thevanx@gmail.com sent you this article.
Comment:
| THE INTELLIGENT INVESTOR | When Past Performance Doesn't Even Predict Past Performance Wall Street likes to tout returns it didn't actually produce. You need to ask some important follow-up questions. This past week, the Dow Jones Industrial Average, born on May 26, 1896, turned 128 years old. Let's pour the Dow a drink from the fountain of youth and see what happens. I'll give you a hint. The lesson here isn't only about markets, but also about marketing—in particular, what's called backtesting, a statistical dirty trick that's central to Wall Street's marketing playbook. And we can use the Dow, which this year has been drastically underperforming its younger cousin, the S& P 500, to help explain how the pros try to pull the wool over your eyes. Consider the way the indexes are constructed. The Dow is priceweighted: The higher a company's share price, the more it contributes. Stocks in the S& P 500, by contrast, are weighted by their total float-adjusted market value: share price multiplied by the number of shares that trade publicly. (The Dow, originally owned by Dow Jones, publisher of The Wall Street Journal, is now owned by S& P Dow Jones Indices, but two Journal editors sit on the index committee.) So Boeing, for instance, is 2.9% of the Dow. That puts it barely behind Apple at 3.3%—because Boeing's $172 share price is close to Apple's $190. That's why Apple doesn't have a huge impact on the Dow. But Apple's adjusted market value, $2.76 trillion, is 28 times greater than Boeing's $98 billion. This makes it far more meaningful to the S& P 500. Add up the share prices of the Dow's 30 stocks, and you get something like $5,850. A new stock at $1,000 a share would constitute 17% of the basket—single-handedly unbalancing the entire benchmark. That's what Howard Silverblatt, senior index analyst at S& P Dow Jones Indices, calls "1896 mathematics." It effectively precludes the Dow from adding such stocks as Eli Lilly (recent price: $815, up 40% year-to-date) or Netflix ($648, +33%). All this got me thinking: How would a fantasy Dow consisting of the 30 highest-priced stocks have performed? Let's call it the Supersized Dow. Out with Intel (share price: $30), Verizon Communications ($40) and Cisco ($46). In with NVR ($7,455), Booking Holdings ($3,755) and 28 other stocks with supersized prices, including Lilly, Netflix and Nvidia. Over the past 10 years, this Supersized Dow would have returned an average of 30.2% annually, versus 11.4% for the real-world Dow and 12.8% for the S& P 500, according to John Jacques of AJOVista, an investment firm based in Boston. A $10,000 investment would have mushroomed to almost $140,000 in the Supersized Dow— but to less than $34,000 in the S& P 500. What's the catch? Why can't you pulverize the market just by buying the stocks with the highest share prices? Because of backtesting. The Supersized Dow would have outperformed only if, at the beginning of the period, you bought the stocks that happened to have the highest share prices at the end of the period— something no one could possibly predict. If, 10 years ago, you had bought the stocks with the highest share prices then, you'd have underperformed the S& P 500 by an average of 1.3 percentage points annually. That assumes you adjusted the portfolio each Jan. 1 so it held the 30 highest-priced stocks as of that date—as would be logical if you tried this in real time. You also had to trade for free, reinvest all dividends and pay no taxes. By backtesting—applying hindsight to past data and pretending that a hypothetical portfolio had been run that way all along—financial marketers can make outperformance look ridiculously easy. Take the five target-date funds run by Brandywine Asset Management of Thornton, Pa. These funds seek greater risk when savers are young, scaling back as investors approach or exceed retirement age. Brandywine's funds use put options, which go up when stocks go down, to try mitigating losses in a market crash. In theory, that can provide an edge. If you'd invested $10,000 in January 2013, you'd have had $23,091 in Brandywine Target Retirement at the end of March 2024—but only $16,064 in its average competitor, according to the firm. The only trouble is, you couldn't have invested in the fund in January 2013, because it didn't exist. That track record is backtested and largely hypothetical. The fund didn't begin operating until October 2023. The past returns show how the fund would have performed if it had existed ever since January 2013, partly based on other accounts Brandy--wine ran at the time. From July 2018 until May 2020 and again in March and April 2023, the firm didn't have actual  results from the strategy, so those periods are simulated, says Brandywine's chief executive, Michael Dever. "If you're managing money, the only thing that matters is accuracy," says Dever. "We think we've come up with a very accurate representation of what we expect to be able to achieve performancewise." Since its launch in October, the fund has outperformed its benchmark by 2 percentage points with less fluctuation, according to Brandywine. "So far it's been doing exactly what the back performance says it should," says Dever. "We'll know for sure in another 10 years." The Brandywine funds are only the tip of the backtesting iceberg. Asset managers often roll out exchange-traded funds based on such factors as "liquidity" or "revenue" that appear to have worked in the past, but flop in the real world. Insurance companies have sold billions of dollars of annuities, universal life and other products linked to customized indexes with high past "returns" that soon fade. Of course, you can't completely ignore the past when you project the future. To protect yourself against backtests that foretell nothing, ask questions like these: If this idea is so great, why weren't these guys using it at the beginning of the period instead of only at the end? Why wasn't everybody using it? How long has the strategy actually been used, and with how much money? Has it been tracked over even longer periods than reported in the backtested data? How many other strategies were backtested but abandoned? Do the past numbers include trading costs? If the answers don't make sense, don't invest. |
| |
| This site provides a digital replica of the print newspaper and is intended for the personal use of our members. For commercial reproduction or distribution of Dow Jones printed content, contact: Dow Jones Reprints & Licensing at (800) 843-0008 or visit djreprints.com. |
Copyright 2017 Dow Jones & Company, Inc. All Rights Reserved.
No comments:
Post a Comment