War and Peace: History Rhymes, But Which History?

It is said that history doesn’t repeat itself, but rhymes, but investors need to be careful about what history they study.
 

As I prepare for retirement at the end of March, I want to impress upon readers the importance of looking under the hood of past quantitative history studies in order to understand underlying assumptions.
Consider, for example, this timely study of U.S. equity returns after geopolitical and economic shocks. The accompanying table from Ryan Detrick of Carson Investment Research would lead to the conclusion that investors should ignore shocks and buy the dip, as the stock prices tend to shrug off short-term setbacks and rise over time.

A different table from Jeffrey Hirsch at Almanac Trader tells a slightly different story. Hirsch excluded some of the more minor shocks in his event study such as the Asian Financial Crisis and Brexit. Average and median returns are directionally similar inasmuch as stock prices tend to react to the initial shock and then rise afterwards, but the magnitude of the returns is dissimilar to the Detrick study. In addition, post-shock returns improve significantly if investors focus on the post Iran Hostage Crisis period. The worst of the initial short-term price shocks were attributed to World War II and the early days of the Cold War.

In short, how you choose your data sample will affect your return expectations.

The full post can be found here.

How the Investing Game is Changing

Preface: Explaining our market timing models 

We maintain several market timing models, each with differing time horizons. The "Ultimate Market Timing Model" is a long-term market timing model based on the research outlined in our post, Building the ultimate market timing model. This model tends to generate only a handful of signals each decade.

The Trend Asset Allocation Model is an asset allocation model that applies trend-following principles based on the inputs of global stock and commodity prices. This model has a shorter time horizon and tends to turn over about 4-6 times a year. The performance and full details of a model portfolio based on the out-of-sample signals of the Trend Model can be found here.

  

My inner trader uses a trading model, which is a blend of price momentum (is the Trend Model becoming more bullish, or bearish?) and overbought/oversold extremes (don't buy if the trend is overbought, and vice versa). Subscribers receive real-time alerts of model changes, and a hypothetical trading record of the email alerts is updated weekly here. The hypothetical trading record of the trading model of the real-time alerts that began in March 2016 is shown below. 

 

The latest signals of each model are as follows:

• Ultimate market timing model: Buy equities (Last changed from “sell” on 28-Jul-2023)*
• Trend Model signal: Bullish (Last changed from “bearish” on 27-Jun-2025)*
• Trading model: Neutral (Last changed from “bullish” on 26-Nov-2025)*

* The performance chart and model readings have been delayed by a week out of respect to our paying subscribers.

Update schedule: I generally update model readings on my site on weekends. I am also on X/Twitter at @humblestudent and on BlueSky at @humblestudent.bsky.social. Subscribers receive real-time alerts of trading model changes, and a hypothetical trading record of those email alerts is shown here.

Subscribers can access the latest signal in real time here.

A Lifetime Paradigm Shift

The year is almost over, and it's time to reflect on the tumultous time investors have experienced. In particular, Trump's trade war has caused an unexpected response and volatility. The markets were initially rattled by his "Liberation Day" announcements. Calm set in once it became apparent that major trading partners didn't retaliate, except for China and stocks turned risk-on and bond yields fell. 

 

The trade war was only the beginning. When I announced in March that I was shutting down, I didn't expect the financial markets were going to experience a paradigm shift of a lifetime. The White House release of the National Security Strategy (NSS) is just another manifestation of the paradigm shift that not only affects U.S. foreign policy, but basic assumptions about investing that I am not sure I know how to analyze anymore. 

 

The investing game is changing. It's time for me to leave.

The full post can be found here.

An insightful interview with Scott Bessent

RenMac hosted an interview with Scott Bessent, who is Trump's announced nominee for Treasury Secretary, in early 2021. While Bessent did not talk about policy or politics, I found it highly insightful as he described his career path and his investment process.

 

The interview is useful to listen to in its entirety, but here are some highlights.

 

The full post can be found here.

Audit your trading the way pros do it

Traders are always interested in improving their techniques. Today, I would like to offer a framework for thinking about your trading, using the way fund sponsors evaluate investment managers, called the 5 Ps.

• People: Who are you, and what's your experience and training?
• Performance: How have the returns been, and what kind of risk did you take to achieve those results?
• Philosophy: What makes you think you have an edge?
• Process: How do you implement the edge you have on the market?
• Portfolio: Does your portfolio reflect what you are saying about philosophy and process?

With the preface that there are never any single right answer in investing and trading, I will focus on "philosophy" and "process".

The full post can be found here.

A simple decision vs. a decision process

I got some pushback from a reader to my weekend post (see How to spot the bear market bottom) about the FT Alphaville article indicating that former Secretary of Defense Mattis raised concerns about how the White House lacked a decision making process. The reader went on to defend Trump's decisions.

I try very hard to remain apolitical on this site. Everyone is entitled to their own opinion, but there is a distinction between a decision, and a process. Here is an example from the investment realm. Josh Brown recently ranted about people "who called the correction". Click this link if the video is not visible.

Josh Brown's main complaints can be summarized as:

• Anyone can make a market call. If you are wrong, very few people will remember, or you can delete your articles or tweets.
• Managing a portfolio is a much tougher task. Portfolio managers are measured by actual returns. As an example, if you decide to sell out, what is your discipline for buying back in?
• Just because someone doesn't say anything, it doesn't mean that they are unprepared for market volatility. Most firms have compliance guidelines about what individual portfolio managers or advisors can or cannot say or publish. 

Despite my own efforts at transparency (see A 2018 report card) where I have published my track record, and owned up to bad calls, I sympathize with Brown. Josh Brown's rant amounts to distinguishing a decision (market call) to an investment process. A timely market call means little if there is no investment process behind it.

The full post can be found at our new site here.




A Special Announcement
We told you so. We told you the market was going down.

Here is the track of Humble Student of the Markets, where we are neither perma-bulls nor perma-bears. Most recently, we have been correctly bullish since the correction of 2015, and turned cautious in August 2018 (see Market top ahead? My inner investor turns cautious, August 5, 2018).

We were also timely at the 2009 bottom. We issued a call to buy beaten up low-priced stocks with high insider buying a week before the ultimate bottom (see Phoenix rising? February 24, 2009).

The out-of-sample record of our model trading portfolio in 2018 was up 42.9%. For more details, see our weekly updates here.

The recent market volatility has brought a flood of new subscribers, and we are announcing a price increase, and a number of other changes in order to better control the growth of our community. However, all subscribers will be grandfathered at their old prices.

The following changes will occur as of March 1, 2019:

• The annual subscription price will rise from US$249.99 to US$356 per year.
• The monthly subscription price will rise from US$24.99 to US$35.60 per month.
• The 24-hour subscription will no longer be offered.
• The embargo period for free content will change from two weeks to four weeks.

Remember, if you subscribe now, you will be grandfathered at the old price - permanently.

A 2018 report card

The year is nearly over, and it is time to issue a report card for my investor and trading models. Overall, both had good years, except for the trading blemish at year-end.

My inner investor could not have asked for much more. He was correctly bullish during the run-up from early 2016, and turned cautious at the January top. He turned bullish again as the market corrected in February, and became cautious again in August (see A major top ahead? My inner investor turns cautious).

The cautiousness turned into bearishness in early December when my Ultimate Market Timing Model flashed a sell signal after a 10% drawdown (see A bear market is now underway). As a reminder, the Ultimate Market Timing Model is a very slow turnover model that changes its views only every few years. It was designed for by investors with long term horizons, with the intention of avoiding the worse of equity drawdowns associated with major bear markets:

I am indebted to the blogger at Philosophical Economics who suggested a macro overlay to trend following systems (see Building the ultimate market timing model). Major bear markets generally occur under recessionary conditions. Why not ignore moving average signals until your macro model is forecasting a recession?

This “Ultimate Market Timing Model” is ultimately beneficial for long-term investors. If you could cut off the left tail of the return distribution and avoid the really ugly losses, you could run a slightly more aggressive asset mix and receive a higher expected return with lower risk. For example, if the standard risk-return analysis dictates a 60% stock and 40% bond asset mix, you could change it to a 70/30 mix with this model, and get downside risk similar to the 60/40 portfolio. To be sure, this system isn’t perfect, and anyone using such a model will have to incur “normal” equity risk, and it would not have kept you out of the market in the 1987 Crash.

Little did I expect the market to fall so dramatically after that sell signal, but I can't ask for much more in an asset allocation model, either on an intermediate or long term perspective.

The full post can be found at our new site here.

Pax Americana, or America First?

December is the season for investment advisors and portfolio managers to meet with their clients. Here are some thought on your an allocation framework as you prepare for those meetings. As a cautionary message, let's begin with a "buy and forget" portfolio featured in Fortune in 2000 and how they performed by 2012.

Haha. Experienced portfolio managers and advisors don't make those kinds of mistakes. We all know that diversification is the only free lunch in investing.

For the simple answer on personal investing, I refer readers to a Business Insider article by Chelsea Brennan, "I spent 7 years working in finance and managed a $1.3 billion portfolio — here are the 5 best pieces of investing advice I can give you":

1. Understand your goals
2. Index fund investing is the easiest way to win
3. Be in it for the long term
4. Don't assume you have it all figured out
5. Be prepared for anything

Unfortunately, I see American investors making the diversification mistake, as well as mistakes 4 and 5 again and again. Much of what passes for financial planning in the US is based on the mistaken assumption of a backtest that has severe survivorship problems. This will becoming increasingly evident as American policy changes from the era of Pax Americana to America First.

The full post can be found at our new site here.

Peak smart beta?

A recent comment by Michael Mauboussin of Credit Suisse that nailed the dilemma of active managers, namely that using traditional approaches to alpha generation is akin to mining lower and lower grade ore:

Exhibit 1 shows that the standard deviation of excess returns has trended lower for U.S. large capitalization mutual funds over the past five decades. The exhibit shows the five-year, rolling standard deviation of excess returns for all funds that existed at that time. This also fits with the story of declining variance in skill along with steady variance in luck. These analyses introduce the possibility that the aggregate amount of available alpha—a measure of risk-adjusted excess returns—has been shrinking over time as investors have become more skillful. Investing is a zero- sum game in the sense that one investor’s outperformance of a benchmark must match another investor’s underperformance. Add in the fact that in aggregate investors earn a rate of return less than that of the market as a consequence of fees, and the challenge for active managers becomes clear.

I got into quantitative investing back in the 1980`s when ideas and models were fresh and plentiful. Today, factor investing has become increasingly mainstream, and so-called "smart beta" may have exceeded their best before date.

The full post can be found at our new site here.

A passive index fund built to outperform?

A long time reader sent me this Seeking Alpha article entitled "Monish Pabrai Has Created An Index Fund Built To Outperform", which described a "passive index fund" built using the following three investment themes deployed in three portfolio buckets:

• Share buybacks: Companies that are buying back their own shares
• Selected value manager holdings: The holdings of 22 selected value managers, based on their 13F filings
• Spin-offs: Companies that were recently spun off from their parent

It's difficult to have a detailed opinion on the pros and cons of this fund. That's because the article only described what this "index fund" would hold, it did not describe the portfolio construction method, or how much of each stock it would hold. So it`s impossible to understand the risk profile of the fund, the size of its factor exposures, as well as its sector and industry exposures.

All the marketing hype aside, this investing approach is really a re-packaged form of factor investing, otherwise known as "smart beta". Therefore investors who buy into such a vehicle should expect similar kinds of results as "smart beta", though in a multi-factor format.

The full post can be found at our new site here.

The ways your trading model could leading you astray

I have had a number of discussions with subscribers asking for more "how to" posts (see Teaching my readers how to fish). This will be one of a series of occasional posts on how to build a robust investment process.

For traders and investors, one of the challenges is how to build a robust discipline that works well through different market regimes. As a case study, consider this study from Simple Stock Model that generates signals based on the cash flows in and out of the SPY ETF as a sentiment signal. The trading rule is: "If the 4-week average of the 3-month change in SPY's percentage of shares outstanding is greater than +5%, be out of the market."

The chart below shows the equity curve from this trading system (white line = buy and hold, blue line = trading system). The results look pretty good, especially for a relatively low turnover model. (Incidentally, it's on a sell signal right now).

SPY shares outstanding trading system

Not so fast! Don't jump to conclusions before digging into the data and reading the fine print.

The full post can be found at our new site here.

Thanks, but I'm not that good!

It's always nice to get positive feedback from subscribers. One subscriber praised me for my trading model and wanted real-time updates of signal changes (which I already provide but wound up in his spam folder).

Another subscriber complimented me on my series of tweets indicating an oversold market on Thursday, which suggested that the market was poised to rally should the Jobs Report on Friday morning was benign (click links to see tweet 1, 2, 3, 4, and 5).

Thanks, but I'm not that good.


Teaching my readers how to fish
Humble Student of the Markets is not intended to be a trading service. I addressed this issue in my post Teaching my readers how to fish.

Think of a building a boat as like building a portfolio. The portfolio management process consists of the following steps:

1. Deciding on what to buy and sell;
2. Deciding on how much to buy and sell; and
3. Deciding on how to execute the trade.

While we discuss step 1 endlessly in these pages and elsewhere, the other steps are equally important. Step 2 is also a reason why what I write in these pages is not investment advice, namely I know nothing about you:

• I know nothing about your cash flow, or spending needs;
• I know nothing about your return objectives;
• I know nothing about how much risk you are willing to take, or your pain threshold;
• I know nothing about your tax situation, or even what tax jurisdictions you live in;
• And so on…

If I know nothing about any of those things, how could I possibly know if anything I write is appropriate for you? I was asked recently why I don’t post my portfolios and their performance. While posting my trades represent a disclosure of any possible conflicts in my writing, my own portfolios are a function of my own cash flow needs, my return objectives, my own pain thresholds, etc. How could any portfolio that I post be appropriate to anyone else?


A terrific call, or terrible call?
Consider the following example. On February 24, 2009, a week before the ultimate market bottom, I made a call to buy a high-beta portfolio of low-priced stocks, which I termed a Phoenix portfolio (click link for post):

• Stock price between $1 and $5 (low-priced stocks)
• Down at least 80% from a year ago (beaten up)
• Market cap of $100 million or more (were once "real" companies)
• Net insider buying in the last six months (some downside protection from insider activity)

Was that a terrific call, or a terrible call? You be the judge.

At one level, the call to buy a high-beta portfolio a week before a possible generational bottom for stocks could be a career making call. On the other hand, the market fell -11.9% based on closing prices before the final bottom was reached.

For investors, the Phoenix portfolio was well-timed and it went on to roughly triple its value in about a year. For short-term traders, the 11.9% drawdown was a disaster.

This brings me to my point. Don't blindly follow what I do. My return objectives are not the same as yours. My pain threshold will be different from yours, which affects the placement of stop loss orders.

Your mileage will vary. I can only teach you how to fish, not fish for you.

A copy of this post can also be found at our new site here.

If machines are human, would you let one marry your daughter?*

Several months ago, the internet was all abuzz over the victory of Google's AlphaGo program beating Go grandmaster Lee Sedol (see story here). As the game of Go is a computationally and mathematically complicated game and the number of variations in the game is an order of magnitude higher than chess, it was a great victory for the kinds of "deep learning" artificial intelligence (AI) techniques pioneered by Google's Deep Mind team.

Indeed, there have been great strides by AI research teams in the fields of pattern recognition and natural language processing. As an example, the Washington Post chronicled a startup called Viv designed to be a natural language AI bot that can order you pizza, among other tasks:

In an ordinary conference room in this city of start-ups, a group of engineers sat down to order pizza in an entirely new way.

“Get me a pizza from Pizz’a Chicago near my office,” one of the engineers said into his smartphone. It was their first real test of Viv, the artificial-intelligence technology that the team had been quietly building for more than a year. Everyone was a little nervous. Then, a text from Viv piped up: "Would you like toppings with that?"

The engineers, eight in all, started jumping in: “Pepperoni.” “Half cheese.” “Caesar salad.” Emboldened by the result, they peppered Viv with more commands: Add more toppings. Remove toppings. Change medium size to large.

About 40 minutes later — and after a few hiccups when Viv confused the office address — a Pizz’a Chicago driver showed up with four made-to-order pizzas.

The engineers erupted in cheers as the pizzas arrived. They had ordered pizza, from start to finish, without placing a single phone call and without doing a Google search — without any typing at all, actually. Moreover, they did it without downloading an app from Domino’s or Grubhub.

The full post can be found at our new site here.

* The title was inspired by an old science fiction short story entitled "If all men were brothers, would you let one marry your sister?"

Teaching my readers how to fish

In the past week, I had discussions with several different people about the operating philosophy of Humble Student of the Markets, The objective of the website can be summarized by a variation of an old adage:

Give a man a fish, he'll eat for a day.
Teach a man how to fish...he'll want to get a boat.

I don`t want to just give my readers a fish for the day, I would rather help them build their own boat.


Why my boat is different from yours
Think of a building a boat as like building a portfolio. The portfolio management process consists of the following steps:

1. Deciding on what to buy and sell;
2. Deciding on how much to buy and sell; and
3. Deciding on how to execute the trade.

While we discuss step 1 endlessly in these pages and elsewhere, the other steps are equally important. Step 2 is also a reason why what I write in these pages is not investment advice, namely I know nothing about you:

• I know nothing about your cash flow, or spending needs;
• I know nothing about your return objectives;
• I know nothing about how much risk you are willing to take, or your pain threshold;
• I know nothing about your tax situation, or even what tax jurisdictions you live in; 
• And so on...

If I know nothing about any of those things, how could I possibly know if anything I write is appropriate for you? I was asked recently why I don't post my portfolios and their performance. While posting my trades represent a disclosure of any possible conflicts in my writing, my own portfolios are a function of my own cash flow needs, my return objectives, my own pain thresholds, etc. How could any portfolio that I post be appropriate to anyone else? Your mileage will vary.

Don't look for a fish

Here is an example of what I am talking about. I had been recently bullish on stocks and both my investment account (inner investor) and trading account (inner trader) got long. My trading account sold and got stopped out of its long position as a result of my risk control discipline, which is a function my risk profile and pain threshold. Subsequent market action indicates that my inner trader got faked out and the market rallied. In that case, it appears that my inner trader was wrong by getting stopped out of his position, while my inner investor was right.

This incident also illustrates the point of the do's and don'ts of reading the content on this website. Anyone blindly following my trades is in effect looking for a fish. But there is no fish. The markets are not easy. You have to build a boat that's right for you.

I have two boats (used by my inner investor and inner trader). Taking a ride on either of mine by blindly following my trades means adopting my investment objectives and risk profile, which you know nothing about.


My two boats

The chart below shows an example of how my inner trader thinks about the stock market. He isn't always right, but he has been more right than wrong. His portfolio turnover averages 200% per month, which is not appropriate for everyone.

By contrast, here is an example of how my inner investor thinks about the market. The time horizon is longer. Turnover is much lower, but drawdown risk and pain threshold is higher.

For full details, see:

• Why I am bearish (and what would change my mind) May 2015
• Relax, have a glass of wine August 2015
• Why this is not the start of a bear market September 2015
• The reason why the bulls should be cautious about a January hangover December 2015
• Why this is a correction and not a bear market January 2016

Neither of those boats may be right for you. The purpose of Humble Student of the Markets is not to give anyone detailed trading advice, including the specific timing of trades. I can only make suggestions, but you have to decide if those suggestions are right for you.

I am not here to give you a fish. I am here to teach you how to fish and help you build your own boat. That way, you can eat for a lifetime.

The price of financial success

Last year, my Valentine's Day post was about how Facebook analyzed the behavior of who fall in love (see Falling in love, the Facebook version). This year, I thought that I would highlight the price of success for hedge fund managers.

I highlight a research paper entitled Limited Attention, Marital Events, and Hedge Funds. Here is the abstract (emphasis added):

We explore the impact of limited attention on investment performance by analyzing the returns of hedge fund managers who are distracted by personal events such as marriage and divorce. We find that marriages and divorces are associated with significantly lower fund alpha, during the six-month period surrounding the event and for up to two years after the event. Relative to the pre-event window, fund alpha falls by an annualized 8.50 percent during a marriage and 7.39 percent during a divorce. Busy fund managers who manage larger funds and engage in high tempo investment strategies are more affected by marriage. Fund managers who depend on interpersonal relationships in their investment strategies are more affected by divorce. We show that behavioral biases may partially explain the connection between inattention and performance deterioration. The difference between the proportion of gains realized and the proportion of losses realized widens during a marriage and a divorce, indicating that inattentive hedge fund managers are more prone to the disposition effect. Taken together, our findings suggest that limited investor attention can hurt the investment performance of professional money managers.

To be successful, hedge fund managers should consider staying away from personal relationships, any of them.

Ah, the price of success!


Site notice
We have temporarily stopped taking monthly and annual subscriptions in order to better control the rapid growth of our community. However, if you are interested, you can either subscribe

• To a 24-hour day pass; or
• Subscribe for notification of free posts, which are open and available to the public two weeks after publication.

How patient an investor are you?

I received a lot of feedback to my post, How Valeant revealed the dirty little secret of fund management. Some of it was off the mark, as the post was not intended to be a discussion on the merits of Valeant (VRX) as an investment. Instead, it was an illustration of how impatient investors are forcing managers to closet index in order to manage their own business risk.

In the post, I cited an example that showed the simple act of a single wrong decision on one stock has the potential to sink an entire investment management practice - and we had't even begun to discuss the myriad of other ways that risk can rear its ugly head.

It all boils down to the question of how patient investors are with their managers and how much rope they are willing to give their managers to succeed, or fail.


The Value Investor example
Consider, for example, a style of investing that is known to reward patient money - value investing. At Euclidean Technologies, John Alberg and Michael Seckler demonstrated how difficult the value style can be. First, they did a backtest of a simple value discipline:

In this analysis, the value strategy is simply to buy companies that are most inexpensively priced in relation to their prior year’s EBIT (earnings before interest and tax). The top chart plots the simulated performance of this simple approach to value investing in context of the SP500’s total return. The bottom chart shows how much and for how long the value strategy fell behind the SP500 at each point in time.

Across this particular simulation, over the period January 1973 to June of 2014, the value approach achieved a compound annualized return of 17.2% while the SP500’s total return (price change plus dividends) did 10.3%. Sounds amazing!

Here's the catch:

However, as the bottom chart shows, an investor using this value strategy would have had to endure 14-years (1988 through 2001) where he wouldn’t have received much feedback that he was on the right path. Soon after recovering from falling behind the market by 30% from 1988-1991, he would have lagged by nearly 50% across a grueling 6 years.

Managing career risk
How many people can endure 14 years of poor performance, whether it's their own money or with client money?

We can all sit around and nod sagely that investment discipline is important, but what actually happens in the real world? Ben Carlson recently wrote about the difference between academic investment research and real world research:

When new investors are just starting out in the markets they’re often told that a paper portfolio is a good way to test out a strategy without putting real money to work. This one sounds good in theory but is fairly useless in practice.

The thing is that there are no simulations that can prepare you for the emotions you feel when investing actual money in the markets. The feelings you get from making or losing money can’t be simulated. The same is true of those who try to turn research into an investable strategy.

He highlighted a quote from an interview with Cliff Asness of AQR (emphasis added):

Well the single biggest difference between the real world and academia is — this sounds overly scientific — time dilation. I’ll explain what I mean. This is not relativistic time dilation as the only time I move at speeds near light is when there is pizza involved. But to borrow the term, your sense of time does change when you are running real money. Suppose you look at a cumulative return of a strategy with a Sharpe ration of 0.7 and see a three year period with poor performance. It does not phase you one drop. You go: “Oh, look, that happened in 1973, but it came back by 1976, and that’s what a 0.7 Sharpe ratio does.” But living through those periods takes — subjectively, and in wear and tear on your internal organs — many times the actual time it really lasts. If you have a three year period where something doesn’t work, it ages you a decade. You face an immense pressure to change your models, you have bosses and clients who lose faith, and I cannot explain the amount of discipline you need.

Managers face incredible career risk under those circumstances. Even Warren Buffett isn't perfect. CNN Money recently ran a story entitled Warren Buffett's top stocks are dogs this year, which pointed out that Berkshire top holdings (IBM, WFC, USB, GS, KO, AXP, PG, WMT) have all performed poorly this year. Now imagine that your manager is Brand X instead of a well-known name like Warren Buffett, and your Brand X manager followed a Buffet-like discipline. How much patience would you have with that manager after a bad year?

The following example should not be interpreted my "I told you so" victory lap, as I have been badly wrong in my investment career. When I factor in the sheer volume of hate mail that I received from reader when I was bearish on stocks early this year and when I turned more constructive on equities after the August sell-off, many investors are not very patient at all:

Why I am bearish (and what would change my mind) May 2015 (red arrow below)
Relax, have a glass of wine August 2015 (blue arrow below)
Why this is not the start of a bear market September 2015 (purple arrow below)

Key lessons
There are two key takeaways here. For investors, your lack of patience, both at an individual and institutional level, is forcing investment managers to take steps to control their own business risk and become sensitive to benchmark tracking error. If you want to know who to blame for closet indexing, look in the mirror.

Managers also have to recognize that, like it or not, client time horizons are very short. Even for institutions, the typical grace period for poor performance is no more than 2-3 years. In that case, you need to be aware of business risk when sizing your bets in a portfolio.

Notwithstanding the possible solvency problems that come with bad performance should AUM plummet, the way a manager controls business risk also affects the culture of the organization. For example, sales and marketing staff tend not to have the same level of conviction as the investment staff on the investment philosophy. Managed improperly, they can become "order takers" during good times, which is an easy job that practically anyone can do, but abandon the organization during bad times.

Is that the kind of organization that you want?

How Valeant revealed the dirty little secret of fund management

How would you feel if your equity fund manager lagged the index by 5% in a year? Supposing that you hired the manager, or bought his fund, as part of the diversified equity portion of your portfolio and he missed by 5% in a year. Let's say that you be patient, then he underperforms by another 5% in the next six months. Would you fire someone who lags the market by 10% in 18 months?


Returns vs. business risk
I have met people who say that they love managers who hit the home run and loathe benchmark huggers, but investors get very nervous when their managers lag the market by as little as 5-10% in a relatively short (1-3 years) time frame. From the perspective of the manager, this level of risk tolerance brings up the issue of business risk. How do you maximize performance using your process, or "secret sauce", without taking on excessive business risk that sinks the entire firm?

Risk comes in all shapes and sizes. For an equity portfolio, common sources of risk are sector and industry risk, market cap risk and stock specific risk. To illustrate my example, consider the lowly issue of stock specific risk, which should be diversifiable (at least according to theory).

The recent case of price volatility in Valeant Pharmaceuticals (VRX) is an instructive example in risk control. In the past few weeks, I have spoken to a number of Canadian portfolio managers whose performance was blindsided by specific risk from that one single stock, The outlook for VRX is highly divisive and talking about it is the equivalent of bringing up touchy topics like religion or gun control in polite company. (Incidentally, I have no opinion on the stock.) The chart below depicts the price of VRX in the top panel and the TSX-VRX ratio in the bottom panel, which shows what would have happened to relative performance had a manager had no position in VRX for the past few years.

Despite the fact that the stock got hit today, VRX has shown remarkable returns in the past few years. In the space of about 5 years, the stock has become a 10-bagger and anyone who didn't own it would have underperformed the index (bottom panel). Consider the following effects for a manager who did not hold VRX:

• If he didn't own it at the end of 2013, relative return shortfall would be over 8% when VRX hit its peak this year. For some investment organizations, that kind of shortfall could be a near-death experience.
• Even with the pullback, relative performance shortfall would only be back to 2013 levels and he have not made up for the shortfalls in the previous years.

The art of business risk management

The analysis brings up a key question for the business risk for investment management operations. Yes, we would all like to have the courage to bet our investment convictions, but how much business risk is the practice willing to take? Supposing that an operation were to lose half its clients because of a single decision on a stock, what does that do to the bottom line? Revenues would go down by about 50%, but there are fixed costs such as rent, salaries, systems, legals, etc. Profitability would plunge in such an instance, is the investment management business willing to take that kind of risk?

If not, then there are a couple of steps a manager can do. First, he has to decide the appropriate level of stock specific risk he is willing to take against the benchmark. Supposing a stock has a 3.5% weight in the benchmark, would you hold a 0% if you ranked it a "sell" (-3.5% bet), a non-zero weight, such as 2.5% (+/- 1% vs, benchmark) or 1.5% (+/- 2% vs. benchmark)? On the other hand, if your ranked it a "buy", would a 5.5% weight (+/- 2% vs. benchmark) be appropriate? What about 8.5% (+/- 5% vs. benchmark)?

Another way of approaching the problem would be to try and determine the median competitor weight in the stock (with techniques that I have written about before). Then set benchmark weight to be the median competitor weight instead of the index weight.

I show this example as just how a simple decision on a single stock can crash an entire investment management practice. I haven't even gone into all the other ways that risk can rear its ugly head, such as macro factor, sector, size and so on.

Asking too much of managers?

This post also illustrates the dirty little secret of fund management. Investors are asking too much of managers and managers are consequently reacting rationally by closet indexing.

Investors have to ask themselves: How much rope are you willing to give a manager to succeed? Is John Hussman flaming out, or is he a brilliant thinker going through a bad patch? Other well-known managers like Bill Miller and Ken Heebner have had their ups and downs, how patient are you willing to be? If you have a low level of patience, then you are forcing managers to become benchmark huggers because you are not giving them enough room to win.

For managers: Given the realities of the market, how much risk are you willing to take so you don't crash your firm?

Blame the algos for the sell-off!

I've been pondering the reasons for the US stock market's unusual behavior in this latest panic. In particular, the speed and magnitude of the sell-off has been astounding. Consider the puzzle presented by this hourly chart of the SPX over the last two months.

In early July, the market sold off and fear gauges rose. VIX term structure inverted and NYSE TRIN spiked above 2 (blue circles), which triggered two of the three components of my Trifecta Bottom Model (see the chart in Is this the Ashley Madison market panic?). The Trifecta Model has had a 100% record of spotting short-term bottoms since 2010, but it failed in August 2015.

On Friday, August 21, 2015, similar market conditions prevailed. The market had become oversold, according to RSI(21), VIX term structure had inverted, which indicated rising fear, TRIN had risen above 2 and the SPX had tested a key support level at 2020. These were the classic conditions for an oversold market. Instead of rallying, or even pausing at about the 2040 level, the SPX proceeded to crater over 100 points in the next few days.

Why?

One clue came from the elevated levels of NYSE TRIN (circled in red), which had the characteristics of forced, price-insensitive selling.

It is time to blame the algos.


Algos for newbies
An algorithm, or algo, is simply a recipe for doing a certain task. Here are some simple examples of simple algos that would result in forced selling:

• Margin accounts are required to maintain sufficient equity. If the value of the assets in the account falls below a key level, we ask the account holder to come up with more cash to maintain minimum equity levels. If he can't, we take control of the account and sell the assets.
• A firm targets a certain Value-at-Risk (VaR) for its capital commitment. When VaR falls because volatility is low, traders are free to (and possibly encouraged) to lever up their allocation. When VaR rises, traders are required to scale back their positions.

During the Crash of 1987, portfolio insurance algos, which had become extremely popular at the time, exacerbated the downside volatility once the selling had started. For the uninitiated, you can replicate a long stock-long put option position with delta hedging. Unfortunately, delta hedging required that you buy stock when it goes up and sell stock when it goes down. So when the market tanked, the portfolio insurance programs (algos) were forced to sell stock, which pushed prices down, which forced them to sell more stock...

The algo culprits

Today, delta hedging is not a big part of the market, but others have stepped in to take their place to heighten market volatility. First, there were the HFT algos, which created a minor level of havoc in ETF pricing on Monday, August 24, but HFTs cannot take all of the blame. Marko Kolanovic, JP Morgan quant, explained the violent selling in the context of an unwind from three main types of strategies (via Zero Hedge, emphasis added):

Volatility Targeting (VT) Strategies follow fast signals (such as short term realized volatility) and rebalance quickly (e.g. 1-5 days). Selling pressure from these strategies (estimated to be $50-75Bn) peaked last week and is largely out of the way now. Short-term realized volatility increased from ~10% to ~30%, indicating these funds already reduced their exposure by ~70%. If volatility were to increase further (e.g. in-line with the peak realized volatility in 2011 of ~50%), these funds would have to sell progressively smaller amounts (e.g. an additional ~10-15%). The question is when these funds will start buying. Our view is that the re-levering of these funds is not imminent (e.g. not in the next few days). Leverage in these strategies is a function of trailing volatility (e.g. moving average of the VIX or 1M realized volatility), and even if the VIX were to start declining now, trailing realized volatility would stay elevated for the next ~3 weeks.

CTAs – Following our report last week, we have been getting questions about CTA equity exposure and the timeline of CTA flows. Our CTA replication models suggested that CTA equity exposure at the end of July was very high, at approximately 30% (or $80-$100Bn notional). This allocation to equities is also consistent with the performance of CTA indices in August (Bonds and Commodities were roughly flat, while Equities were down 8% - thus, a 30% equity allocation would match the -2.5% CTA observed performance). As the trend following signals (e.g. 1M, 3M, 6M, 12M price returns) started turning negative in August, CTAs started de-levering equities (in early August). As of September 1st, our CTA replicator indicates that the strategies should be ~25% short equities (short ~$70bn). This would indicate a ~$150bn swing (selling) of CTAs’ equity allocation.

However, CTA strategies don’t rebalance as quickly as VT strategies, and it often takes 1-4 weeks to achieve their target exposure. To assess where we are in the process of CTA de-leveraging, we have calculated the daily beta of a broad CTA index (HFRXSDV) to the SP 500 shown in the figure below. Note that the CTA SP 500 beta dropped from record levels at the beginning of August to zero currently, indicating that CTAs may have completed more than 50% of the expected equity selling (as noted above, target equity exposure is negative ~25%). We estimate that CTAs may continue selling equities for another ~2 weeks, and that the flows may total ~$40-$60bn. Once CTAs establish their September 1st target positions (short equities), they will no longer be selling, and risk becomes skewed towards CTAs buying equities. We estimate CTA flows in other asset classes include a reduction of USD exposure (from $40bn to zero), no change in bond positions, and some short covering of Oil and Gold (from short 40bn to short 30bn).

Risk Parity (RP) –We studied this allocation method in our Primer on Systematic Strategies, and argued that it is one of the soundest approaches to managing portfolio risk.

Risk Parity strategies de-lever when asset volatility and correlation increase. In our report last week, we estimated that risk parity outflows from equities may total $50-100bn on account of the increase in market volatility and risky asset correlations. These rebalances have started, but, given their typically slower rebalance frequency (e.g. monthly), are largely incomplete. We believe the bulk of the risk parity flows are yet to come, and this may add selling pressure to equities over the next 1-3 weeks. To illustrate this point, one can look at a sample multi-asset Risk Parity strategy such as the Salient Risk Parity index. The beta of this index to the SP 500 (shown in the figure above) reached highs of 60% in early August, and has dropped to about 45% currently (compared to a beta of 0% during some of the previous episodes of market volatility).

Please note that in our estimate of Risk Parity assets we have included funds that use Risk Parity as a risk management overlay and tactical allocation, and not just the dedicated Risk Parity (Quant) hedge funds. One could perhaps even broaden the definition of Risk Parity funds to include investors that change their strategic allocation based on expected volatility (e.g. such as CalSTRS announcement of plans to reduce equity exposure in the near future, recently reported in the media).

Kolanovic's main conclusion as of September 3, is to expect more selling:

In summary, we estimate that only about half (or slightly more than half) of total technical selling was completed to-date (mostly completed by VT funds, half by CTAs, and a smaller fraction by RPs). We estimate that a further ~$100bn of selling remains to be completed over the next 1-3 weeks. As a result, we expect elevated volatility and downside price risk to persist. In our view, the risk/reward for equity investors remains in favor of waiting, rather than being fully invested until there is more clarity from macro data and central banks.

Risk-parity strategies have been the focus as the cause of the selling. Here is the view from BoAML (via Business Insider):

Bank of America Merrill Lynch equity derivatives strategists, led by Chintan Kotecha, who write that "Risk parity is not the risk, vol[atility] control is." As they note, it's worth differentiating between risk parity investors with a fixed amount of leverage, and those who tie the amount of leverage applied to their portfolios to market volatility. The latter begin life with a target portfolio volatility level and then apply a certain amount of leverage to their portfolio based on their forecast of future portfolio volatility. An easy example used by BofAML is two times leverage applied to a portfolio with a target volatility of 10 percent and expected to have a volatility of 5 percent. Conversely, a portfolio with forecast volatility of 20 percent could achieve a target volatility of 10 percent by deleveraging its portfolio to 0.5 times levered.

BoAML went on to explain that the amount of selling is a function of how the risk-parity strategy is implemented and the level of leverage allowed in the strategy:

If the leverage applied to risk parity is via target volatility, then the change in component allocations due to dynamically adjusting the portfolio’s leverage could potentially lead to a collective significant deleveraging of assets tracking risk parity. The amount of deleveraging will be a function of the fund’s target volatility and maximum leverage allowed. But more significantly, the deleveraging will also be a function of the prevailing volatility prior to the volatility spike and the magnitude of the daily moves within the volatility spike. There are a variety of different target volatility levels and leverage caps that are often applied by risk parity funds. Typically, they tend to target a volatility level between 6 percent to 10 percent with maximum leverage ranging from 1.5x to 3.0x.

Regardless of the target volatility and max leverage limits within a risk parity fund, however, the recent and unusually violent spike in equity volatility from depressed levels ... alongside a relatively muted diversification benefit from fixed income ... led to a significant spike in the volatility of, and likely a subsequent deleveraging from, some risk parity strategies.

Here is an example:

In this hypothetical example the current deleveraging would be the 7th largest (Table 3) but could be the most impactful on markets given the growth in assets tracking risk parity in recent years.

I am not here to bash risk-parity strategies. While naysayers can point to the $80 billion Bridgewater All-Weather Fund, which is a key flagship RP fund, was -4.2% in August and -3.8% YTD, Ben Carlson has also correctly pointed out that these strategies have had a terrific long-term record. They have largely worked and done their job (also see this highly useful primer from FT Alphaville on the risk-parity strategy).

Market implications
Nevertheless, the latest volatility storm, which was sparked by a number of volatility-related strategies, has a couple of important market implications.

Tactically, Kolanovic's analysis suggests that there will be more equity selling in the next 1-3 weeks, which makes me more comfortable my constructive long-term call on equity market, but a warning of a re-test of the 1820-1870 level on the SPX in the near future (see Constructive on stocks, but waiting for the re-test).

Longer term, however, the role played by RP in exacerbating market volatility are likely to reduce their popularity, just as portfolio insurance programs lost some of their shine after the 1987 Crash. There have been many investors warning about the instability created by this class of strategy. Leon Cooperman of Omega Advisors recently complained that the volatility created by "the machines", which include RP, were responsible for his poor performance. Others like Ben Inker of GMO, who characterized these strategies as being "price-insensitive investors" who could create wild swings in market prices (emphasis added):

Another group of price-insensitive investors are managers of risk parity portfolios. These portfolios make allocations to asset classes not with regard to pricing of assets, but rather their volatility and correlation characteristics. Their price-insensitivity comes out in a couple of ways. First, as money flows into the strategies, they are levered buyers of bonds and inflation-linked bonds in particular. Like most strategies, if the money flows out, they are forced sellers of a slice of their portfolio. Second, unlike many other investors, they will also tend to buy and sell based on changes in volatility. As the volatility of an asset falls, these strategies will tend to lever it up further, and as the volatility rises they will sell. 

Bloomberg interviewed leading MIT finance academic Andy Lo, who said that volatility-related strategies are exacerbating market volatility  (emphasis added):

Question: Are volatility targeting strategies part of the story? Have they become so popular that they’re exaggerating the moves?

Lo: Not only are they exaggerating the moves, but I think they are creating volatility of volatility. So it’s making the market quite a bit more complicated and the dynamics now are much more different and much more difficult to manage if you’re not aware of how these dynamics play out.

Lo also believed that RP is becoming a crowded trade:

Question: Is risk parity looking like a crowded trade?

Lo: I think there’s definitely a case in point of the idea of alpha becoming beta. The idea that once you start popularizing a particular investment approach, and it becomes so popular, that in and of itself creates these kinds of shock waves. So for example if the strategy itself underperforms, now we have a larger number of investors that are going to be unwinding that strategy and that will create a kind of cascade effect where the strategy will underperform even more as people start to take money out of the strategy. There are a number of examples. Risk parity, of course, is the most recent. But before that trend following, before that value investing, growth investing, earnings surprise, earnings momentum, any kind of a strategy can become a crowded trade. And when it does you have to just make sure that the risk premium associated with that trade is commensurate with the potential risks of getting hit with these unwinds.

Question: Are volatility targeting strategies part of the story? Have they become so popular that they’re exaggerating the moves?

Lo: Not only are they exaggerating the moves, but I think they are creating volatility of volatility. So it’s making the market quite a bit more complicated and the dynamics now are much more different and much more difficult to manage if you’re not aware of how these dynamics play out.

While the criticism coming from Bridgewater competitors like Cooperman and Inker might be viewed as self-serving, academics like Andy Lo are adding to the chorus that the risk-parity strategy is becoming a case of "alpha turning into beta". This view will prompt the pension and investment consultant community to start to pull back from this investment approach and assets will start to dwindle in the years to come.

This brings up a second important investment implication. My assessment that this was a once in 20 year tail event will likely turn out to be true. We will not see a repeat of the violent downdraft in stock prices caused by these kinds of volatility related strategies, which makes me also comfortable with my assertion that my Trend Model trading strategy was trading off better average returns for tail risk, which happens every 20 years or so (see Trend Model August report card: An invaluable lesson in model design).

In 20 years, the lessons from this volatility storm will be forgotten and a new generation of quants and financial engineers will have dreamed up other ways to crash the market.

Can you actually replace an analyst with a robot?

Last week, the WSJ had an article entitled "Can You Tell the Difference Between a Robot and a Stock Analyst? Wall Street tries out research reports written by artificial intelligence":

Each day, Wall Street churns out millions of words encouraging investors to buy or sell stocks, bonds and mutual funds.

In the future, more of those words might not be written by humans.

As automation in financial services grows, computers and algorithms have taken on some of the traditional work of traders, clerks and financial advisers. Now, a host of startups that use artificial intelligence to write news stories and other reports have set their sights on writing work at banks and financial-service companies.

The article claimed that Artificial Intelligence, or AI, is taking over much of the task of analysis:

Narrative Science Inc., which launched in 2010 with computer-generated news articles, added products for financial-services businesses in 2013. Those firms now account for 60% of the company’s client base. Last year, Automated Insights added insurance company Allstate Corp. to a roster of clients that includes Yahoo! Inc.and the Associated Press. Other startups offering automated reports for financial-services firms include Yseop, Capital Cube and Goldman Sachs GS 0.37 % -backed Kensho Technologies Inc.

As artificial intelligence takes on ever more tasks, Wall Street is getting more comfortable putting it to use. Services launched in recent years are now gaining traction because the technology has become more sophisticated and banks are looking for ways to cut costs and increase efficiencies.

What do analysts do?
To evaluate the claim that AI could take over the job of company analysts, let`s consider the job of a sell-side equity analyst, from the tasks that are the easiest to replicate to the ones that are the hardest:

1. Report on industry and corporate developments
2. Issue buy and sell recommendations
3. Communicate insights about a company and its industry

Most individual investors focus on (1) and (2), while institutional investors appreciate and pay for (3). Indeed, the job of reporting on industry and corporate developments could be automated today. A number of news organizations have experimented with using software to put out company earnings reports using a standardized template.

The "quant" investing robot

The task of issuing buy and sell recommendations have also been computerized for years. It's called quantitative investing, but differs from fundamental investing in a key way. Typical quantitative investing techniques rely on factors, such as the insight that "low P/E stocks tend to beat the market". So what you do is buy a whole bunch of low P/E stocks.

Where quantitative investing differs from the kind of fundamental investing done by company analysts is the degree of confidence in the forecast. Quants tend to take a statistical approach and buy a large and broad portfolio of low P/E names, whereas a company specific fundamental analyst will have a far higher level of confidence in his forecast.

Richard Grinold pioneered a principle of portfolio management called "The Fundamental Law of Active Management" which boiled down to the idea that you should bet in proportion to the degree of confidence you have in your forecast alpha (see my previous discussion Examining your assumptions: The Fundamental Law of Active Management). Based on those ideas, a typical quant portfolio will hold between 100-300 stocks, because it is betting on factors or models, whereas a typical fundamentally driven stock picking portfolio will have 20-60 names.

Today, you can buy quantitative system insights from outfits like Value Line, which has had an impressive multi-decade track record, and ranks stocks with Timeliness rankings of 1 (best) to 5 (worst). Does that count as a robot replacement? (Yes, I know that Value Line has human analysts, but not all ranking systems do and I use Value Line purely as an example of a quantitatively driven buy-hold-sell system.)

What is fundamental analysis?

It is the more "creative" part of fundamental analysis that will be difficult for AI software to replicate and model. 

Let me give you an example. Early in my career, I had the privilege to work with a small cap analyst who went on to be a portfolio manager and now the co-head of equities of a major asset management firm. From watching what "Doug" did, I gained invaluable insights that made me a much better quant later in my own career. Small cap analysis is different from being an industry analyst because the drivers for each company and industry are different. When he picked up coverage of a company, Doug would spend several weeks chasing down what he believed the important drivers of value for a company by speaking to the company, its competitors, suppliers, customers and industry association. He would often waste days and weeks chasing down blind alleys only to discover that the issue that he was researching was irrelevant. Only once he had developed a framework for understanding how the business worked would he actually build the spreadsheet for valuing the company`s shares.

Can AI software do that? I doubt it, largely because we don't fully understand how human creativity works yet.

No doubt, we can build software to perform analysis according to certain standardized templates. Perhaps smart software might have been able to read financial statement footnotes to figure out the kinds of risks that Enron had been taking, as the Street seemed oblivious. Can software do channel checks to gain insights as to the public perception of a new product launch, maybe. Those levels of sophistication are probably one or two generations ahead - for now.

As well, investment insights are multi-dimensional and different fundamental analysts may have different insights that institutional investors find valuable and will pay for. As an example, I recall that as a portfolio manager, I derived different levels of understanding of tech darling Nokia during the late 1990's from different analysts. 

The American analysts were much better at business strategy. Nokia was the dominant producer of cellphones and had enormous economies of scale advantages to their then rivals like Motorola, Ericsson, Samsung and others. Nokia's handset operating margins were in the high teens whereas the margins of their best competitors were in either high single digits or low teens. That, reasoned many US analysts, was a source of competitive advantage for Nokia and what made the stock a buy.

By contrast, the European analysts were much better at on-the-ground analysis of the success (or failure) of new handset models rolled out by the major European handset makers. That kind of channel check insight was also a source of tactical advantage. In effect, I read research from the American brokers for the big picture and spoke to the local European brokers because they knew where all the bodies were buried.

Risk is multi-dimensional. Alpha is multi-dimensional. We are not at that stage of software development where AI can give us insights for everything. True, many of the basic reporting functions can be replaced by software - that's not where the real analytical insight comes from.

How to make your first loss your best loss

I have always found that the time when I have learned the most about an investment process is when it does not perform well. After all, it is during periods of drawdown that the blemishes that appear in a model and no amount of backtesting will warn you of those shortcomings. If you are willing to learn from periods of underperformance, you can, in effect, make your first loss be your best loss. This is a case study of how I learned to diagnose a model`s shortcomings and learn from that experience.


Trend Model underperforming
My Trend Model has seen some stellar returns for the past 18 months (last report card here), but had experienced some spotty performance in the past few months. While the returns were not disastrous, they were not up to the levels seen in earlier periods. A look at buy and sell signals in 2015 show a pattern of market choppiness and signal whipsaw:

In a way, that`s not a surprise. Regular readers will know that the Trend Model is based on the application of trend following techniques to global stock and commodity prices. The price volatility experienced for most of 2015 has led to an environment that is unfriendly to trend following models.

Is this just a "feature" of these kinds of models that has to be endured?


A long and short term regime change
Maybe not. In a recent post (see 3 secrets from the Book of (Trend Following) Revelations), I highlighted a study by James Paulsen of Wells Capital Management showing that the long-term trend in equity prices were getting overdone. Paulsen found that stock prices had more or less gone up in a straight line in the last few years and with only minor corrections.

He went on to calculate the rolling three-year R-squared of stock prices and found that they follow cycles of high and low levels of price trends. We just happen to be at the top end of a trending period, which is likely to end soon. When it ends, it will signal an intermediate term top for the US stock market.

In a more recent post (see Calling an audible (for more choppy markets)), I reproduced the Paulsen study and further extended some of the conclusions of that research.

Instead of just looking at the R-squared of 36-month rolling regression, I also examined the R-squared of a 6-month rolling regression as a measure of the short-term trend, largely because my Trend Model uses a much shorter lookback period and it is therefore more correlated with short-term trends than long-term trends. In my last post, I reported that I found that we are in the period where the long term trend is strong (top panel), but the short-term trend is weak (bottom panel).

I further showed that stock prices are likely to roll over in the next few months, given how extended the long-term trend is. In addition, the rollover of the short-term trend is likely a leading indicator of the long trend.

Going back to 1900
I am grateful for all the comments, feedback and suggestions that I have received since I started writing on this topic. One of the comments that I received is that the sample size of this study is absurdly low. I could extend the lookback period of the study using data for the Dow, which goes back to 1900, instead of the SP 500, which only goes back to 1950.

With that suggestion in mind, I reproduced the study using DJIA going back to 1900. Here is the R-squared chart. While the readings are slightly different from the SP 500 study, the longer time horizon revealed some interesting insights. First, the incidence of high 36-month R-squared readings was higher in the pre-1950 period. Nevertheless, the current reading of 0.957 is comparable to the reading of 0.930 in pre-Crash 1987 and 0.911 in pre-Crash 1929. Similarly, the spread between the 36-month and 6-month R-squared readings (bottom panel) is also at a similar order of magnitude when compared to 1987 and 1929.

There is an important caveat to remember! This model measures the direction of the move and not the magnitude. Just because the trend is so extended today doesn't mean that a market crash is around the corner. Other episodes have resoled themselves in 10-15% corrections.

Nevertheless, based on the current 36-month to 6-month R-squared spread of 0.844, I looked at what the return pattern of the DJIA was during past episodes with similar characteristics. The sample size was a more reasonable 16, compared to the minuscule N=4 in the SP 500 study that went back to 1950. The market outperformed initially, but rolled over at between 3-6 months after the first time the spread went above 0.8 (which was March 2015).

And if the trend got even more extended and the 36 to 6 month spread went to 0.9? The results were more dramatic, as the market declined almost immediately.

Historical analysis from Dana Lyons found a number of narrowly range-bound markets that appeared to coincide with tops in 36-month R-squared readings, though the samples did not totally overlap. Lyons had some good news and bad news for stock investors:

First, the good news. All 9 of the prior ranges saw the SP 500 eventually go on to make new highs, though some initially broke the range to the downside first. 2 months later, 8 of the 9 instances saw the SP 500 not only higher but at a new 52-week high. Only the 2007 instance saw the index lower 2 months later, although it was at a new high 3 months afterward (and it had made a new high immediately following the range break).

Now the not-so-good news. On 5 of the 9 occasions, the new highs were very small and very short-lived. Following occurrences in 1965, 1976, 1983, December 1993 and 2007, the market’s upside “breakout” resulted in tops shortly afterward that would predominantly hold for the following year. Only the occurrences in 1951, 1964 and 1995 saw the market persist at new highs for an extended period of time. Furthermore, only 1995 saw the SP 500 continue on to double digit returns over the following year.

In other words, the market did perform well initially, just as my analysis shows, but they more often than not marked a significant intermediate term top.

Weakening long-term trend, choppy short-term trend
My research results using DJIA data going back to 1900 confirm the conclusions of past studies. We are in an environment where the long-term trend is strong but starting to weaken. The weakness is evidenced by a faltering and choppy short-term trend, or price momentum.

For someone using a trend following techniques with lookback periods similar to the shorter 6-month trend, this suggests that the current unfriendly period for this kind of trend following model is temporary. Under these circumstances, I can choose from four options going forward:

1. Status quo: Continue to run the Trend Model as is and accept the drawdowns as a "feature" of the model.
2. Wait for a friendly environment: Go to cash and wait for signs that the trending environment has re-established itself.
3. Focus on the long-term trend and ignore the short-term trend: Most trend following models use a long term moving average to define the trend, e.g. 200 days, and a short-term moving average for risk control, e.g. 50 days. This approach would throw away the shorter (50 dma) and focus on the longer (200 dma) for trading signals.
4. Focus more on counter-trend models. Examples would be contrarian sentiment models looking for crowded longs and shorts, as well as overbought-oversold trading models.

I rejected options 1 and 2 out of hand. Following the status quo is an example of not learning about the investment process and not allowing your first loss to teach you a lesson. Going to cash is a cop-out and detracts from learning. I did consider option 3 seriously, but rejected it as it would result in excessive return volatility because of it strips away the risk control element out of the investment process.

I wound up adopting a version of option 4. I would focus on one side of the counter-trend models by fading strength (but not buying weakness). As my analysis indicates that the long term trend is turning and the decline from an intermediate term top could only be weeks away, buying weakness is the equivalent of picking up pennies in front of a steamroller. Selling strength when the market is overbought, on the other hand, is likely to be a higher percentage play.

The acid test of learning
In conclusion, this is a case study of how I learned about a model from drawdowns. When I interviewed investment managers in the past, I have always asked the acid test question about how they have learned, "Under what circumstances would your investment approach fail?"

If the manager has thought sufficiently about his strategy and he has learned from past mistakes, he will give an intelligent answer. It also shows that he sufficiently understands the kinds of bets that he is making, why it works and when it might fail.

That`s how you learn to make your first loss your best loss.