Powerpoint slides

394 views

Published on

Published in: Business, Economy & Finance
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
394
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
2
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Powerpoint slides

  1. 1. Using Sector Valuations to Forecast Market Returns A Contrarian View February 27, 2003 Lewis Kaufman, CFA Cira Qin Justin Robert Shannon Thomas Vidhi Tambiah
  2. 2. Table of Contents <ul><li>Overview Using Sector Valuations to Forecast Market Returns </li></ul><ul><li>Methodology A Contrarian View </li></ul><ul><li>Regression Results The Model’s Predictive Power </li></ul><ul><li>Out-of-Sample Limited Data, Promising Results </li></ul><ul><li>Trading Strategy A Long-Short Approach </li></ul><ul><li>ARCH Using Conditional Variance </li></ul><ul><li>Conclusions </li></ul>
  3. 3. Overview Using Sector Valuations to Forecast Market Returns <ul><li>Stock market is a discounting mechanism </li></ul><ul><li>Expectations drive stock prices, change over time </li></ul><ul><li>Sector valuations reflect these expectations </li></ul><ul><li>Assume markets driven by fear, greed </li></ul><ul><li>Use sector valuations to gauge sentiment </li></ul><ul><li>Build model to forecast returns </li></ul><ul><li>Key Takeaway: Sector valuations reflect expectations that can be used to forecast market returns </li></ul>
  4. 4. Methodology A Contrarian View <ul><li>Establish Framework </li></ul><ul><ul><li>High P/Es might indicate exuberance, despair depending on sector </li></ul></ul><ul><ul><li>Take contrarian view: sell greed, buy fear </li></ul></ul><ul><ul><li>Use P/E spreads to the market to normalize the results </li></ul></ul><ul><li>Identify Factors, Select Variables </li></ul><ul><ul><li>Investor sentiment Food Producers </li></ul></ul><ul><ul><li>Economic expectations Retailers </li></ul></ul><ul><ul><li>Geopolitical risks Oil and Gas Producers </li></ul></ul><ul><li>Test Intuition by Predicting t-Stats </li></ul><ul><ul><li>Food Producers (+), wide spread suggests fear, should be bought </li></ul></ul><ul><ul><li>Retailers (-), wide spread suggests high consumer confidence, should be sold </li></ul></ul><ul><ul><li>Oil and Gas Producers (+), wide spread suggests external shock, should be bought </li></ul></ul><ul><li>Forecast 1-Year Returns for the S&P 500 </li></ul><ul><ul><li>Identify whether sector valuations can forecast returns </li></ul></ul>
  5. 5. Methodology A Contrarian View <ul><li>Independent Variable Plot: Food Producers </li></ul><ul><ul><li>Suggests (+) relationship between spread, future returns </li></ul></ul>
  6. 6. Methodology A Contrarian View <ul><li>Independent Variable Plot: Retail </li></ul><ul><ul><li>Suggests (-) relationship between spread, future returns </li></ul></ul>
  7. 7. Methodology A Contrarian View <ul><li>Independent Variable Plot: Energy </li></ul><ul><ul><li>Suggests (+) relationship between spread, future returns </li></ul></ul>
  8. 8. Regression Results The Model’s Predictive Power <ul><li>Regression Output </li></ul><ul><ul><li>Adjusted R-square of 25.6% </li></ul></ul><ul><ul><li>Two of three t-stats significant at the 95% level; signs consistent with intuition </li></ul></ul><ul><ul><li>Low Correlation among independent variables </li></ul></ul>
  9. 9. <ul><li>Graphically Appealing </li></ul><ul><ul><li>Model does credible job of forecasting returns </li></ul></ul><ul><ul><li>More effective in recent years: access to information, trading volumes, hedge funds </li></ul></ul>Regression Results The Model’s Predictive Power
  10. 10. <ul><li>Encouraging Scatter Plot </li></ul><ul><ul><li>Graph suggests linear relationship between forecasted and actual returns. </li></ul></ul>Regression Results The Model’s Predictive Power
  11. 11. <ul><li>Other Observations </li></ul><ul><ul><li>Graph suggests linear relationship between forecasted and actual returns </li></ul></ul><ul><ul><li>Systematic positive bias in-sample, results encouraging out-of-sample </li></ul></ul><ul><ul><li>Strong predictor of directional change, implications for trading strategies </li></ul></ul><ul><ul><li>Model more effective in recent years: access to information, hedge funds, volume </li></ul></ul><ul><ul><li>Considered fitting in-sample data to more recent years and using an earlier period as out-of-sample. Better results for R-square and T-statistics. Dismissed idea because out-of-sample from past periods may not be indicative of success </li></ul></ul>Regression Results The Model’s Predictive Power
  12. 12. Out-of-Sample Promising Results <ul><li>Limited Data, but Encouraging Results </li></ul><ul><ul><li>Predicted curve clearly trends with actual returns </li></ul></ul><ul><ul><li>Promising given limited sample horizon; correctly predicted decline in 2000 </li></ul></ul><ul><ul><li>Model has a positive bias, expect predictability to improve when market rises </li></ul></ul>
  13. 13. Trading Strategy A Long-Short Approach <ul><li>Basic Strategy: Long-Short Approach </li></ul><ul><ul><li>Invest $1 in 1/73, invest $1 in 2/73, invest $1 in 3/73,… </li></ul></ul><ul><ul><li>Reinvest proceeds from 1/73 on 1/74, reinvest 2/73 on 2/74,… </li></ul></ul><ul><ul><li>Long-Short investment decision based on model’s predictions </li></ul></ul><ul><ul><li>Compare against benchmarks, market return and risk-free return </li></ul></ul><ul><li>Five Strategies </li></ul><ul><ul><li>Trading Strategy I: Basic Long-Short </li></ul></ul><ul><ul><li>Trading Strategy II: Long-Short with Risk-free </li></ul></ul><ul><ul><li>Trading Strategy III: Long-Short with Momentum </li></ul></ul><ul><ul><li>Trading Strategy IV: Conservative Long-Short with Conditional Variance </li></ul></ul><ul><ul><li>Trading Strategy V: Long-Short with Conditional Variance </li></ul></ul>
  14. 14. ARCH Using Conditional Variance <ul><li>Rationale </li></ul><ul><ul><li>Needed measure of future volatility to create trading strategy based on volatility prediction </li></ul></ul><ul><ul><li>ARCH is employed in strategies IV,V </li></ul></ul><ul><ul><li>We found lags 1,7 and 11 most significant </li></ul></ul><ul><li>The Results </li></ul>
  15. 15. Trading Strategy A Long-Short Approach <ul><li>The Results </li></ul><ul><ul><li>Out-of-Sample returns all outperform the market, with less volatility </li></ul></ul><ul><ul><li>Strategy III performs best across whole sample and in-sample. </li></ul></ul><ul><ul><li>Strategy IV dominates other strategies out-of-sample </li></ul></ul><ul><ul><li>Trading strategies outperform benchmarks in all data sets </li></ul></ul>
  16. 16. Conclusions <ul><li>Sector valuations reflect investor sentiment </li></ul><ul><li>By taking a contrarian view, we can make abnormal profits </li></ul><ul><li>Model supports thesis, outperforms both in-sample and out-of-sample </li></ul><ul><li>Systematic positive bias, though out-of-sample results are encouraging </li></ul>

×