Friday, April 02, 2010

Clarifications Regarding Price Target Calculations

I've been inundated with questions regarding the first and second posts detailing methods of calculating price targets. Here are responses to the more common questions:

* All calculations are in Excel; Excel functions will calculate medians and averages for you. No complicated spreadsheet programming or statistical software is needed;

* The data come from my archives, which is why they go back to late 2002. It's a sample of convenience that cuts across a variety of market conditions;

* Other values can be used, and I encourage readers to experiment with their own formulas. In place of the day's average price, you could use the pivot level for the day or the day's VWAP. A promising variation is to use today's open and calculate price targets around that;

* The basic approach from my second post can be used for any stock or ETF. You'll simply need to adjust the constant in the formulas for R1/S1, R2/S2, R3/S3, etc. Instead of using .60, .80, and 1.0 as in SPY, for instance, you'll need to define the proper constants for each market;

* Using weekly data instead of daily data will give you price targets for the following week. (Those constants need to be adjusted as well). That is very useful for swing traders. I post weekly price targets for SPY each Monday morning via Twitter;

* Formulas for the ES futures will look different, because the pivot and volatility calculations will incorporate overnight trading data. With SPY, there is no overnight data embedded in the formulas.

Hope that's helpful. My goal in providing the formulas is to encourage you to think of trading in a different way, with an emphasis on exits and targets, not just entries. If my posts raise questions and lead you to explore the data on your own and find relationships different from the ones that I have shared, I will have succeeded in my mission!

Thanks as always for the interest and support.
.

9 comments:

ES_Addict said...

Dr. Brett.

Thank you for an insightful series of thought provoking articles regarding target planning and identification.

Your blog in general is an asset to any trader. However, some of your content really hits it out of the park and the "Bonus" series is a fine example.

Thanks again.

The Sanch said...

You're an excellent man, Brett.

Will said...

Thank you again Dr Brett. Looks good on my chart.

Dana said...

Can someone explain to me how dividing the range of the day by the opening price affects the value of the Volatility Estimate? Why not just use the actual range on its own without the added step of the divider? I'm not mathematically inclined enough to see the purpose, yet I don't doubt the validity of the formula. Thanks in advance to the math guy that can explain this to me!

Norm said...

@dana: I don't know much but I think the reason the for the dividing is as follows. Imagine an stock which had a spread of $1 during a day. Is that a lot of volatility? Well it is if the stocks price is only $2 but what if the stock was Berkshire whose price is $12000. Well $1 variance on $12K stock not too volatile is it. So by dividing by the price equalizes the volatility for all price ranges.

flan4 said...

Brett,

How do you determine your constants, i.e. .60, .80, etc. You mentioned that one would need to taylor the constants to different markets. How would one go about determining what constants to use, based on what?

Thanks for sharing!

bleurain said...

Has anyone figured out how to determine the constants for other stocks or weekly time frames?

Brett Steenbarger, Ph.D. said...

Hi Bleurain,

The constants for weekly figures and for individual stocks and ETFs simply have to be established empirically, based on the percentage of times you want R1/S1, R2/S2, and R3/S3 to be hit. Once you determine those percentages, it's easy with some trial and error to establish the proper constants.

Brett

Joe said...

Brett,

Regarding this post, "The constants for weekly figures and for individual stocks and ETFs simply have to be established empirically, based on the percentage of times you want R1/S1, R2/S2, and R3/S3 to be hit. Once you determine those percentages, it's easy with some trial and error to establish the proper constants."

How do you go about testing the data? Do you create S1's, R1's, etc. using different constants by hand for a variety of randomly selected days and record the percentage of times they're hit? Do use excel? Thanks.