johncw_97yahoo'com 1

1
Volumes Value episode II

• Continuation of April 26th 2005 meeting

2

• Revisit abnormal trading days
• Look at a few normalization indicators
• Relationship between price and volume

3
Martin J. Pring1 summarized the following

• Volume is measured in trends, and the trends are
  always interpreted in relations to the recent
  past.
• It is normal for volume to expand with rising
  prices and contract with declining ones. Anything
  to the contrary is abnormal and warns of an
  impending trend reversal.

4
Martin J. Pring1 summarized the following
(cont)

• During bullish trends, it is normal for volume to
  lead the price.
• Selling climaxes clear the air. They do not
  necessarily signal the final low for the move but
  are almost always followed by a rally.
• Record volume coming off an important low usually
  signals a strong rally

5
Martin J. Pring1 summarized the following
(cont)

• A parabolic expansion in price and volume
  represents an exhaustion move, which is typically
  followed by a sharp decline.
• Pring2 has expanded material on volume.

6
Singal3 (in chapter 4) concluded

• In general, there is no evidence of tradable
  price regularities following large price events.
  If large price changes are accompanied by other
  traits of information, such as high volume and
  public dissemination of news, the patterns become
  stronger. The annualized return after
  transactions costs can be estimated at 36
  annually for positive price changes and 15 for
  negative price changes. This strategy works even
  in bear markets!

7
Average Return on Day of Large Change for stocks
gt 10

• Singal3 states also that the average return on
  day of large change in 1990 to 1992 in increasing
  relative return on day 0, was 7.13 and 0.08
  from day 1 to day 20. For average decline on day
  0 of -7.13 there was an return of -0.48 by
  day 20.

8
Brian R. Bell4 states

• Normalizing an indicator allows you to do several
  things.
• Allows historical analysis to become easier since
  the values of the indicator are more consistent
  over long periods.
• Cross-market analysis becomes possible, since the
  values of the indicator are more consistent
  across different instruments.

9
Brian R. Bell4 states

• Analysis over several time frames becomes easier,
  since the values of the indicator can be made
  consistent.
• Effects from sudden changes in volatility can be
  removed.

10
Some normalization techniques

• Bell4 gave example of a 4/8 moving average
  price oscillator normalized to the 1st 4 methods
  techniques listed below
• Average price
• Standard deviation of price
• Average true range of price
• Range of the oscillator itself
• Using log function as seen (aka Jeff) in TASUG
  meetings. See also Mandelbrot 5 chapter 4
  Images of the Abnormal for price examples.

11
Examples of what is abnormal volume?

• One of the criteria Bollhorn6 uses to check for
  Abnormal Activity to Predict New Upwards Trends
  is to compare the volume today, relative to the
  20 day volume average.

12
Examples of what is abnormal volume?

• Swing7 produced a scan based on Pritamani et
  al8. Criteria is as follows
• Liquidity of 100,000 traded/day on average.
• Twice the average volume.
• Positive short-term momentum.
• Exceptionally strong intra-day trend.
• Note that reference 3 and most likely 8 are
  based on information in reference 9.

13
Examples of what is abnormal volume?

• // Translation of Swing7 scan using in
  Amibroker10 afl
• // liquidity 100,000 shares traded per Day, on
  average.
• MAv MA(V,20)
• // require the High Volume, as in Pritamani AND
  Singal8.
• // Specifically, twice the average Volume.
• Filter_v MAvgt100000 VgtMAv2
• // Require positive short-term momentum.
• // Use the directional indicators and the 1-day
  change as proxy for this.
• Filter_d PDI(14)gtMDI(14) (C/Ref(C,-1)-1) gt
  0.05
• // Finally, we require an exceptionally strong
  intra-Day trend,
• // as calculated by how much of the intra-Day
  volatility translated
• // into upwards movement.
• Filter_t (Close-Open)/(High-Low) gt 0.75
• // Exploration scan using criteria above
• Filter Filter_v Filter_d Filter_t

14
Examples of what is abnormal volume?

• Some one had mentioned that one should look at 3
  times the average volume

15
Examples of what is abnormal volume?

• Bajo11 computed the normalized abnormal volume
  (NAV) with Number of Trading Days (NTD66) using

16
Examples of what is abnormal volume?

• Bajo11 computed the normalized abnormal volume
  (NAV) with NTD66.
• It is just a normalized standard deviation
• technique. It is written in Amibroker afl as
• nav (V - MA(V,66))/StDev(V,66)
• In statistics, this formula is known as
• Z-score.

17
Using volume standard deviation method as in
Bajo11

• Using standard deviation(SD), we know that 95.5
  of the change in volume should be within 2SD and
  99.7 within 3SD.
• Bajo11 used the levels of over 2.33SD and 3.1SD
  for indicating abnormal volume.
• Since there are about 250/251 trading days in the
  North American markets, using 20 (NTD60) or 21
  (NTD63) trading days per month may be
  appropriate. References 3,6,7,8,9 uses 20 days
  per month in SD or average calculation.

18
Normalization Consideration?

• When normalizing, one may consider not using the
  current price or volume in the calculation. For
  moving average normalization, instead of
  2MA(V,20), one can use twice the previous
  average volume. i.e. 2ref(MA(V,20),-1).
• Period used.
• Advantages/Disadvantages?

19
Example Charts using AmiBroker10

• Next set of slides are example charts of some
  normalization methods with different periods.

20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
Normalization summary

• Volume Standard deviation
• Vsd (V ref(ma(V,period),n))/ref(stdev(V,period
  ), -n)
• Volume Moving Average
• Vma V/ref(ma(V,period),-n)
• Volume with Range
• Vhr 100(V - LLV(V,period))/(HHV(V,period)-LLV(V
  ,period))
• For starters, try using value period from 20 to
  60 and n should be at least 1. On page 507 of
  Kaufman12, he uses n5 for formula (2) above.

29
Misc

• Misc
• Bollhorn6 examples AAPL, MSFT
• Normalized price
• Scans
• Other normalized indicators

30
Statistically analyzing volume

• One method of visualizing data distribution is by
  plotting all the data in something like a scatter
  plot.
• By using methods like in Goodman13, one may be
  able to help answer question like Is an increase
  in the activity of a stock a meanful indication
  of the direction of the price?

31
Statistically analyzing volume

• Following two slides are scatter like plots. The
  second one is zoom in view of the first.

32
(No Transcript)
33
(No Transcript)
34
Statistically analyzing volume

• One can tabulate the data in groups and display
  the result as a spreadsheet.
• Note to make things simple, when you see the
  range like n-m, the values are greater than n and
  up to m. i.e. 0-2 means greater than 0 and
  including 2.

35
USvolumeTest_61_F20030101T20041231_0sv_P20_V1.csv
36
Statistically analyzing volume

• As a map

37
(No Transcript)
38
Statistically analyzing volume

• As a bar chart

39
(No Transcript)
40
Statistically analyzing volume

• As a colour bar chart

41
(No Transcript)
42
Filename/chart meaning

• Note that the data file names end in V.dat. When
  is
• 0 then the volume of at least 100000.1 then
  there is also a bullish candle requirement of
  (C-O)/(H-L)gt0.75
• 2 then it uses the basic requirement of Swing7
  see slide 13.
• The first day is day 0. So if we were to buy,
  we use the next day (i.e. 2nd day) close
• 0sv denotes standard deviation volume vs
  absolute price change on that day
• 1mv denotes moving average volume vs absolute
  price change on that day
• 2pr denotes price range vs price change
• 3di denotes standard deviation volume vs price
  change range
• 4fp20d denotes price range vs future price
  change (20th day close over 2nd day close)
• 5fq20d denotes price range vs maximum future
  price change in 20 days over 2nd day close
• 6fr denotes price change vs the close of n day
  over the 2nd day close

43
Filename/chart example

• The file name
• USvolumeTest_61_F20030101T20041231_0sv_P20_V1.dat
  
• denotes that
• the watchlist is called USvolumeTest
• the watchlist number is 61
• analysis is from 2003-01-01 to 2004-12-31
• (0sv) standard deviation volume vs absolute price
  change chart
• P20 using a 20 period in the normalized volume
• V1 with additional bullish candle requirement
  of (C-O)/(H-L)gt0.75

44
US sample example

• Sample of 1814 US stocks.
• Minimum volume of 100000
• Period of 2003-01-01 to 2004-12-31.
• Following slides are colour charts of the
  results.
• The outputs are produced by running Amibroker 3
  times (i.e. V0,1,2) on the watchlist
  USvolumeTest and then running Gnuplot14 on the
  charts.

45
(No Transcript)
46
(No Transcript)
47
(No Transcript)
48
(No Transcript)
49
(No Transcript)
50
(No Transcript)
51
(No Transcript)
52
(No Transcript)
53
(No Transcript)
54
(No Transcript)
55
(No Transcript)
56
(No Transcript)
57
(No Transcript)
58
(No Transcript)
59
(No Transcript)
60
TSX sample

• Using a sample of 725 TSX stocks and applying the
  following
• Minimum volume of 100000
• Period of 2003-01-01 to 2004-12-31.
• Following slides are colour charts of the results.

61
(No Transcript)
62
(No Transcript)
63
(No Transcript)
64
(No Transcript)
65
(No Transcript)
66
(No Transcript)
67
(No Transcript)
68
(No Transcript)
69
(No Transcript)
70
(No Transcript)
71
(No Transcript)
72
(No Transcript)
73
(No Transcript)
74
(No Transcript)
75
(No Transcript)
76
Discussion

77
References

• Pring, Martin J., Volume Basics, Stocks and
  Commodities, July 2000, Volume 18, No. 7, pages
  36-41.
• Pring, Martin J., Technical analysis explained,
  McGraw-Hill 2002.
• Singal, Vijay, Beyond the random walk a guide
  to stock market anomalies and low-risk
  investing, Oxford University Press 2004, Chapter
  4 Short-Term Price Drift, pages 56-77.

78
References

• Bell, Brian R., Normalization, Stocks and
  Commodities, October 2000, Volume 18, No. 10,
  pages 58-68.
• Mandelbrot, Benoit et al, The (Mis)behavior of
  markets, Basic Books 2004, pages 88-94.
• Bollhorn, Tyler, Using Abnormal Activity to
  Predict New Upward Trends, 2000.
  http//www.smallcapcenter.com/help/pdf/abnormal.pd
  f

79
References

• Swing, Larry, Breakout Stocks, May 12 2005.
  http//www.mrswing.com/artman/publish/article_1029
  .shtml
• Pritamani, Mahesh and Singal, Vijay, Return
  Predictability Following Large Price Changes and
  Information Releases, Journal of Banking and
  Finance, April 2001, Volume 25, No. 4.
• Pritamani, Mahesh, Return Predictability
  Conditional on the Characteristics of Information
  Signals, PhD Dissertation 1999.
  http//scholar.lib.vt.edu/theses/available/etd-042
  399-112528/

80
References

• AmiBroker 4.70, http//www.AmiBroker.com
• Bajo, Emanuele, The Information Content of
  Abnormal Trading Volume, May 24 2005 draft
  paper. http//papers.ssrn.com/sol3/papers.cfm?abst
  ract_id313582
• Goodman, William M., Statistically Analyzing
  Volume, Stocks and Commodities, November 1996,
  Volume 14, No. 11, pages 465-470.

81
References

• Kaufman, Perry J., New Trading Systems and
  Methods, Wiley 4th Edition Feb 2005.
• Gnuplot data and function plotting routine.
  http//www.gnuplot.info