Last few weeks we talked about clustering in context of MCAP and a simple relative ranking percentile score. We also talked about a “Growth Bias”, something that defines the stock market aka running after winners, winner’s bias, momentum or plainly speaking rich-get-richer. Today, we extend the idea further into dynamic mechanisms.
The stock market is a dynamic system characterized by continuous fluctuations and complex interdependencies among its constituents. Understanding these dynamics requires analyzing stock returns vs. Relative percentile ranking scores over different periodicities, observing changes in shapes, slopes, and curvatures, and interpreting the underlying mechanisms.
Enthused by what we saw in annual periodicities pointing to slope towards the end of top performers viz. the growth bias, we decided to extend the test case to weekly, monthly, quarterly, and multi-year intervals. We had an intuition of what we would find, but the results were more compelling. Each periodicity provided unique insights into market behavior. Weekly returns ranged between +5% to -20%, a relatively low band, occasionally exhibiting high volatility and noise. This was expected as returns need time to build into a persistent trend and psychological amplification (herding). Monthly and quarterly returns expanded the ranges while smoothing out some of the volatility. The longer term translated into higher slopes compared to weekly maps. However, it was the multi-year maps, both 2 year and 5 years that had pronounced curvatures.
The evolution of the visuals across different periodicities indicates a dynamic mechanism at play, reflecting the interconnectedness and continuity of market behavior over time. The performance of stocks over short-term intervals influenced their medium-term which in turn influenced the long-term trends. The relative percentile ranking clusters revealed distinct biases: a growth bias where higher percentile stocks consistently show higher returns, a core bias representing moderate returns, and a value bias indicating lower returns and as the time increased, the relatively linear action turned into a wavy non-linear, sinusoidal behavior. These shapes and forms illustrated the interconnectedness between relative percentile scoring states across time.
Since, there is limit to the integrity of data, more than a decade is a long time for average winner in S&P 500 holdings, and investors have average holding constraints, it is safe to assume that the wavy nature of growth bias - value bias maps, is driven by an underlying dynamic mechanism which not only knows how to perpetuate but also how to replenish itself.