X hits on this document

PDF document

Maximizing Equity Market Sector Predictability in a Bayesian Time Varying Parameter Model* - page 45 / 46

279 views

0 shares

0 downloads

0 comments

45 / 46

Figure 3

Returns on Dynamic Sector Allocation Model Cummulative returns for a dynamic sector allocation model using 5 different econometric specifications are charted. For each case, total available capital is split evenly between a long portfolio based on the highest predicted sector return and a short portfolio based on the lowest predicted sector return. For the expanding sample strategy, parameters are updated by OLS every period beginning at T = 250 using the full available sample through time t. For the discrete rolling strategy, parameters are updated by OLS every 50 periods for the sample from t-T to t. For the 5yr rolling sample strategy, parameters are updated by OLS weekly using a sample from t-T to t. For the 1yr rolling sample strategy, parameters are updated by OLS weekly using a sample from t-49 to t.For the Bayesian beta strategy, variance parameters are estimated every 50 periods and the beta parameters are updated using the Kalman filter. All results are based on forecasts for period t+1 using parameters and data estimated through period t. The total sample is 429 observations. Estimation results for each strategy begin in period 250 (October 14, 1994). The starting capital level for the strategy is normalized at 100.

1yr Rolling OLS

Ka lm a n

5yr Rolling OLS

Discre te Rolling OLS

Ex pa nde d Sa m ple

450

400

350

300

250

200

150

100

50

28/10/02

28/07/02

28/04/02

28/01/02

28/10/01

28/07/01

28/04/01

28/01/01

28/10/00

28/07/00

28/04/00

28/01/00

28/10/99

28/07/99

28/04/99

28/01/99

28/10/98

28/07/98

28/04/98

28/01/98

28/10/97

28/07/97

28/04/97

28/01/97

28/10/96

28/07/96

28/04/96

28/01/96

28/10/95

28/07/95

28/04/95

28/01/95

28/10/94

44

Document info
Document views279
Page views279
Page last viewedSat Dec 10 22:17:11 UTC 2016
Pages46
Paragraphs1927
Words14448

Comments