October 2022
David Kelly on “sticky inflation”
Inflation Illusion
Nothing new here
More unLikelyhoods from Robert Kiyosaki
Crypto hedge? UnLikely.
Eyeing a bottom.
A counterproductive endorsement.
Lesser Known Statistics: Inferential Statistics
When most people hear “statistics” they think of descriptive statistics, which is the branch that seeks to quantitatively describe data in a meaningful way - things like averages, spread, and other summary calculations that describe a data set. There is no error in these calculations, the average is determined precisely without error, it is what it is. Commentators who write pieces like How Statistics Can Help You Beat The Market are employing descriptive statistics.
This is not the only branch of statistics however. Inferential statistics emerges when we need to generalize from limited data sets to broader populations. Think of sampling voters to predict the outcome of an upcoming election - asking every voter for their preference is expensive, not to mention that is the whole point of the actual election. So instead we survey a small subset of voters, then extrapolate those findings to the entire population while trying to control for sampling error. There most definitely is error when extrapolating beyond the limited data set one collected, error that statisticians understand but many others do not. This error has led people to irrationally lose confidence in polling, a disillusionment that saw a resurgence after the 2016 presidential election (Why 2016 election polls missed their mark | Pew Research Center).
There are many flavors of statistical errors.
One of the most common statistical errors is extrapolating from a too-small interval of time. You know, like when hedge fund managers and their apologists promote the industry’s outperformance in the first two months of the year: Hedge funds significantly outperform broad markets through a bad start to 2022. Of course long term studies show the perils of extrapolating from short term non-representative samples of performance, because inevitably this short-term “luck” runs out. Here’s CNBC’s Noah Sheidlower illuminating this point, Despite success this year, underperformance rates are 'abysmal' for large-cap active managers for the long run:
S&P Global recently published its Mid-Year 2022 SPIVA U.S. Scorecard, which measures how well U.S. actively managed funds perform against certain benchmarks. The study found that 51% of large-cap domestic equity funds performed worse than the S&P 500 in the first half of 2022, on track for its best rate in 13 years — down from an 85% underperformance rate last year.
…
Despite the promising numbers, long-term underperformance remains, as Pisani noted, “abysmal.” After five years, the percentage of large caps underperforming benchmarks is 84%, and this grows to 90% and 95% after 10 and 20 years respectively.
Although presenting complex models and sophisticated strategies makes managers sound smart, the long term performance records tell a different story. The issue is that the data in limited samples are noisy and may mask the underlying effect. Inferential Statistics helps answer questions like “How plausible is the claim that active managers do in fact outperform passive indexing, given data showing their underperformance?” Anything can happen in the noisy short term, but in the long term inferential statistics helps illuminate the signal more and more clearly.
Hypothesis testing helps determine whether a particular observation is Likely to be from a true effect, or whether the observation can be explained by random variation alone. It is why teachers will give the benefit of the doubt to potentially lying students the first time, but are not so forgiving with future suspected lies: What is Hypothesis Testing? Why Don’t Teachers Believe the Liar Students?
The classic statistical example of hypothesis testing is to determine whether a particular coin is fair. We start with a “null hypothesis” that the coin is fair, then look for evidence against this hypothesis of fairness. If we flip the coin 10 times and get 5 heads with 5 tails, there is no evidence to support the claim that the coin is biased, so we are okay with the null hypothesis. But is 6 heads and 4 tails evidence that the coin is biased, or could that outcome happen merely by random variation of a fair coin? At a reasonable acceptable error level of 5%, 6 heads and 4 tails does not provide sufficient evidence that the coin is biased, because truly fair coins will produce results as or more extreme as 6 heads and 4 tails more often than the 5% error level. In fact, the Likelyhood that a fair coin will produce the perfectly fair 5 heads and 5 tails is only 24.61%, meaning that a statistically uninformed observer will see an outcome other than 5 heads and 5 tails the other 75.39% of the time and suspect incorrectly that the fair coin is actually biased.
But if that same coin was flipped 1,000 times and produced 600 heads and 400 tails - the same 60% / 40% outcome as when we flipped the coin only 10 times - an outcome that extreme is very unlikely from a truly fair coin. To be precise, if we flip that fair coin 1,000 times, we will get an outcome at least that extreme (either 600 or more heads or 600 or more tails) just 0.0000000002722% of the time. So we reject the null hypothesis of a fair coin and conclude that an outcome of at least 600 heads or tails is strong evidence of a biased coin.
There are variations on hypothesis tests, including one-sided tests versus two-sided tests, non-binary outcomes, and others. But we can readily extend these findings to active management in investing. The default null hypothesis is that traditional active management does not outperform the S&P 500 Index net-of-fees, and the alternative hypothesis is that traditional active management does outperform the S&P 500 Index net-of-fees. The observation is the long history of reports like SPIVA showing active management’s underperformance over increasingly long time horizons, which leads us to the same conclusion as with the coin flipping: that such an underperformance is very unlikely from active managers who do in fact outperform the index, and therefore the active management “coin” is very Likely biased against investors.
Lesser Known Statistics: Subjectivity
Nassim Taleb’s Fooled By Randomness, Michael Mauboussin’s The Success Equation, and Nate Silver’s The Signal and the Noise are mathematical must-reads for any investment manager. While grounded in solid mathematics, each book avoids complex formulas and calculations - there’s simply no need. A conceptual understanding of statistics goes a long way in making sense of fundamental principles of investing.
Here’s a quote attributed to Warren Buffett:
"We don't use complicated valuation models because we want investments that are so obvious as to not need one."
Buffett elaborates in the 2007 Berkshire Hathaway annual meeting:
“If they close down the stock market for a couple of years, if interest rates go up another 100 basis points, or 200 basis points, we’re still happy with what we bought. And above that, I really, I know it sounds kinda fuzzy, but it is fuzzy.”
This is no surprise, Buffett is an extremely successful buy and hold value investor. Given the long history of U.S. stock indices appreciating in value, the logic behind the seismic reallocation out of active management into passive indexing is clear. As Buffett also says, “Never bet against America”.
It is true that some investors have profitably employed complex mathematical trading systems over long time horizons, usually to shorten the time frames of their tradable positions. Jim Simons has been called the “Quant King”, “The Man Who Solved the Market”, and “the most successful hedge fund manager of all time” for good reason - the Medallion fund managed by his quantitative hedge fund Renaissance Technologies achieved a 66.1% average gross annual return (39.1% average net annual return) between 1988 - 2018 (U.S. News). Notably, Renaissance Technologies hires mathematicians, statisticians, physicists, and other academics without any financial industry experience at all. Their trading system has been characterized as predicting short term technical price movements without conducting much or any long term fundamental analysis, so it is not necessary for analysts to have traditional financial industry certifications or qualifications.
As investors lengthen their time horizons, however, fundamental analysis becomes more salient than technical analysis in effective investment management - and also more subjective.
ONE MORE THING…
David Kelly on “sticky inflation”. Why the Fed should worry less about sticky inflation (but probably won't)
Inflation Illusion. If you're selling stocks because the Fed is hiking interest rates, you may be suffering from 'inflation illusion'
Nothing new here. All-or-Nothing Stock Churn Exacts Harsh Toll on Would-Be Market Timers.
More unLikelyhoods from Robert Kiyosaki. Set your revisit reminders for January. Robert Kiyosaki Predicts US Dollar Will Crash by January — Suggests Buying Bitcoin
Crypto hedge? UnLikely. Gold and crypto have been called 'inflation-proof' investments—so far in 2022, neither seems to be a great hedge
Eyeing a bottom. The conditions needed for a stock market bottom are forming as investors get overly bearish, JPMorgan says
A counterproductive endorsement. Not only Be wary of investing advice from celebrities like Kim Kardashian, but Ethereum Max Price Analysis: EMAX price takes a huge hit after Kim Kardashian joins the rat race to promote it.
The information and opinions contained in this newsletter are for background and informational/educational purposes only. The information herein is not personalized investment advice nor an investment recommendation on the part of Likely Capital Management, LLC (“Likely Capital”). No portion of the commentary included herein is to be construed as an offer or a solicitation to effect any transaction in securities. No representation, warranty, or undertaking, express or implied, is given as to the accuracy or completeness of the information or opinions contained herein, and no liability is accepted as to the accuracy or completeness of any such information or opinions.
Past performance is not indicative of future performance. There can be no assurance that any investment described herein will replicate its past performance or achieve its current objectives.
Copyright in this newsletter is owned by Likely Capital unless otherwise indicated. The unauthorized use of any material herein may violate numerous statutes, regulations and laws, including, but not limited to, copyright or trademark laws.
Any third-party web sites (“Linked Sites”) or services linked to by this newsletter are not under our control, and therefore we take no responsibility for the Linked Site’s content. The inclusion of any Linked Site does not imply endorsement by Likely Capital of the Linked Site. Use of any such Linked Site is at the user’s own risk.