Expected Returns: An Investor's Guide to Harvesting Market Rewards

24

Tactical return forecasting models

• Systematic forecasting models may focus on (1) timing a single asset or an asset pair or (2) selecting a portfolio across a cross-section of assets.

• Such models often use multiple indicators, such as value, carry, and momentum.

• The trading horizon can vary from within-day to a year. A monthly or weekly trading frequency is common.

• Numerous techniques have been applied to return forecasting. In out-of-sample evaluations, simple models often perform just as well as more complex ones.

• Data mining and crowding problems are inevitable. They can be mitigated but not fully avoided.

24.1 INTRODUCTION

Valuation indicators are effective in providing a long-horizon (multi-year) view on an asset’s prospects. In this chapter, I turn to dynamic trading strategies and forecasting models that have a shorter horizon (one week, month, or quarter). I first describe the generics—model types, assets traded, indicators—and then comment on possible improvements and pitfalls for the systematic trading style. I keep this chapter brief so as to retain some of my proprietary trade secrets, but Chapters 8 through 10 review several publicly known market-timing indicators for equities, duration, and credit, while Chapters 12 through 15 review four popular dynamic trading strategies: equity value, foreign exchange carry, commodity momentum, and volatility selling.

What types of models are used? Single-trade timing models and multi-asset selection models are ...