Monday, August 15, 2016

Evidence-based investing - Using prior beliefs in decision-making

I believe in evidence. I believe in observation, measurement, and reasoning, confirmed by independent observers. I'll believe anything, no matter how wild and ridiculous, if there is evidence for it. The wilder and more ridiculous something is, however, the firmer and more solid the evidence will have to be. 
-Isaac Asimov, scientist and writer (1920-1992)

Evidence-based finance can be viewed as a subset of quantitative investing. While the  pure quant will develop models and do all of his own testing, an evidence-based manager may use other information to drive the process. The evidence-based manager will scour as many sources for information of some relationship and use the work of others to serve as a prior for forming an idea or make a decision. Inherently bayesian, evidence-based decision-making weighs all of the prior evidence to derive an opinion on a decision rule. Actual new testing will be used to further support these priors from others.

Evidence-based decision-making is being used extensively in medicine and is used more often than admitted by many in finance. The key to good evidence-based decision-making is effectively weighing different evidence based on the quality of the tests and the analysis that is used. Now many quants will review the literature before they undertake their own tests, but evidence-based decision-making is more than just a simple review of past tests. It is means of weighing the work of others. 

An evidence-based approach can follow a simple process:
  • Form a specific question to be analyzed;
  • Find all of the existing evidence on the question;
  • Categorize the quality of testing and data employed as evidence; 
  • Form a decision rule based on the power of the evidence available;
  • Test using your own database and specification
  • Apply the developed rule based or prior and new analysis to your investment process;
  • Evaluate relative to past evidence;
  • Adjust the decision rule or eliminate.
This approach may be viewed as simple but the value is in forming a defined research structure that can stand up to testing and a weighing process. 

No comments: