FundVision Classic Collection - Dale's

Author: Dale

Salil,

When you talk about breadth indicators, you are considering three different data types and three different indicator types. I think almost all of the breadth measures I have seen are various combinations of these with different smoothings.

Data Streams:

• advancing/declining issues
  
• advancing/declining volume
  
• highs/lows

Recognizing that everyone starts with the same data, the question becomes what is the best way(s) to cut and slice it to find what you are looking for. It also points out that what you do for one data stream might work well for the others too. Another technical note is over what time period the advance/decline or high/low is being measured. For example in Barrons, you can see the number of issues that advance/decline over a daily basis or a weekly basis (there can be notable differences). Similarly, while highs/lows are generally calculated now over the last 52-weeks (one year), some people (Telecharts 2000 comes to mind) also have highs/lows over 4, 13, and 26 week periods which again measure subtly different things.

Indicator Types:

• differences: these are the most common with usually some type of smoothing over multiple periods or differences between two smoothings (McClellan Oscillator). One problem to be aware of is changing scales over time. There are a lot more issues and volume than there was ten or twenty years ago. Consequently if you scale it as a difference between the raw numbers, the scale should expand over time. An easy way to address this is to express it in percentage terms. For example Zweig created a measure he called breadth thrust which was a ten-day exponential average of advances divided by advances plus declines. This way, the scale remains the same 0 to 100%.
  
• ratios: these are indicators where they divide one stream by another (advances/declines). TRIN is an example here with the extra twist of it including issues and volume for advances and declines. A problem of which you should be aware is that the calculations for these usually introduce a indicator which is not symetric. For example, 5 advances / 4 declines = 1.25, but 4 advances / 5 declines = 0.8. Although the indicator appears to be balanced around 1, it is not symetric. Some people have addressed this. Tushar Chande created a balanced form of TRIN which I think he called market thrust. Arguably the percentage version might be better thought of as a balanced ratio rather than a difference. Both the differences and ratios are typically used as oscillators to highlight strength/weakness and turning points (overbought/oversold).
  
• cumulative: these indicators are where the calculations from either differences or ratios are simply added cumulatively over time. The advance/decline line is the most obvious example but the McClellan summation index is another. The summation indexes are typically used to spot divergences with some index.

I think you can categorize all breadth indicators using this little 3x3 matrix and would find not surprisingly a lot of similarities. Personally, I prefer putting the data into percentage terms and using differences/ratios. But these are personal preferences based on what works for me.

I hope this was what you were asking. If instead you wanted a laundry list, I can try to dig up some more but I think you can much of this on web sites like the one you found or ones associated with programs which list formulas or old issues of TASC.

dale

As an aside, I came across this web site which shows Metastock formulas for a number of indicators some of which I hadn't seen listed before. Might be of intrest.

• Indicator formulas

Author: Dale

The discussion on average daily change patterns around days of the week prompted me to take a closer look at the data. For this post I want to focus on the earlier historical record and leave aside the question of the last few years for another post. As already noted, historically Monday tends to be a bad day actually averaging a negative return while the other days of the week have been positive on average, with Friday looking the best. (see BobSenn's post link below). Sometimes averages can be helpful but sometimes they can mislead us with a picture that is not complete. I think the day of the week issue merits a closer look.

The data I'm looking at cover the period January 4, 1915 to February 20, 1990 for the Dow Jones Industrial Average (this was a quick data set I could get off the internet).

If you look at the first table some interesting things pop up. First, in the average daily change column we can see the general pattern Bob noted earlier holds with Monday being the only negative day on average. For this time period, Friday still looks good but is even with Wednesday. When we look at the percent of time the market gained, the overall pattern is the same: Monday makes gains less than half the time while the other days of the week are all above 50% though only slightly. The fourth and fifth columns suggest a different conclusion, however. When we look at only days that went up, Monday averaged the highest gains, 0.84%. When we look at only the days that went down, Monday was far and away the worst, averaging 0.94%. Taken together this suggests to me that Monday is most likely to be a more active day (relatively speaking) in either direction and the key difference to explain the overall negative returns on Monday is the lower percent of days where the market gained. Of the ten worst days during this period, six were Mondays! It is important to note here, that the probability differences are not very large. You could probably make a system to take advantage of this very small edge but the gains would seem slight and would probably disappear if you had any transaction costs. Furthermore, sometimes Monday is better. Of the 75 full years covered, in seven of those years, Monday was the best day of the week.

 

	Average Daily Change	Percent of Time Market Gained	Average Gain During Up Day	Average Loss During Down Days
Monday	-.11%	46%	0.84%	-0.94%
Tuesday	.04%	51%	0.76%	-0.70%
Wednesday	.08%	53%	0.78%	-0.72%
Thursday	.04%	51%	0.73%	-0.67%
Friday	.08%	54%	0.73%	-0.69%

If you think the market is mostly random, you might stop here and simply note this as an interesting anomaly. However, what we all know (smart people that we are) is that you need to think about context.

The simplest way to reframe this is to ask were there any differences when we take into consideration what happened the prior week. And this is where the more revealing picture unfolds as seen in the second table. When the prior week saw a gain in the DOW, the likelihood of any day of the week making a gain is about the same (50-54%). Monday fits in the range and the average gain made on Monday for such weeks is about even with Wednesdays and Fridays. HOWEVER, when the prior week was negative, the probability of Monday rising dropped to 40% and on average these Mondays lost 0.32 percent. But the other days of the week maintained their same rough probability of making a gain although the average gains were higher. This paints a more complex but useful picture about days of the week. It suggests that Monday is important but mostly when the prior week is down. At such times sellers apparently are much more likely to come out in force. This could be because many people only focus on weekly data or that weekends provide a time to fret if the market has dropped. Furthermore, the imbalance apparently mostly gets settled on Monday so that the rest of the week can resume the general pattern of price movement (apparently random or subtly trending depending on your viewpoint). However, when the prior week was up, the urge to sell is predictably not as strong so Monday looks pretty normal.

	Prior Week Positive, Percent of Time Gained	Prior Week Positive, Average Gain	Prior Week Negative, Percent of Time Gained	Prior Week Negative, Average Gain
Monday	51%	0.06%	40%	-0.32%
Tuesday	50%	0.01%	52%	0.08%
Wednesday	52%	0.06%	54%	0.09%
Thursday	50%	0.02%	52%	0.07%
Friday	54%	0.06%	54%	0.10%

If you substitute the word "trend" for prior week, I think you can see what most of us probably already know. Day of the week is not nearly as important as the larger trend. There may be useful patterns to exploit here but don't lose sight of the bigger picture. Also don't forget that these are still relatively modest probability differences. To say that for every five Mondays following a down week, 3 times the market will fall and 2 times it will rise paints a more complex situation to navigate.

For what it's worth

Author: Dale

It may help to have two types of exit strategies in mind: Proactive and Protective. The line won't be black and white but I think a distinction can be made and having both might be helpful.

A Protective exit strategy would be something like a trailing stop, resistance line, or a moving average penetration. Here effectively the exit is to protect you from losing more. I say more because by definition, these are all designed to get you out only after price as already fallen some amount from its last peak. You may still make money but it will be less than it was a few days/weeks earlier. This kind of exit strategy is basically an insurance policy. You are saying I bought (was holding) a fund/stock that I thought would go up (continue to go up) but it didn't so now I'm going to get out. This is good because as Art has shown, losses can only be made up with gains of greater magnitude. The problem is twofold. First, for many of us to have to invoke such a protective exit strategy at some level requires us to say I was wrong, the fund didn't go up. In a world where we are trained to think we have to be right all or at least most of the time, this can be psychologically very hard. Hence, you have people hold on to losing investments long after it seems all reason says sell. There has been a lot written about how to deal with this. Accepting "errors", calling it the cost of business, seeing this as successful corrections etc. I think it helps to realize you don't have to be right all the time. Look at Bob's Pony Express report on his investing contest. The last time I looked, exactly 50% of his trades lost money. However, he is still doing very well and winning the contest because the losses are small and his big gains are very big. The second problem is that as soon as you sell the fund may turn right around. Regardless of whether you buy right back in you may question yourself about making a new "mistake". Just don't forget that you did this as insurance and insurance has a cost too. But we still buy it to protect ourselves against something worse. At the end of each year do you beat yourself up if you didn't die (life insurance), have your house burn down (fire insurance), or crash your car (auto insurance)? No, but these insurance policies all cost you something. You can make "errors" with your exit strategy (just like Bob's entry strategy) and still come out way ahead because you are protected against catastrophes and you don't lose sight of what is important. Not losing all your money because then you can't play anymore.

A Proactive exit strategy would be exiting early, perhaps even while price is still rising, because you believe you are near a top or the fund is not making the gains you want. No we can't time the exact top, but we frequently get lots of clues about turns that might suggest we should exit. For example, if the rate of change is slowing, you might choose to exit. You could see this with MACD, steepness of your moving average, various ROC indicators, etc. You still might use a moving average or trendline to set the exit, but here you would be using a tighter line. For example, you might use the 50 day moving average as your protective stop line, but if momentum seems to declining, you might switch to a 10 day moving average for your exit. You might tie all this in with cycles, Elliot waves, or a variety of other such techniques. The underlying notion is that you will be getting out sooner than your protective exit strategy would dictate. If you don't like opposites, think of this as a more aggressive protective exit strategy rather than a different beast. It is even possible you may exit while price is still rising, although slowly. This reflects a saying you hear often, that "you have to leave something on the table" (probably applies more to stocks than funds). It is important to realize that here too you will not be perfect. You will sell and it will become apparent later that holding on would have been better.

Keys in my mind are to

1. Do this thinking ahead of time, not in the heat of the moment.
   
2. Think through what you want to do with your exit strategy and why. Pay attention to your feelings, risk tolerance, and objectives to design something that will fit you.
   
3. Write it down and pull it out when you get ready to buy and throughout your trade.
   
4. Remember why you are doing this and don't be hard on yourself if it seems you were wrong at the time.
   
5. Revist your exit strategy from time to time to assess how your exits are working and see whether it still fits you and your objectives. If not modify or completely change it. Repeat as necessary.
   

dale

P.S. As far as technical tools, I use a combination of different time periods for averages, trendlines, bollinger bands, rate of change indicators, and time cycles to help me set what I want.