Wednesday, March 31, 2010

Bonus Post: Calculating Price Targets

As I mentioned earlier today, in appreciation of the generous readership, I thought I would share some of my ideas and methods for calculating price targets. If you're new to this topic, it would be helpful to review my prior posts on hidden volatility assumptions and defining effective price targets with the previous day's data.

What we saw in that latter post was that using the previous day's high, low, and average prices provides us with relatively high probability targets for the current trading day.

In my own work, I do not use the average price as defined in the post (H+L/2). Rather, I use (H+L+2C/4). This is the "pivot" level that I post each morning for SPY via Twitter. This overweights the closing price relative to the prior day's high and low, so that--on average--the pivot price will be closer to the current day's open. Going back to late 2002 (N=1894 trading days), my Excel calculations show that we have touched the previous day's pivot on 70% of all trading days.

For this reason, the previous day's high, low, and pivot prices are key near-term price targets for my trading. As I mentioned previously, even closer price targets are the overnight high and low prices from the ES futures.

If I anticipate a slow trading day with a narrow price range and we open in the middle of the overnight and prior day's ranges, I will look for trades to take out the overnight high or low price and then the previous day's high or low. If I anticipate a slow trading day and we open nicely above or below the overnight and prior day's pivot levels (for overnight "pivot" I use the day's VWAP), I look for a move back to VWAP and then the previous day's pivot if buying or selling can't be sustained.

If I anticipate an average or busier trading day, I look toward more distant profit targets. Below is one way of calculating those that builds on the previous post.

FORMULAS FOR CALCULATING PRICE TARGETS

* Let us call the difference between yesterday's high and low prices R, for range. That means that the difference between yesterday's average price and yesterday's high is 1/2 R and the difference between yesterday's average price and yesterday's low is 1/2 R. (We're using average price, not the pivot level, for this calculation. More on pivot-based calculations in the next post in the series).

* If we calculate (yesterday's average price + 3/4 R), we will get a price level above yesterday's high that we'll call R1. If we calculate (yesterday's average price - 3/4 R), we will get a price level below yesterday's low that we'll call S1.

* Going back to late 2002, the odds of hitting R1 or S1 during today's trade are 67%. Two-thirds of the time, we'll hit R1 or S1. It's a high probability target if volume is average or better.

* If we calculate (yesterday's average price + R), we will get a price level above R1 that we'll call R2. If we calculate (yesterday's average price - R), we will get a price level below S1 that we'll call S2.

* Going back to late 2002, the odds of hitting R2 or S2 during today's trade are 41%. We want to see above average relative volume (and today's volume > yesterday's volume) to assume that we'll touch R2 or S2.

* If we calculate (yesterday's average price + 5/4R), we will get a price level above R2 that we'll call R3. If we calculate (yesterday's average price - 5/4R), we will get a price level below S2 that we'll call S3.

* Going back to late 2002, the odds of hitting R3 or S3 during today's trade are 26%. We would need to see significantly above average relative volume (and today's volume significantly > yesterday's volume) to assume that we'll touch R3 or S3.

VARIATIONS OF THE ABOVE WORTH RESEARCHING:

* Instead of using yesterday's average price as a base for calculation, you can use the traditional pivot formula of (H+L+C)/3.

* Instead of using yesterday's average price as a base for calculation, you can use today's open. That is especially helpful when the overnight session leads to an opening price far from yesterday's average price.

* Instead of using R values based on yesterday's trading range, use the average trading range from the prior N days. My research shows some benefit to going out several days, but returns are diminishing out to a five-day lookback.

Regardless of your calculation method, you will find that R increases as the market's volatility increases and decreases as the market's volatility wanes. This automatically adjusts your price targets for the market's most recent volatility.

Going back to late 2002, yesterday's volatility correlates with today's volatility by a whopping .75. That means that we can predict more than half of the variance in today's volatility simply by knowing the prior day's trading range. If we go out to a five-day period, the correlation between the prior five-day's average range and today's range has been .80.

Once you become good at tracking today's volume relative to yesterday's (or the prior five days'), you can make very reasoned estimates as to which levels we're likely to hit during the day. That considerably strengthens our exits and helps us maximize our risk/reward.

This post and the next one (tomorrow) will remain on the blog for a limited time. If the research is of interest, you might want to print out the post or copy the relevant data.

Thanks again for all the interest and support--

Brett
.

21 comments:

tt said...

Thanks Dr Brett. Rewarding your loyal readers this way is a class act.

Student of the Market said...

Second that. Thanks for all your efforts

dsorense56 said...

Dr. Brett,
Thank you for sharing the detailed reasoning for your pricing strategy. This already sheds some light onto why my price determinations are frequently a "near miss".

wildman said...

Very sound set of formulas. Indeed I had compared with others and had found solid. Thank you. Best regards.

BirdMan said...

Dr Brett
As always... It is us that should be - and are - thanking you!
Dave

Dr Bill said...

Wow! All the cool variations to play with and correlate. For math guys like me, this is catnip.

Dr K

j said...

entertaining MV too =)

SSK said...

Thanks for your insight as always! Keep up the great work! Best, SSK

j said...

Where/how do I get yesterday's average price? (I have the day's volume but not the value)

George said...

Thanks for sharing all that, and for taking the time to write it all up.

Charles said...

Doc the pleasure of patronage is all mine.

Curious...why the change in font?


Charles

RoachApproach said...

For calculating the pivot you could also use price/volume and market profile data. This is especially easy if you have only one POC.

Cheers,
Markus

TST said...

Long live Dr Brett

Don said...

Thank you Dr. B. As always, your knowledge and sharing is greatly appreciated. Also, I am enjoying "The Daily Trading Coach."

Michael said...

Thanks for all that you do Dr. Brett. Do you consider overnight highs and lows to be prices reached prior to the open? Just wanted to clarify. Thanks again!

Mike

Brett Steenbarger, Ph.D. said...

Hi Michael,

None of the formulas for SPY as outlined in the post make use of overnight data--

Brett

edwin said...

Hi Dr. Brett,
Volatility is proportional to volume. I've known that for the longest time. However your description took my understanding from a 2 dimensional cognitive understanding of a map to a 3 dimensional experience of the territory.
Thanks,
Edwin

Buck said...

Dr. Brett,
Great stuff as always. Could you briefly touch on "evidence of directional bias to the day's trade". I have usually stayed away from trading the open, but it's frustrating watching the TF regularly move multiple points in one direction from the open. Any comment would be appreciated.
Thanks again.
Joseph

Eagle said...

thank you , this is a noble gesture, as important as your mentoring online

OptionsVic said...

Dr Brett,

In your 'bonus post #2' you have a pivot as :
Pivot = (H + L +2C)/4

while in this one you it as
(H+L+2C/4)
Trust this one has typo.

Thanks for sharing these valuable nuggets with your fans.

JackAz said...

Dr. Brett,

You're writing is helping me be a better trader. Thanks for this gift you give to the web, (and to me!)

Question: Why have you limited your research to 2002? I take ES data as far back as I can get it, (1997). I don't have any training in statistical analysis, but my understanding is that the bigger the sample set, the more reliable the results. I am not questioning your results, just questioning why you have chosen to limit the size of the sample.

Thx