Weighing The Week Ahead: Does Historical Analysis Improve Market Forecasts?

Next Week’s Theme

Many market participants will not report back until Thursday. Investment committees may not meet until January 7th. The weeks with a holiday on Wednesday create an unusual trading environment. I’m not sure how the punditry will find fresh topics, but I expect the 2020 forecasts to continue. With that in mind, let’s take a deeper look at the foundation for market forecasts. In particular:

Does analyzing history improve market forecasts?

Background

Since the 2016 election there has been a growing attack on expertise. Populism seemed to encourage the idea that everything is just a matter of opinion, and all opinions are equal. The outcome of the election and the apparent failure of the major polls encouraged this meme. It seems like anyone who is predicting anything is a target for scorn, usually accompanied by a Yogi Berra quote or one about making fortunetellers look good.

In sharp contrast, the same sources report economic data without any attention to confidence intervals or past accuracy of the source. They feature charts of past data and trends, but without making a prediction. What do they expect the reader to do with this information? The sources are inviting inferences. This transfers the forecasting responsibility to the reader, avoiding any possible blame on the part of the author.

If all forecasts are terrible, what is the point of reporting data? It is past time for someone to think carefully about what a useful forecast entails.

My inspiration

Jeff Sommer (NYT) writes He calls forecasts by Wall Street strategists “flagrantly inaccurate” comparing them to a perma-sunny weather forecaster. Based on some calculations from Paul Hickey of Bespoke Investment Group (a reliable source), he summarizes the median forecasts and market returns since 2000. “On average” he summarizes:

The median forecast was that the stock index would rise 9.8 percent in the next calendar year. The S&P 500 actually rose 5.5 percent.

The gap between the median forecast and the market return was 4.31 percentage points, an error of almost 45 percent.

The median forecast was that stocks would rise every year for the last 20 years, but they fell in six years. The consensus was wrong about the basic direction of the market 30 percent of the time.

Mr. Hickey found that the forecasts were often off by staggering amounts, especially when an accurate forecast would have mattered most. In 2008, for example, when stocks fell 38.5 percent, the median forecast was typically cheery, calling for an 11.1 percent stock market rise. That Wall Street consensus forecast was wrong by 49.6 percentage points, and it had disastrous consequences for anyone who relied on it.

He concludes that investors should just buy index funds.

Mark Rzepczynski, one of my favorite sources, writes . Proclaiming weather forecasters to be the best in any field of study, he praises them for using percentages. He also points out that short-term forecasts are better than those for longer times. He suggests topics for inclusion in a year-end outlook report.

Basis for Criticism?

Neither article provides much help to the average investor.

My Approach

I agree with much of the forecast criticism, and I don’t attempt it. Some analysts even specify detailed scenarios showing the path the market will take during the year. Even a forecast of a 30% gain for the year may not be helpful if there is a 25% decline along the way. Such detailed guesses are not needed to improve investment results. I suggest the usefulness of two factors.

1. The attractiveness of stocks relative to other assets.
2. The strength of the economy relative to potential. A cool economy generates more corporate earnings as it improves. An overheated economy eventually does the opposite.
3. Do not insist on a twelve-month time frame and certainly not a specific calendar year. Some sort of rolling forecast method would be better.
4. Focus on risk, not a specific result. Since recessions are associated with the greatest risk to markets, that is a good place to start.

This approach dramatically reduces the chance of big losses and keeps you on the right side of the market most of the time.

Please note that none of these ideas depend upon history. A focus on the prior year history is a losing method, not a basis for forecasting. Here are some examples of what we can and cannot predict.

Can Predict

Can’t Predict

Relevance

There will be tweets

The number or content

Recognize the short-term reactions for what they are

There will be some market swings

When or exactly how large

An opportunity for those confident in fundamentals

Odds of an election outcome

Who will win

Understand that a 70% chance means that the forecast will be wrong 30% of the time

Odds of a recession

Whether a recession will actually occur

Knowing this provides an important edge – enough to be very helpful

Which sectors have the most potential gain

Exactly when the gain will occur

For a long-term investor, this is good enough

The odds of your favorite team winning the Super Bowl

Whether it will actually happen

Packers?

The chances of a freeze tonight in Phoenix

Whether it will actually happen

There is actually a warning tonight. This is a good example of recognizing a risk.

Odds of a hurricane striking mainland

Whether it will actually happen

Very important for reducing risk – life, injuries, property

I hope you can see the recurring theme. Knowing the odds, especially if derived from a solid process, can help your results. Contrast this with a prediction from a big Wall Street firm this week. “There is a 50% chance of a correction.” What do you do with that information? Even after the fact it can be neither proved nor disproved. Now that is a safe forecast!

I’ll have some additional observations in today’s Final Thought.

Does analyzing history improve our market forecasts?

Last Week Recap

In my last installment of WTWA, I accurately predicted continuing discussion of 2020 outlook from some names and faces that are not as familiar to the investment community. That was pretty easy. There was discussion of my theme, complacency, since that is a popular way of describing the market.

The Story in One Chart

I always start my personal review of the week by looking at a great chart. This week I am featuring the version from Investing.com . If you check out the interactive chart, you can see the related news sources and add your own indicators.

The market gained 0.6% for the week. The trading range was only 0.7%. The media portrayed this as a “grind upward,” and I suppose that is technically correct. Not much was happening. Daily volume was less than half that of the prior week. You can monitor volatility, implied volatility, and historical comparisons in my weekly Indicator Snapshot in the Quant Corner below.

Noteworthy

A philosophy professor was concerned about poor student performance. To test an idea, he offered extra credit to students who would give him their cell phones for nine days and then write about the experience. I cannot summarize the results without spoiling the story, but please make your own guess about what happened before reading it. The original experiment was in 2014. Last year he repeated it in a different environment. There was a dramatic difference in the results, but again, not what you might think. (MIT Technology Review ).

The News

Each week I break down events into good and bad. For our purposes, “good” has two components. The news must be market friendly and better than expectations. I avoid using my personal preferences in evaluating news – and you should, too!

New Deal Democrat’s high frequency indicators are an important part of our regular research. The results remain positive in both the long- and concurrent-time frames, and the short-term forecast has improved to neutral. NDD continues to highlight the metrics to watch if concerned about manufacturing weakness spreading to the consumer. I suggest that readers check out the full post to appreciate the comprehensive nature of this work.

The Good

Initial jobless claims edged lower to 222K, in line with expectations but improvement over the prior week’s 235K.

Hotel occupancy increased on a year-over-year basis. (WSJ ).

The Bad

Durable goods orders for November declined 2.0% and the October results were revised lower to a gain of only 0.2%. This result missed expectations for a gain of 1.4%.

New home sales for November were 719K (SAAR), missing expectations of 735K. October’s report was also revised downward to 710K. GEI ) reports this subject in the depth we expect from him. This table is just a sample of his analysis, including several interesting charts.

The MBA Mortgage Index declined 5.3% similar to the prior week’s 5.0% decline. This is a story about seasonal weakness. Earlier this year Calculated Risk reports that applications are up 5% on a year-over-year basis.

The Ugly

Ugliness does not cease during the holiday time, but I’m not going to discuss it this week.

The Week Ahead

We would all like to know the direction of the market in advance. Good luck with that! Second best is planning what to look for and how to react.

The Calendar

The economic calendar is normal with the important reports (ISM Manufacturing, ISM Services, and auto sales) scheduled for a Thursday or Friday release. If last week’s pattern repeats, we will have a quiet week with plenty of time for continued year-end assessments.

Congress is on recess and the President is on vacation, but don’t expect the political news to stop!

Briefing.com has a good U.S. economic calendar for the week. Here are the main U.S. releases.

Next Week’s Theme

Many market participants will not report back until Thursday. Investment committees may not meet until January 7th. The weeks with a holiday on Wednesday create an unusual trading environment. I’m not sure how the punditry will find fresh topics, but I expect the 2020 forecasts to continue. With that in mind, let’s take a deeper look at the foundation for market forecasts. In particular:

Does analyzing history improve market forecasts?

Background

Since the 2016 election there has been a growing attack on expertise. Populism seemed to encourage the idea that everything is just a matter of opinion, and all opinions are equal. The outcome of the election and the apparent failure of the major polls encouraged this meme. It seems like anyone who is predicting anything is a target for scorn, usually accompanied by a Yogi Berra quote or one about making fortunetellers look good.

In sharp contrast, the same sources report economic data without any attention to confidence intervals or past accuracy of the source. They feature charts of past data and trends, but without making a prediction. What do they expect the reader to do with this information? The sources are inviting inferences. This transfers the forecasting responsibility to the reader, avoiding any possible blame on the part of the author.

If all forecasts are terrible, what is the point of reporting data? It is past time for someone to think carefully about what a useful forecast entails.

My inspiration

Jeff Sommer (NYT) writes He calls forecasts by Wall Street strategists “flagrantly inaccurate” comparing them to a perma-sunny weather forecaster. Based on some calculations from Paul Hickey of Bespoke Investment Group (a reliable source), he summarizes the median forecasts and market returns since 2000. “On average” he summarizes:

The median forecast was that the stock index would rise 9.8 percent in the next calendar year. The S&P 500 actually rose 5.5 percent.

The gap between the median forecast and the market return was 4.31 percentage points, an error of almost 45 percent.

The median forecast was that stocks would rise every year for the last 20 years, but they fell in six years. The consensus was wrong about the basic direction of the market 30 percent of the time.

Mr. Hickey found that the forecasts were often off by staggering amounts, especially when an accurate forecast would have mattered most. In 2008, for example, when stocks fell 38.5 percent, the median forecast was typically cheery, calling for an 11.1 percent stock market rise. That Wall Street consensus forecast was wrong by 49.6 percentage points, and it had disastrous consequences for anyone who relied on it.

He concludes that investors should just buy index funds.

Mark Rzepczynski, one of my favorite sources, writes . Proclaiming weather forecasters to be the best in any field of study, he praises them for using percentages. He also points out that short-term forecasts are better than those for longer times. He suggests topics for inclusion in a year-end outlook report.

Basis for Criticism?

Neither article provides much help to the average investor.

A critique should define the characteristics of a good forecast. What level of accuracy would Mr. Sommer find to be helpful? Is there another group that could forecast with a gap of less than 4.1%? Or get the annual direction more reliably?

The Sommers advice to buy index funds would have provided no help during the years of “disastrous consequences.” His advice implies a positive expectation from owning stocks. What is that average gain and why would that expectation be less “disastrous.”

Why should we expect the selected strategist group to be the best forecasters and free of bias?

Weather forecasting benefits from the availability of massive amounts of data. Models are hungry for information, and the weather forecasts get better every year. There are not enough valid years of stock returns and estimates to do a meaningful comparison.

My Approach

I agree with much of the forecast criticism, and I don’t attempt it. Some analysts even specify detailed scenarios showing the path the market will take during the year. Even a forecast of a 30% gain for the year may not be helpful if there is a 25% decline along the way. Such detailed guesses are not needed to improve investment results. I suggest the usefulness of two factors.

1. The attractiveness of stocks relative to other assets.
2. The strength of the economy relative to potential. A cool economy generates more corporate earnings as it improves. An overheated economy eventually does the opposite.
3. Do not insist on a twelve-month time frame and certainly not a specific calendar year. Some sort of rolling forecast method would be better.
4. Focus on risk, not a specific result. Since recessions are associated with the greatest risk to markets, that is a good place to start.

This approach dramatically reduces the chance of big losses and keeps you on the right side of the market most of the time.

Please note that none of these ideas depend upon history. A focus on the prior year history is a losing method, not a basis for forecasting. Here are some examples of what we can and cannot predict.

Can Predict

Can’t Predict

Relevance

There will be tweets

The number or content

Recognize the short-term reactions for what they are

There will be some market swings

When or exactly how large

An opportunity for those confident in fundamentals

Odds of an election outcome

Who will win

Understand that a 70% chance means that the forecast will be wrong 30% of the time

Odds of a recession

Whether a recession will actually occur

Knowing this provides an important edge – enough to be very helpful

Which sectors have the most potential gain

Exactly when the gain will occur

For a long-term investor, this is good enough

The odds of your favorite team winning the Super Bowl

Whether it will actually happen

Packers?

The chances of a freeze tonight in Phoenix

Whether it will actually happen

There is actually a warning tonight. This is a good example of recognizing a risk.

Odds of a hurricane striking mainland

Whether it will actually happen

Very important for reducing risk – life, injuries, property

I hope you can see the recurring theme. Knowing the odds, especially if derived from a solid process, can help your results. Contrast this with a prediction from a big Wall Street firm this week. “There is a 50% chance of a correction.” What do you do with that information? Even after the fact it can be neither proved nor disproved. Now that is a safe forecast!

I’ll have some additional observations in today’s Final Thought.

h1 Quant Corner and Risk Analysis/h1

I have a rule for my investment clients. Think first about your risk. Only then should you consider possible rewards. I monitor many quantitative reports and highlight the best methods in this weekly update, featuring the Indicator Snapshot.Both long-term and short-term technical indicators remain neutral, but continue to show improvement.Recession risk remains in the “watchful” area. There is little confirmation for the risk signals, which we have been monitoring since May.The Featured Sources: . I put this one last since I do not like the slide show approach. Some of the ideas overlap other articles, but this is the only place I saw “digitized diaper monitoring.”Great Rotation Hint of the WeekMy screening process yielded HA as one of our picks. I did not get it from Peter F. Way’s method, but it is nice to see support from the market makers.This chart from MarketDesk illustrates the shift in sector strength that I am monitoring and trading.There was plenty of excitement in the last decade, but you won’t find it from reading about what stocks did well.Take advantage of this time to think about major trends in all fields. Then you can make your own gentle “forecasts.”[Has your portfolio kept up with the changes over the last year? If you are unsure, write for my brief paper on Four Signs of Portfolio Trouble. Just send an email request to info at inclineia dot com].

h1 Some other items on my radar/h1

Get The News You Want

Read market moving news with a personalized feed of stocks you care about.

Get The App

I am trying to put the worries aside for the holidays. These are not the main stories, so I’m going to focus on some new ideas. But don’t call me complacent!!

Written By: Jeff Miller