 10% plus price moves. The 5 th scan tries to capture that. Let me repeat it again, the scan by themselves only function to signal a possible start of big move. There are lot many variables which I use besides this to finally select a trade out of this scan.
3 My holding periods range from few weeks to months. There are no price targets for exit. Exits are based on trailing stops.

mauidiver15 said...

Pradeep, I have a couple of questions about how you enter these positions in the episodic pivot strategy.

First of all, when you enter immediately, do you just pick up the stock on one of your scans, investigate the event, and then enter right away no matter how high the stock jumped in value?

Second, when you wait for a better entry: what characteristics make you decide to wait to enter, and how do you know when it is the right time to enter.

Thank you very much for your help.

Pradeep Bonde said...

1 Buy based on the catalyst+ previous few years action to check for neglect+ float

2 Large float seldom start rallying immediately, tend to have a pullback and then go after few weeks. Entry on 4% plus breakout on high volume.

Unknown said...

Can you list your 12 Episodic Triggers referred to above? Do you rank or weight them by possible impact on the stock?

Jim Collins said...

Heres the TC2000 code translated:

Bearish
1 ( 100 * (C - C1) / C1) <= ( - 20) AND V > 10000 AND C >= 5

(100 * (Close - Close yesterday) /(Close Yesterday) ) <= (-20)
AND
Volume > 10000 AND Close >= 5