Factor Model

Revision for “Factor Model” created on May 19, 2016 @ 12:45:08

TitleContentExcerptRevision Note
Factor Model
The central idea of modern financial economics: <em>The average return of a stock is the payoff for taking risk.</em>
<h1 id="FactorPremiumFactorExposure">Factor Premium & Factor Exposure</h1>
Given a stock and a risk factor,

<dl class="wiki"><dt>Factor Premium</dt><dd>quantifies the payoff to an investor who takes on this risk by buying the stock.</dd></dl><dl class="wiki"><dt>Factor Exposure</dt><dd>quantifies the exposure of the stock to this risk.</dd></dl><dl class="wiki"><dt>Average stock return per this risk</dt><dd>

average stock return = factor exposure X factor premium
<h1 id="Mathematics">Mathematics</h1>
Suppose we have <em>N</em> stocks whose returns depend on <em>K</em> factors.
<li>the factor premiums for the <em>K</em> factors: f_1, cdots , f_K</li>
<li>the factor exposures of stock <em>i</em>: beta_{i1}, cdots, beta_{iK}</li>
The return on stock <em>i</em>, r_i" , can be written as r_{i} = alpha_{i} + beta_{i1}f_1 + cdots + beta_{iK}f_K + epsilon_{i}

The average stock return is E(r_i) = mathbf{beta_i}'E(mathbf{f})

where, mathbf{f} = {left( alpha_i, f_1, cdots,f_K right)}'mathbf{beta_i} = {left( 1, beta_{i1}, cdots,beta_{iK} right)}'

<h1 id="FindingFactorPremium">Finding Factor Premium</h1>
Since return (r_i) and factor exposures (beta_{i1}, cdots, beta_{iK}) of the <em>N</em> stocks are known, we can compute, at any time point, the factor premiums (f_{1}, cdots, f_{k}) from the stock return equation using the <a class="ext-link" href="http://en.wikipedia.org/wiki/Ordinary_least_squares"><span class="icon">​</span>OLS</a> (Ordinary Least Square) <a class="ext-link" href="http://en.wikipedia.org/wiki/Regression_analysis"><span class="icon">​</span>regression</a>.

r_{i} = alpha_{i} + beta_{i1}f_1 + cdots + beta_{iK}f_K + epsilon_{i}

<h1 id="PanelRegression">Panel Regression</h1>
However, knowing the factor premiums on separate time points (e.g. each month or season) would not be helpful for prediction and analysis. We would like to know whether there exist stable factor premiums for a longer time period (e.g. 2~3 years).

Given there are <em>N</em> stocks over <em>T</em> time periods, left { (r_{11}, cdots, r_{N1}), cdots,(r_{1T}, cdots,r_{NT}) right }

Factor exposures of <em>N</em> stocks over <em>T</em> time periods, left { (beta_{11}, cdots, beta_{N1}), cdots, (beta_{1T}, cdots,beta_{NT}) right }

We can estimate the factor premiums (<strong>f</strong>) from the following equation using OLS regression,

r_{it} = beta'_{it}f + epsilon_{it}

<h1 id="OurImplementationinAlgoquant">Our Implementation in Algoquant</h1>
<a title="AlgoQuant" href="http://numericalmethod.com/up/algoquant/">AlgoQuant</a> has a <a href="http://redmine.numericalmethod.com/projects/public/repository/svn-algoquant/show/core/src/main/java/com/numericalmethod/algoquant/model/factormodel/qepm">package</a> on QEPM to model factors, exposure and premium. We currently support the computation of premiums using cross-sectional or panel regressions (factor premiums on separate time points).
<h1 id="Reference">Reference</h1>
<a class="ext-link" href="http://www.amazon.com/Quantitative-Equity-Portfolio-Management-Construction/dp/0071459391"><span class="icon">​</span>Ludwig B Chincarini and Daehwan Kim, Quantitative Equity Portfolio Management: An Active Approach to Portfolio Construction and Management, McGraw-Hill Library of Investment and Finance, 2006.</a>

Old New Date Created Author Actions
May 19, 2016 @ 12:45:08 webmaster
April 23, 2015 @ 18:04:50 Haksun Li
April 20, 2015 @ 18:45:24 webmaster
April 16, 2015 @ 13:10:37 Haksun Li
April 16, 2015 @ 13:10:30 [Autosave] Haksun Li
April 16, 2015 @ 13:02:43 Haksun Li
April 16, 2015 @ 06:35:56 sanjay
April 16, 2015 @ 06:35:15 [Autosave] sanjay
April 15, 2015 @ 16:05:31 webmaster
April 15, 2015 @ 15:49:53 webmaster
April 15, 2015 @ 15:49:19 [Autosave] webmaster
April 15, 2015 @ 06:06:37 sanjay
April 15, 2015 @ 05:38:53 sanjay