Applying Forecasting in Finance
The success of an investment is dependent on four tasks:
- Detecting causal patterns in a noisy world
- Making well-calibrated probabilistic forecasts about how these patterns will play out in the future
- Factoring these forecasts into investment decisions that deliver risk-adjusted returns
- Conducting post-mortems on investment failures and successes to make increasingly accurate forecasts and better trades
One classic challenge is determining the role of good luck and managerial skill in portfolio performance. Traditional methods either:
- Collect and observe data over many years. This is to allow for stable statistical estimates. However, this process takes time.
- A second method is to use case-study post-mortems, which rely on accurate causal inferences.
Cerniglia and Tetlock propose using the Alpha-Brier system to disentangle skill from luck quicker (Cerniglia and Tetlock, n.d.). This system requires firms to first construct an analytical infrastructure that allows them to capture in real-time the accuracy of forecasting judgments underlying investment decisions. The Alpha-Brier process also facilitates learning by giving investors the explicit accuracy feedback they need for making well-calibrated forecasts on the drivers of market outcomes.
Cerniglia, Joseph A., and Philip E. Tetlock. n.d. “Accelerating Learning in Active Management: The Alpha-Brier Process” 45 (5). Institutional Investor Journals Umbrella:125–35. https://doi.org/10.3905/jpm.2019.45.5.125.