The $100 trillion managed globally as your pension and investment assets is primarily discretionary. The human does the job. With or without the crash, the pie will become bigger. The future is about machines managing your pension and investments. There are many problems to solve, starting from bias to teaching investors how and why to trust the machine. But before anything, we need to pose the right questions for the machines to solve.
Cash is one challenge. Machine Learning (ML) today knows no Cash because nobody explains it why it should know Cash in the first place. The $35 trillion that is managed passively assumes a long-only world, where all crashes eventually come to an end, so why worry about Cash. Ok! that’s a good assumption to work within a world without survivorship bias. The long-only road is full of bankruptcy, junk, fraud, and instrument failures and it also has larger cycles, when something called “what if? deflation” happens. Passive or active, there is no free lunch. We are in an equity correlated world and equity is the preferred asset class, which means, it’s united we stand, united we fall. The challenge is for the active $65 trillion too, who struggle with idiosyncratic and systematic risk.
Classifications of Cash
1: CAPE Cash [1]. When CAPE reaches historical extremes.
2: Growth Cash. When Growth as an investment style delivers 2X vs. Value style ('08).
3: Death to Short Sellers Cash. The airlift that a 10 year and a 70-year bull trend create.
4: Robinhood Cash: When the law of ruin [2] fails. Profits used to be inversely related to transaction costs. Imbalances created by the new micro-market structure.
5: Commodities Cash: When there is a shift from paper to hard assets.
6: Agro Cash [3]: When Agricultural inflation happens at 0% interest rates.
7: Emerging Cash [4]: When emerging markets reach a secular underperformance extremity.
8: % Mutual Fund Cash [5]: Asset managers herd-like everything else in nature. At market tops, the majority of asset managers are invested and vice versa. Now they are all invested.
9: Cash is trash Cash [6]: When everyone knows Cash is trash, it’s in a deep discount.
10: Systematic Cash: When ML anticipates when and how much to go Cash.
ML knows no history because it is data-hungry and history is about sparse low-frequency data. Teaching ML to understand and know Cash involves reading history and understanding how asset, economic, cultural, and the multiplicity of cycles operate together and how it’s not information but informational state that is the answer for ML to know when to Cash.
Readings
[1] CAPE, Robert Shiller
[2] Jovanovic, Franck & Poitras, Geoffrey. (2006). A 19th Century Random Walk: Jules Regnault and the Origins of Scientific Financial Economics.
[3] Cash and Crash Cycles, M. Pal, Business Standard, Oct 2007
[4] All-roads-lead-to-china-and-emerging-markets, CFA Institute, 2018
[5] Ned Davis Research
[6] Ray Dalio Still Thinks ‘Cash Is Trash’, Bloomberg, April 2020