by Bjørnar Mundal, on 28 August 2020 11:11
Buffet Market Index (BMI) looks at the relationship between the total market cap and the GDP. It is a well known valuation indicator and Warren Buffet remarked in an interview that "it is probably the best single measure of where valuations stand at anygiven moment".
We used Deep Alpha to test BMI's accuracy at predicting bear markets on S&P 500. The test is performed on data going back to 01.01.1970, and without revealing too much: it has performed well over the period, and is definitely an indicator you should pay attention to.
Results:
Our analysis shows that the indicator is BEARISH if the 180 days Rate of change of the indicator is lower than -0.08. In other words, if we observe a negative momentum in the indicator, it signals BEARISH. Additionally, it can be viewed on as a leading indicator, as we perform a 30 day shift on the transformed indicator when testing it.
Interesting enough, our methodology finds a different way of assessing the indicator, compared to the norm. Traditionally, people look at the current level of the indicator and believes it signals overvalued markets if it surpasses a certain level (e.g 100%).
The 180 days rate of change is today -0.31. Therefore, todays BMI indicates a bear market. Interesting enough it went from BULL to BEAR in January.
The analysis shows that the F1-score is 0.60, implying a high predictive power. As we can see from the confusion matrix below, the indicator has over 87 % accuracy (True Bull + True Bear). Additionally, the CAGR figures shows a impressive effect of using it as a signal:
We like to see historical patterns and use those to navigate in the future. As you can see from the picture below, BMI has been good at calling bear markets since 1970:
For analytic purposes we have used our methodology and allowed Deep Alpha to find the best transformation of the indicator. If you are curious about the methodology we use to test and find the best transformation of each indicator, you can read more about it here:
