Scribd Logo
Upload

Welcome to Scribd!

  • Num Py Mad CAPM

    Uploaded by

    ivan2
    19 views2 pages
    The document summarizes an approach to stock selection that uses two risk-based filters: 1. The "mad CAPM" uses median absolute deviation as a more robust measure of risk than standard deviation in the Capital Asset Pricing Model to screen stocks. 2. A "liquidity filter" screens stocks based on their liquidity, estimated using geometric mean of daily volumes, with the goal of maximizing liquidity rather than expected returns. 3. The approach involves first screening stocks with the mad CAPM, then with the liquidity filter, to potentially profit from the resulting portfolio.

    Original Description:

    NumPy CAPM example

    Original Title

    NumPyMadCAPM

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    The document summarizes an approach to stock selection that uses two risk-based filters: 1. The "mad CAPM" uses median absolute deviation as a more robust measure of risk than standard deviation in the Capital Asset Pricing Model to screen stocks. 2. A "liquidity filter" screens stocks based on their liquidity, estimated using geometric mean of daily volumes, with the goal of maximizing liquidity rather than expected returns. 3. The approach involves first screening stocks with the mad CAPM, then with the liquidity filter, to potentially profit from the resulting portfolio.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    19 views2 pages

    Num Py Mad CAPM

    Uploaded by

    ivan2
    The document summarizes an approach to stock selection that uses two risk-based filters: 1. The "mad CAPM" uses median absolute deviation as a more robust measure of risk than standard deviation in the Capital Asset Pricing Model to screen stocks. 2. A "liquidity filter" screens stocks based on their liquidity, estimated using geometric mean of daily volumes, with the goal of maximizing liquidity rather than expected returns. 3. The approach involves first screening stocks with the mad CAPM, then with the liquidity filter, to potentially profit from the resulting portfolio.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    of 2

    Stock selection with the mad CAPM and liquidity filtering

    Ivan Idris The Capital Asset Pricing Model ( CAPM ) usually uses variance or standard deviation as a risk metric. I invented a slight modification of the model, which I call the mad CAPM ( well OK maybe I did not invent it, but I dont know if anyone else did ). M.A.D here is median absolute deviation not to be confused with mutual assured destruction or the other thing. It should be a more robust statistical measure of risk. The liquidity filter works almost the same way as the CAPM, but instead of trying to maximize expected return, the goal is to maximize liquidity. Liquidity is after all a good thing. So the plan for today is: 1. Screen stocks with the mad CAPM. 2. Screen stocks with the liquidity filter. 3. Profit!

    The mad CAPM


    The median absolute deviation of a sample is calculated by: 1. 2. 3. 4. Finding the median. Subtracting the median in the previous step from each element. Taking the absolute value of each difference. Determining the median of the absolute values.

    The code.
    1 def mad( arr ): 2 mdn = median( arr ) 3 4 return median( abs( arr - mdn ) )

    and unit test.


    1 ... 2 3 4 ... actual = mad( [ 1, 1, 2, 2, 4, 6, 9 ] ) assert actual == 1, 'incorrect mad' + str(actual)

    Liquidity filter
    The liquidity is estimated by the geometric mean of the daily volumes. So I apply the mad CAPM with this liquidity figure instead of the expected returns.
    1 2 3 4 5 6 7 ... ev = geomean( returns[ 0 ] ) evs.append( ev ) madC = mad( returns[ 0 ] ) mads.append( madC )

    8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

    liq = geomean( v ) liqs.append( liq ) beforeLastReturn = c[ len( c ) - 2 ] / c[ len( c ) - 3 ] - 1 if beforeLastReturn > 0: continue t = file.replace('.csv', ''), ev, madC, liq records.append( t ) ( a,b,residuals ) = fitline( mads, evs ) ( aLiq, bLiq, residuals ) = fitline( mads, liqs ) for t in records: symbol, evC, madC, liq = t if evC > a * madC + b: if liq > aLiq * madC + bLiq:

    ...

    If you liked this post and are interested in NumPy check out NumPy Beginners Guide by yours truly.

    You might also like