2019-05-24

AI & Asset Allocation, Part 1

First part of an opinion in three parts




Part 1  |  Part 2  |  Part 3  |  Appendix 1  |  Appendix 2  >

The Turk marvelously won at chess
Imagine the sun warming you on the beach. You check the financial headlines on your mobile phone. A down day. You put the phone under the umbrella and take a dip in the sea. Returning to the umbrella, you ask the waiter to bring another cold one for you and your companion. This feels good. The next day, you check the financial headlines on your mobile phone. Stocks gained on the exchange. You log in to your brokerage. A few down and a few up, and a respectable increase overall for the last few years. You put the phone under the umbrella and take a dip in the sea. Returning to the umbrella, you ask the waiter to bring another cold one for you and your companion. This feels good. Just another couple of days for your kind of investor.





This triplet of articles is about the roles of AI (Artificial Intelligence), cash, pensions and Social Security, bond funds, stock funds, and individual stocks in your personal investment portfolio, that is, asset allocation.

Ample research shows almost no humans can consistently beat the S&P500 in the long term, when adjusted for risk (price fluctuation). Those humans who seem to do so (Warren Buffet, George Soros, a few others) can be explained by the luck of the draw.

I'm skeptical about the value of AI investing.

Parijat Garg asks good questions about AI investing. Considering the Turing Test, the message I take away is that when robots choose stocks like humans choose stocks, then we should expect similar results.

To take one example, some of the benefits identified by Schwab, a highly reputable firm, for their “Intelligent Portfolios” product include these (enhanced by my opinion):

1. Tax-loss harvesting: it's probably good during relatively stable markets, but I doubt what value this provides, compared with waiting till the end of the year, selling your losers, waiting long enough to avoid IRS wash-sale rules, then buying them back. I wonder what effect tax-loss harvesting will have on your net worth during a crash and recovery.

2. Daily portfolio rebalancing: Rebalancing assures portfolio allocation to avoid excessive concentrations of relatively risky assets. I question whether there's significant value produced, compared with quarterly rebalancing, and I doubt the value of quarterly versus annually, and even if you miss a year, there's not much to gain by remembering it instead. An advanced researcher at Mellon, Mr. Jeff Ricker, studied the effect of rebalancing and asset reallocation on taxable accounts and found negative effects on portfolio value. That is, the implicit advice is: Make the best decision available to you when you buy, then don't sell (unless you have major personal need for cash or major change in investment strategy). If you think rebalancing gets more valuable with increasing frequency, then rebalance daily, or perhaps hourly, and AI robotic methods do rebalance correctly with little cost.

3. Portfolio content selection: Using AI has no value compared with human choices or buying a copy of Barron's and throwing darts at the stock list pages. There is some dispute about the value of choosing investments by throwing darts (Mr. Ricker, when at Wells Fargo Investment Advisors, preferred a professional competition grade blowgun) or taking financial advice from blindfolded chimpanzees. Rick Ferri writes in Forbes that the monkeys easily beat the Dow Jones Industrials, time after time. Alex Mayyasi writes that the professionals beat the monkeys, though maybe by having their picks published in the Wall Street Journal or by taking risks or by cheating.

I skeptically want to see how AI portfolio management does in a crash before I do anything more than a minimal exploratory speculative investment.

Before putting you into their AI product or any other investment product, your broker, with some prudence and regulatory compliance, puts you through a risk preference interview or checklist. If called "AI", that is pure marketing nonsense. Some one or some computer looks up your answers in a table or chart or calculates a simple linear formula or merely applies human judgement to the risk and reward preferences of a person “like you”. Very like putting a small person inside an "automaton" that marvelously wins at chess.

My own advice for fortunes under $10m in the USA, taking a long 5+ years view:

Bonds and cash...

Your Social Security (national pension in the United States) and other pension payments provide the effect of a long-term government bond with the present value of the future payments. You already own these future payments. An investment in bonds will do no better for you. Bond prices fall when interest rates rise, a mathematical equivalence. Currently we remain at generational low levels of interest rates. The rate on a 30-year US Treasury bond is about 3% at April 30, 2019. Expect bond values to fall as interest rates rise in the future. Even if there is no change in rates and value, you may want a higher return on your invested money. If your social security payment is $1,000 per month, and if you want 5% per year or more from your investments, then your future social security payments are about equal to a bond that pays $12,000 per year forever (or until your death, whichever comes first), and the value of your forever bond is about $240,000 (=$1,000*12/.05), or very slightly less for the probability of death in 30 years or so. If you aren’t old enough to collect Social Security, then the forever bond is worth somewhat less, and will grow as you approach the qualifying age.

In Part 2, we will consider using Bond ETFs (Exchange Traded Funds) and Stock ETFs.

Part 1  |  Part 2  |  Part 3  |  Appendix 1  |  Appendix 2  >





Images

Fischchen, Breeding Ostrich in Berlin Zoo (May 27, 2007, shared with attribution under GNU Free Documentation License, Wikimedia.org, https://commons.wikimedia.org/wiki/File:Breeding_Ostrich_Berlin.jpg)

Mark Hirschey, Warren Buffett portrait (2005, used with permission, retrieved Jun 14, 2019 from https://en.wikipedia.org/wiki/Warren_Buffett#/media/File:Warren_Buffett_KU_Visit.jpg)

Lord Kelvin (circa 1900, Messrs. Dickinson, London, New Bond Street, Public Domain, https://en.wikiquote.org/wiki/William_Thomson)

Edwin Lefevre snapshot (published in “The Bookman”, v.25, 1907, Public Domain, retrieved Jun 14, 2019 from http://babel.hathitrust.org/cgi/pt?id=uc1.$b623112;view=1up;seq=137;size=200)

Maersk Line, The Denmark Staff (1914, Wikimedia.org, shared with attribution under Creative Commons Attribution-Share Alike 2.0 Generic licensehttps://commons.wikimedia.org/wiki/File:The_staff_(1914)_(7312784848).jpg)

"The Turk:, engraving, published in Gottlieb von Windisch, "Inanimate Reason" (1784, Wikipedia, https://en.wikipedia.org/wiki/The_Turk, retrieved May 24, 2019, Public Domain)

Images in Appendix 2 are from Wikipedia.org and in the Public Domain or made available via Wikipedia's Creative Commons Attribution-ShareAlike License


Sources

Warren Buffett, Chairman’s Letter to Sharelholders, Feb 27, 2009 (published in Berkshire Hathaway, Inc., “2008 Annual Report”, letter copyright 2009 by Warren Buffett, retrieved Jun 14, 2019 from http://www.berkshirehathaway.com/2008ar/2008ar.pdf)

Department of the Treasury, “Daily Treasury Yield Curve Rates”, (https://www.treasury.gov/resource-center/data-chart-center/interest-rates/Pages/TextView.aspx?data=yield, retrieved Apr 30, 2019)

Rick Ferri, “Any Monkey Can Beat The Market” (Dec 20, 2012, Forbes, https://www.forbes.com/sites/rickferri/2012/12/20/any-monkey-can-beat-the-market/#4acf7191630a, retrieved Apr 29, 2019)

Parjat Garg, “Discretionary Investing in the Age of Artificial Intelligence” (retrieved Apr 29, 2019, CFA Institute, https://www.the-right-question.org/en/discretionary-investing-in-the-age-of-artificial-intelligence/)

Johnathan Hoenig, “Find Your Stock Market 'Sleeping Point'” (May 9, 2011, Marketwatch, retrieved Jun 16, 2019 from https://www.marketwatch.com/story/find-your-stock-market-sleeping-point-1304958203774)

Mark Hulbert, "Almost all retirees make this mistake" (Jul 27, 2019, Marketwatch, retrieve Jun 26, 2019 from https://www.marketwatch.com/story/almost-all-retirees-make-this-mistake-2019-07-26?siteid=nbch)

Edwin Lefevre, “Reminiscences of a Stock Operator” (1923, https://www.amazon.com/Reminiscences-Stock-Operator-Edwin-Lefèvre/dp/0471770884/ retrieved Jun 14, 2019)

Burton G. Malkiel, “A Random Walk Down Wall Street” (1973-2019, https://smile.amazon.com/Random-Walk-Down-Wall-Street/dp/1324002182)

Alex Mayyasi, “How Well Do Blindfolded Monkeys Play the Stock Market” (Jan 31, 2014, https://priceonomics.com/how-well-do-blindfolded-monkeys-play-the-stock/, retrieved Apr 29, 2019)

Charles Schwab & Co., "Intelligent Portfolios" (2019, https://intelligent.schwab.com/)

Alan Turing, pp. 3-5 (1951, transcript, BBC broadcast 1952, conversation with M.H.A. Newman, AMT, Sir Geoffrey Jefferson and R.B. Braithwaite, The Turing Archive, http://www.turingarchive.org/viewer/?id=460&title=5)

Dickson G. Watts, “Speculation as a Fine Art (, https://www.amazon.com/dp/1607962659, retrieved Jun 14, 2019) ; nb. The story has been retold many times, of course, and attributed to various persons, no doubt apocryphally, and may predate Watts by centuries. As Watts told it, it's much more terse: "One man told another that he could not sleep on account of his position in the market; his friend judiciously and laconically replied: 'Sell down to a sleeping point.'" As the legend came down to me, a man approached J.P. Morgan on Wall Street, saying "I worry I am overinvested in bonds. Is the market going down? Should I sell it all and hold the cash? Please tell me what to do, Mr. Morgan. I can't sleep at night!". Morgan replied "Sell until you can sleep at night." And Mr. Hoenig, writing in Marketwatch, seems to have heard the same attribution, but a slightly different story than I. 

Wikipedia, “The Turk” (retrieved Apr 29, 2019, https://en.wikipedia.org/wiki/The_Turk)



Image by Daniel Brockman, May 2019