Welcome Guest, please sign in to participate in a discussion. Search | Active Topics |

Positive and Negative divergences--benefits, drawbacks, and case studies Rate this Topic:
Previous Topic · Next Topic Watch this topic · Print this topic ·
rmr1976
Posted : Thursday, February 1, 2007 11:38:21 AM
Registered User
Joined: 12/19/2004
Posts: 457
I've been doing some preliminary research on divergences, which is a concept I find particularly useful. I hope to quantify divergences in the same way that Charles Bulkowski did some quantitative research on chart patterns.

What I've noticed is that divergence analysis can be useful even if the stock is in a strong trend, whether up or down.

It also appears to me that positive divergences are more accurate in calling bottoms than negative divergences are in identifying tops. I've made many (small) mistakes making bets on stocks declining based on negative divergences. This is consistent with technical lore that bottoms are more easily identified than tops, which can be long, drawn out affairs.

With positive divergences, you can wait for some small amount of confirmation--ie. a candle reversal, a test of a significant support level, etc.

With negative divergences, you have to get short before you are sure. Most of the time, if you don't get in when the stock looks like it is ready to go to the moon, the stock sells off fairly sharply to a support level (ie a moving average, or a test of a prior horizonatal level), and then buyers can and often do, push the price up a bit higher.

In strong uptrends, a negative divergence identifies a time when it would be wise to wait for a pullback before entering long positions.

For risk control with divergences, especially negative divergences, the best bet seems to me to be a time based stop. Divergences seem to pick up when price is going to revert back to the trend you are monitoring.

If the signal doesn't confirm itself in a fairly short amount of time (ie. 20 days--which is a VERY long time for a divergence on daily data), then the signal is likely a false one.

Positive divergences lend themselves to stops based on breaks of support. Bottoms are more easily identified that tops, so the signal that a trend is resuming to the downside is fairly clear. I've noticed that these breaks of support are often accompanied by spikes in volume.

For negative divergences, stops based on price are likely to get you whipsawed if they are set fairly close. I've seen numerous instances of a divergence signal being given, having the stock spike up to a slightly higher high, and then fall sharply in a short amount of time.

Case Studies

I use an easy scan for positive and negative divergences, and ADM came to my attention on 1/19/07.

It has suffered a pretty severe haircut since the peak in May-down almost 30% using closing prices.

I looked at the chart, and noticed that there had been multiple divergences in the past few months that lead to short term pops in the stock price. Would those divergences have been profitable? Some would have been--if you were careful about the buy price, making sure you bought near the LOWER Bollinger band.

I typically avoid divergence signals when the price is very close to the moving average (within 10 BB in Telechart), which I set to identify support and resistance.

I noted a clustering of divergence signals at the January lows, with volume spiking on updays, suggesting accumulation. The weekly and daily targets, based on support and resistance--suggested a potential move to 35.

I figured this stock might draw some buying interest if there is rotation to some defensive issues. I was a bit hesitant to take a position, since I knew that earnings were being announced today. But I finally bit the bullet and bought a small lot of June 30 calls yesterday at market open. I was going to be out on a close below 30.50, and the major downtrend since May likely priced in all of the bad news that could be released.

I had been burned before by waiting for news to be announced before taking a position based on this type of trading method. The signals are short term, and if it is a reliable signal, the low risk buy price won't last very long.

Today, earnings were announced, and the stock gapped up to my target. If only all of my trades could end this way. I put in a sell at market close for those calls. I might pick up more calls again should this trade back down to fill the gap.

Just so this doesn't look like I'm patting myself on the back, here are a few trades that didn't work out as well.

NVDA: There were multiple divergences leading up to the major price peak in December. I had bought some puts on NVDA in October, betting on a decline to at least the lower bollinger band (21 SMA, 20 width).

At each of those divergence signals, there was a pullback to support (the 21 day moving average), which were nice moves if you were short term swing trading, and had a very fast trigger finger to enter and exit.

I was looking for a longer term move down, and didn't get it. Any of those small moves did not do much for the options I had bought (slightly OTM). Puts I bought on 10/9/06 were closed out for a loss on 10/16/06--which happened to be the top of a short term swing.

This is a prime example of a major top taking a significant time to unfold. In October, it should have been clear to me that the uptrend was still dominant. Volume tended to coincide with large updays, if you look at a candle chart. Things started to change in November, where volume started to slack off at the highs, and price formed a congestion pattern. There was the spike up in December, but volume was no where near the levels reached at the November price peak.

The clear sell signal came on 1/3/07, with a large spike down on huge volume. You would have needed a trend following method to permit you to get short from that point.

Craig_S
Posted : Thursday, February 1, 2007 11:47:24 AM


Worden Trainer

Joined: 10/1/2004
Posts: 18,819
QUOTE
It also appears to me that positive divergences are more accurate in calling bottoms than negative divergences are in identifying tops.

You should consider the bullish bias in the market as a possible reason for this. It is often sighted as a reason why "long" signals appear more accurate than "short" signals.

- Craig
Here to Help!
nyquil
Posted : Thursday, February 1, 2007 12:06:36 PM
Registered User
Joined: 1/23/2007
Posts: 13
I don't know why but most of the time I look for negative divergence instead of a positive one. Most of the time at least I can catch a little pull back. I wouldn't bet it as catching a trend reversal thou because lots of times it goes higher and higher. But when that happens I would think it is good time to go short again.

Basically, I like to see the price make a new high, then have a fairly good pull back due to whatever reason, then runs significantly higher again, but due to the previous pull back the indicator will be lower, hence a divergence.

Most recent play was ONNN, it gap up 3 days ago with negative divergence, I catch a little pull back then get out. Others I can think of was NTGR, NVDA, STEC, and TRT. IBM as well, I thought it was a good short when it top 100's but I didn't get in that one thou.
rmr1976
Posted : Thursday, February 1, 2007 12:17:41 PM
Registered User
Joined: 12/19/2004
Posts: 457
Craig,

You make a good point, which I have considered. That only leads me to ask another question--why the bullish bias in stock prices?

The nature of stock prices is that long positions have the potential for unlimitted profits, while short positions are limitted to 100% declines. It is much easier for buyers to ID points for value, than it is for sellers to ID spots where the prices is "overvalued."

Price changes at higher levels become more volatile, when looked at from an arithmetic scale. A $1 change in a $100 stock is much more likely than a $1 change in a $10 stock.

Just looking at charts indicates to me that you need to be more selective when trying to catch a downswing with a negative divergence. You can't wait for candle reversals, because the reveral usually brings the price to a short term support level.

You also need to see how this short term signal fits into the big picture. Are you getting the signal when the stock has just broken out from a major downtrend or resistance level? If so, I'd think that the stock is more likely to pull back (or trade sideways) somewhat, then continue on up.

bruce51
Posted : Thursday, February 1, 2007 1:39:07 PM
Registered User
Joined: 2/3/2006
Posts: 113
hi whats your easy scan for neg and pos divergence
bruce51
Posted : Thursday, February 1, 2007 2:05:32 PM
Registered User
Joined: 2/3/2006
Posts: 113
and do you have a formula for Macd divergence
rmr1976
Posted : Thursday, February 1, 2007 3:57:12 PM
Registered User
Joined: 12/19/2004
Posts: 457
Bruce,

I'd prefer to keep my specific parameters to myself, but I'll reveal the general idea.

A knowledge of some concepts from cycle analysis would be useful to identify parameters that would work.

If you think about it, an Easyscan for divergences isn't all that hard to come up with. I don't know why it took me so long to figure out.

The method that the Worden trainers show involves using linear regression sorts, which on a slow computer can take a lot of time. The method is OK, but I spot a number of instances where it missed a divergence that I easily spotted.

Another problem: visual analysis of divergences compare only 2 points on price with 2 points on an indicator, that can be spaced apart an arbitrary (ie. random) number of days. There can also be multiple divergences on a chart, that somewhat complicates matters.

A linear regression compares all of the data in price with all of the data in an indicator, most of which isn't relevant to spotting a divergence.

What I do is the following:

Create an easy scan that searches for stocks that have

1. made a new high (low) in the past N days AND
2. have an indicator that is lower (higher) today than some period in the past.

I use 2 separate easy scans: one for positive, another for negative. The criteria for positive divergence are the exact opposite for negative ones.

The lookback shouldn't be longer than the period you used in your search for a high or low. If you use 100 days for the high or low, you shouldn't use more than 100 days (and probably much less) in the criteria for the indicator.

Using some random numbers:

1. Search for a 25 day high/(low) AND
2. RSI 14 Today less/(more) than RSI 14.25.

Plot whatever indicator you like underneath the chart, then plot your easyscan as a custom percent true indicator in the bottom window.

When your indicator signals a divergence, it will spike up. I use green for positive (buy) and red for negative (sell).

Using MACD in this easyscan is not too hard, and it is the indicator I use to search for divergences.
diceman
Posted : Thursday, February 1, 2007 6:40:30 PM
Registered User
Joined: 1/28/2005
Posts: 6,049
Quote:"That only leads me to ask another question--why the bullish bias in stock prices?"

The term :"GROWTH" would come to mind.

What possible use would a stock market be if prices
had fallen for the last 100 years. (and what company
would want to be listed on it?)

Quote:"Price changes at higher levels become more volatile, when looked at from an arithmetic scale. A $1 change in a $100 stock is much more likely than a $1 change in a $10 stock"

You are comparing apples and oranges.
A $1 move in a $100 stock is 1%.
A $1 move in a $10 stock is 10%.

(it will always be easier for something to move 1%
vs. 10% at any price)

When ever a price move is looked at it must be asked
in relation to what?

If I tell you an index "moved" 352 points. Unless you know
what its value is.you cant make a judgment on the size
of that move.

Thanks
diceman
rmr1976
Posted : Thursday, February 1, 2007 8:41:23 PM
Registered User
Joined: 12/19/2004
Posts: 457
Diceman,

You write:

Quote:"That only leads me to ask another question--why the bullish bias in stock prices?"

The term :"GROWTH" would come to mind.


You seem to take for granted that stock prices are essentially guaranteed to rise over the long term.

While it is true that U.S. stocks have risen over the past 100 years, what about stock markets in Europe over that span of time?

I don't care that stocks might be higher 50-100 years from now. I'm concerned with what stocks are likely to do in the next few quarters to years. The long term trend does nothing to help me figure that out.

Simply buying and holding is not the easy strategy its proponents make it seem.

You are comparing apples and oranges.
A $1 move in a $100 stock is 1%.
A $1 move in a $10 stock is 10%


That was essentially my point. It is percentage returns that matter, and at higher prices, it is harder to identify where a stock is "overvalued", while bottoms are relatively easier to identify, because the point values are larger, making stop placement more difficult.

There are a few charting methods that do not take this volatility relationship into account, however.
kokoda
Posted : Thursday, February 1, 2007 10:08:00 PM
Registered User
Joined: 1/12/2006
Posts: 296
Historically, the stock market has a 75% bullish bias. The reason is partly greed (not using this in a negative sense). Individuals that want to increase their assets have choices; one of these is a savings account that pays paltry returns. We have the stock market and also the realm of real estate. We have a financial system that allows various savings connected to the market - 401k, IRA's, 403B's, etc. This financial system component plus the greed factor accounts for much of the activity.

Apples/Oranges? .... Somehow the comparisons between high/low priced stocks within this forum wind up with the low priced stock being gifted with the 10% return and the high priced stock pegged with the 1% return. I just don't get it.

Hey, just my opinion
bruce51
Posted : Thursday, February 1, 2007 10:30:20 PM
Registered User
Joined: 2/3/2006
Posts: 113
Hi,rmr 1976 can you help me with setting up divergence with macd with green signal and red so i can use macd and can you show me how to go back 25days because my indicator just goes to new highs or lows and if you can help as much as you can it would be nice i dont know if you will get this message or a trainer will.well i hope some one can help me set this up with a little extra exsplenation thanks so much.
nyquil
Posted : Thursday, February 1, 2007 11:09:57 PM
Registered User
Joined: 1/23/2007
Posts: 13
QUOTE (rmr1976)

That was essentially my point. It is percentage returns that matter, and at higher prices, it is harder to identify where a stock is "overvalued", while bottoms are relatively easier to identify, because the point values are larger, making stop placement more difficult.

There are a few charting methods that do not take this volatility relationship into account, however.


QUOTE (rmr1976)

Positive divergences lend themselves to stops based on breaks of support. Bottoms are more easily identified that tops, so the signal that a trend is resuming to the downside is fairly clear. I've noticed that these breaks of support are often accompanied by spikes in volume.

For negative divergences, stops based on price are likely to get you whipsawed if they are set fairly close. I've seen numerous instances of a divergence signal being given, having the stock spike up to a slightly higher high, and then fall sharply in a short amount of time.



I don't get it. Why would a bottom be easier to identify than a top?

Let's look at AMD, OVTI, RHT, and XMSR. Where's the bottom for AMD? I have no idea. OVTI? I got in couple days ago when I saw a triple doji, but had to get out quickly 2 days later when I saw a long upper shadow reversal signal, there's no luxury of a divergence to tell me, ok we may be near a bottom, and even if there is a signal, should I really trust it? RHT gapped and sell down pretty badly due to news from competitor, and seems that sell off was an over reaction and it recover nicely the next day, but would I be that confident to catch a falling knife that way? I sure as hell was very tempted thou :) but anyway. XMSR did give a reliable divergence signal.

So the same question, is the bottom really easier to spot than the top? I find it more difficult.
diceman
Posted : Thursday, February 1, 2007 11:15:19 PM
Registered User
Joined: 1/28/2005
Posts: 6,049
Quote:"I don't care that stocks might be higher 50-100 years from now. I'm concerned with what stocks are likely to do in the next few quarters to years. The long term trend does nothing to help me figure that out."

Growth does not have a timeframe associated with it.

The same stocks I trade for a period of 4 to 10 days
are up for the same reasons. If you look at a chart
that has any growth over a period of time. It had to
happen in days.

There is no requirement for "buy and hold".
-----------------------------------------------------------------------------------------
Quote:"While it is true that U.S. stocks have risen over the past 100 years, what about stock markets in Europe over that span of time? "

There are many factors that impact the business climate.
(This is basically investors saying: "Europe we don't like
your policies)

Climate, resources, politics, tax-policy, socialism,
dictatorships.

There is no guarantee of economic success. It can only
be hoped that countries will adopt policies that promote it.

(there has to be some common sense associated with investment this is why large quantities of investment capitol
don't flow into Antarctica)

I guess my puzzlement came from that fact that you are
"surprised " by it. The equation is very simple survive and
grow or you die.

Thanks
diceman




bruce51
Posted : Thursday, February 1, 2007 11:17:50 PM
Registered User
Joined: 2/3/2006
Posts: 113
how do u set up the scan can u give detail word for word.
bruce51
Posted : Thursday, February 1, 2007 11:20:10 PM
Registered User
Joined: 2/3/2006
Posts: 113
rmr 1976 can you give me detail word for word to set up the scan on the pos and ng divergence scan so can set it up to try it.
rmr1976
Posted : Thursday, February 1, 2007 11:36:24 PM
Registered User
Joined: 12/19/2004
Posts: 457
Set up the PCF first:

Positive Divergence
1. Search for a 25 day low
2. MACD today > MACD 25 days ago

L < MINL25.1 AND
(AVGC5-AVGC25) > (avgC5.25 - avgC25.25)

Do the opposite for a negative divergence

H > MAXH25.1. AND
(avgc5-avgc25) < (avgc5.25 - avgc25.25)

Then calculate the PCF's. Finally, plot each one in the same window as Percent True custom indicators. Then plot with a 5.25.5 MACD.

Keep in mind, these aren't the exact criteria I use, but the principle is the same. Tinker with the formula.

Divergences are very sensitive to the parameters you set up. It is helpful to know some principles of cycle analysis, so you know what parameters are more likely to work.
bruce51
Posted : Thursday, February 1, 2007 11:45:35 PM
Registered User
Joined: 2/3/2006
Posts: 113
what indicator do you use for 25day low or do u use

a 52 week low indicator and when plot this in do i put it like thisL < MINL25.1 AND
(AVGC5-AVGC25) > (avgC5.25 - avgC25.25)

is this correct the hole thing
bruce51
Posted : Thursday, February 1, 2007 11:50:12 PM
Registered User
Joined: 2/3/2006
Posts: 113
how do i plot just go to macd and put it in
bruce51
Posted : Thursday, February 1, 2007 11:52:55 PM
Registered User
Joined: 2/3/2006
Posts: 113
search a 25day low how do i do that and thanks for all your support
bruce51
Posted : Thursday, February 1, 2007 11:55:06 PM
Registered User
Joined: 2/3/2006
Posts: 113
QUOTE (rmr1976)
Set up the PCF first:

Positive Divergence
1. Search for a 25 day low
2. MACD today &gt; MACD 25 days ago

L &lt; MINL25.1 AND
(AVGC5-AVGC25) &gt; (avgC5.25 - avgC25.25)

Do the opposite for a negative divergence

H &gt; MAXH25.1. AND
(avgc5-avgc25) &lt; (avgc5.25 - avgc25.25)

Then calculate the PCF's. Finally, plot each one in the same window as Percent True custom indicators. Then plot with a 5.25.5 MACD.

Keep in mind, these aren't the exact criteria I use, but the principle is the same. Tinker with the formula.

Divergences are very sensitive to the parameters you set up. It is helpful to know some principles of cycle analysis, so you know what parameters are more likely to work.
can you share some of your parameters im getting started i can use some help of exsperience thanks so much again
bruce51
Posted : Friday, February 2, 2007 12:07:19 AM
Registered User
Joined: 2/3/2006
Posts: 113
what are the criteria you use can you help me out thanks
bruce51
Posted : Friday, February 2, 2007 12:50:55 AM
Registered User
Joined: 2/3/2006
Posts: 113
rmr 1976 thanks for your help tomorrow and if you can help me with whatever criteria that you can pertaining to divergence any thing that help
bruce51
Posted : Friday, February 2, 2007 10:40:01 AM
Registered User
Joined: 2/3/2006
Posts: 113
hi for support and resistance what does the term 10 bb stand for
rmr1976
Posted : Friday, February 2, 2007 7:59:42 PM
Registered User
Joined: 12/19/2004
Posts: 457
Bruce,

Have you experimented with the parameters I gave you? I don't want to tell you what parameters are right or wrong. That is for you to decide.

If you know how to look for divergences visually, then you should understand the logic of the parameters above.

Another thing to remember--divergence trading is considered a counter-trend strategy, so the price at which you get a signal (and get filled) is important.

If you get a divergence signal, but price already popped from outside the Bollinger bands to the moving average, it may be wise to see how the stock acts as it pulls back to test support levels and resistance levels. It is possible that the move suggested by the divergence is already over, and the trend is about to resume.

This is not a complete trading strategy. At best, it is only a set up. It is up to you to figure out how to trade the divergence.

Do you want to take a short swing, using price targets?

Do you want to take failed divergence signals, and trade against them (ie. with the trend)?

Do you use other indicators (ie. volume) to confirm or dispute the divergence?

How do you handle multiple divergence signals? What determines your stop level? How much do you bet on a trade?

These are the things you need to think about.
Users browsing this topic
Guest-1

Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.