System with 100% Accuracy Over 55 Years: "Bear Market Has Started"

Research
21

Marc, I think you have me all wrong. I read your post in another thread saying essentially that ‘market timing doesn’t work.’ You have written statements like that fairly often and I felt someone should respond. I was simply pushing back against your declaration and showing examples of market timing definately working.

In fact, I created a very simple market timing system for that purpose, which I and another p123 member tested using another website with 55 years of S&P500 data. The system has settings that anyone can put together and use on p123. That backtest ran from 1960 to present and it showed that when using a very simple 3-indicator system, the result was 100% winners.

My motivation was to provide p123 members with a simple market-timing system that could help them avoid severe losses that make so many give up on investing. Now you have responded back with what sounds to me like an accusation of fraud:

I launched IntelligentValue.com in September 2004 and have had the previledge of serving many hundreds of subscribers in the last 11 years. If I were defrauding people, you would think that there would be complaints against my company. However, if you check our record with the Better Business Bureau, you’ll see that we have an A+ rating without a single complaint in 11 years.

Also, I would like to clarify one thing for everyone: I am not selling anything! I am not accepting subscribers! If you go to the join page on my site, you will see that I am turning people away… and referring them to the free subscription page if they are interested in receiving my free take on the markets whenever I am sufficiently motivated to write a newsletter. As I stated previously, I am moving on to something else as I prepare for early retirement, something that in the last 10 years was made possible by p123.

My motivation in starting this post was to a) try to show that your blanket statement saying “market timing doesn’t work” is not the only point of view, and b) give back to p123 members with a little food for thought. A bear-market signal is being given by my very simple technical analysis system (settings for which I freely shared in my two previous posts).

I am not (as you said) “insist(ing) with such vehemence that 100% success is so easy to achieve.” Are you kidding me? Success at anything in life is damn near impossible to achieve. But for the community of p123 members who use this site and read these blogs and don’t know better, I feel they are being shortchanged when told by a p123 expert/representative that “market timing doesn’t work.”

P123 has some very good tools that can be used to implement a well-conceived market timing strategy, and thanks to the dedication and perseverance of Marco, it’s getting better all the time. By combining the world-class, p123-based quantitative stock selection with some basic market timing to mitigate significant losses, we can achieve outstanding results; in my experience, results that are dramatically better than stock-selection used alone.

Marc, I know you have dedicated your career to fundamental stock analysis and converting that analysis into language a computer can understand. As I said in my earlier post, in my opinion you have an incredible body of work to show for your efforts. Heck, you ‘wrote the book,’ as they say, back before most investors had ever heard of quantitative stock selection. Your tutorials and advice here on p123 are excellent and I have made posts saying so several times over the years since you joined the team.

You might think it’s getting late in your career, but I hope you will consider keeping an open mind about technical analysis and market timing. To advise members that it just flat ‘doesn’t work’ is short-changing those individuals who don’t have much experience and are impressionable.

We have similar career paths. I also started my investing career as a fundamental analyst, working for Drexel Burnham and Merrill Lynch in the 1980s and I thought TA was hogwash up until 2006, when I decided to give it another try at almost 50 years old. Robust market timing systems that significantly enhance a portfoli’s returns can be created, and even ‘old dogs’ like us can learn new tricks.

From the tone of your last post, it sounds like I have offended you. That was not my intention, sir. I was simply trying to prompt you into considering the facts that I presented as evidence of the possibility that market timing can work. If you feel I insulted you in some way, I apologize.

22

However good one’s intentions might be, no system is 100% accurate, and to imply that it is easy to time the market is misleading. It would probably only take one false positive for the average investor to abandon ship, with a huge opportunity cost that would be very difficult to recover.

Market timing can be highly susceptible to data mining, largely due to the relative infrequency of significant market corrections. In contrast, a quantitative model based on fundamentals (e.g., value) would potentially generate hundreds of trades over a 15-year backtest, which gives a much higher degree of confidence in the empirical results, which are also backed by academic research that spans over many decades.

Why not focus on the true strengths of P123, which are factor-based strategies based on fundamentals, and a large investable universe that is inaccessible for most professional asset managers?

23

Great article, perfectly relating to this discussion:
https://blog.thinknewfound.com/2015/09/trend-following-market-timing/

24

Chris,

you did not prove to us, you did not even show us, that your market timing systems are very successful. This could be done with, say, 10 OOS signals, which overall generated convincing market outperformance or at least DD reduction. I read nothing like that from you. To construct a successful in-sample backtest with only 30 or 40 signals (= 15 or 20 (closed) trades) is no feat.

I am not against market timing, and I hope, it can improve some strategies at least a little.

Matthias

25

Actually, I’m all for market timing, and have even written favorably on it on Seeking Alpha several years ago.

What I’m against is data-mined market timing; reliance on signals that have worked in the past even when it’s becoming clear that the world is changing in ways that can, have, and are likely to continue rendering the old signals ineffective. Specifically, I was addressing the trend-based SP500 estimate signals Marco, and widely adopted by many on p123. What I said is that these have, indeed, captured large fundamentally-driven declines in the past, that it’s reasonable to assume they’ll continue to do likewise in the future, but that we need to also be worried by the emergence of increasingly scarier rapid non-fundamentally driven declines caused, it seems, by the increasing concentration of money and the increasing concentration of decision making in the market. It’s no accident that the signals failed in 2011. I’ve also encouraged users to try to discover new signals, and I suggested that the next generation of successful signals may involve to one degree or another use of the economic data now available on the platform. As we move forward, I think it will become increasingly important to look, when timing, beyond the market’s internal dynamics and to also try to get a handle on factors that may be influencing the algorithms designed for the big machines, the ones that not only read numbers but even text.

On that basis, I’m definitely hostile to market timing in R2G not because I believe it can’t work but because it appears to me that its practitioners are naively relying on simulated results from a model that has not shown itself likely to be successful in many kinds of potential future bear markets and that by doing so, they are kidding themselves and subscribers into believing they can deliver protections they can’;t really deliver.

As to TA per se, I don’t have much talent for it but have always respected it and worked for ways to combine it with fundamentals, as for example when I was at Reuters and in the time I spent in Salt Lake city with the folks at Meta Stock working with them to fund ways to combine fundamentals into their product and in an article I once published in one of the TA magazines entitled “Sleeping With the Enemy” in which I discussed how I, a fundamental guy, uses technicals and suggested ways for the TA crowd to do likewise (and did a seminar on that at a Traders Expo).

If you go back over the forums and really look at the things that get me riled up, you’ll see that it’s not any idea per se. My problem is with bad process often manifested as a process that disrespects the adage that “past performance doesn’t assure future outcomes,” and adage that isn’t just a legal thing but one that’s very very much true. Another and related manifestation of bad process is population to discover what has worked when in fact our challenge is to use our work within the fixed data set as a basis for jumping off into (extrapolating into) a new dataset, one which may be similar or which may, as is often the case, be different. Financial research is not the same as medical research, physics research, etc. and I’m gratified to see nowadays that some big-time quants off p123 are similarly criticizing “physics envy” that leads to bad work in finance. It has to do with good quant vs, bad quant. That’s why I pound the table on fundamentals. Without understanding why factors operate as they do, one’s ability to extrapolate into the future owes itself to nothing more than market inertia (the extent to which the future continues to resemble the past) or dumb luck (which, if you rely on that, always bites you sooner or later because the market never remains inert for too long). As powerful as many think p123 is, I find it bothersome that so many see and benefit from only a tiny portion of what’s available to them. My aim is not to get anyone to conform to anything but to expand their use of the arsenal sitting right at their finger tips. The market can be a brutal place, and people shouldn’t battle it with one hand tied behind their backs.

I hope this clarifies where I’m coming from. And thank you, I appreciate your having heard me on my objections to easy and 100%.

26

m:

 I may be wrong, but I don't think Chris is claiming 100% accuracy going forward.

Bill

27

I’m more interested in the “Bear Market Has Started” than the “100% accuracy over 55 years.” Anyone can build a backwards-looking model with 100% accuracy.

I think the stock market will continue to climb over the next few years, because the economy is slowly improving and China’s slowdown won’t affect the US too much.

At some point in the future, when everyone is over-leveraged and overly optimistic, and the last bear has switch to bull, then we will have a massive crash in stocks. But we are not there today.

But what do I know, I’m probably totally wrong.

28

First, I would like to say “thank you” to Chris for sharing his findings with everyone. Another “thank you” goes to Tobias for supporting the analysis.

Second, I do not appreciate the hostile comments made by some of us. While there might be some genuine concerns about the thread title, it is quite ridiculous to accuse Chris of

  • Lack of integrity
  • Plagiarism (using the 200-day moving average and monthly MACD)
  • Not selling at each high and buying at each low over the last 55 years
  • Showing us nothing

Honestly, I am quite surprised that Chris still replies. Even if I did not believe in market timing, it would be sad to see him go silent and not challenge my thinking in the future. I would love to see constructive builds, such as running his system in Germany/UK/Japan or with US data pre 1960, ideas to improve the getting-back-in-the-market sensitivity, etc.

If anyone is still interested in learning more about market timing, fellow P123er Fred Piard has published a few interesting articles on seekingalpha.com recently. He uses several indicators and looks at the total of negative/positive indicators to arrive at his market timing signal. http://seekingalpha.com/article/3184376-a-new-systemic-multi-valued-market-timing-indicator

I do not understand how quants believe they can successfully time the buying/selling of individual stocks, but not the market. Isn’t this a contradiction?

Cyberjoe

1 Like
29

Chris, I too appreciate your contributions here, on market timing or other topics, and hope you will continue to express your views. I have used market timing to great success in the past. For one example, I rode the bull market from mid 1990’s to the internet bubble years using a crude moving average system, inspired by my background in digital signal analysis, and it played a role in my early financial independence. I hope P123 can continue to maintain an open forum where subscribers feel comfortable to express diverse or contrarian views, and where participants can maintain appropriate decorum.

30

I think if Chris wants us to believe his market timer is not data-mined, he needs so share the details of the proprietary portion of it. Then we can each decide if it is data-mined. Since this is not the timer he uses on his web site for paid subscribers, there should be no problem sharing this with us.

-Debbie

31

For many years I had a fascination with finding a perfect market timing system. After my own extensive testing as well as interviewing many trading firms and hedge funds (used to be involved in selecting providers for my companies pension plan) I determined that none exists. Markets adapt so a perfect system would need to adapt as fast as the market. Even Goldman Sacs, Bridgewater, etc…, with their supercomputers that can perform over a billion simulations in a short time period and their army of quants have taken substantial drawdowns over the last few years, including the recent drawdown. In general what I have found is that faster systems have more whipsaws, slower systems have larger drawdowns, and “perfect systems” based on backtests underperform out of sample due to data mining/curve fitting because the number of available data points are limited and as I previously mentioned, conditions change. I am not suggesting that market timing does not have a role in risk management but one should keep their expectations in check and be humble. For the awesomeness of ones system will always be less in the future than it was in the past.

Scott

1 Like
32

I found Chris’s post interesting and welcome his contribution, but he’s not doing himself any favours by appearing to make overblown claims. His market timing system does not have 100% accuracy, because it doesn’t pinpoint the exact bottom and top of previous market cycles. Market timing using technical analysis isn’t easy. He previously posted about reliably making 100% a year, which most investors of any experience would scoff at. I’m not surprised that someone making great claims about their approach is treated with a great deal of scepticism.

33

Chris Michaels,

I don’t think there’s a single member of this site that would disagree that the study of price action via technical analysis can contribute in a meaningful and significant way to improving returns in both up markets and down. However, this Ford-vs-Chevy-style debate in this and other articles has not been constructive. I also share Dodge1664’s recollection of the 100% return claim post you made and the sour exchange that followed. That post was subsequently removed from the site.

In the more recent post that inspired this thread, named: “Anybody caught this downturn?”, on page 8 , you posted the following on 9/2/15:

I responded, approximately 1 hour later that I was unable to access this section of your website as it requires a user name and password. That link now seems to take users to a generic landing page. You never responded as to why a password was required or that the issue had been resolved. Is newsletter signup required to view the free content? If your goal is to enlighten the P123 community, then I humbly request you follow through with your offer to make the Intelligent Market Risk Analysis free on your site. Or preferably, reprint it here on the P123 forums. Let the light shine in, Chris.

Lastly, I would also echo Marc’s challenge to begin demonstrating your marketing timing and stock selection ideas in an R2G model. Your stated goals would be easier to achieve with this audience if you had some OOS performance data to share.

34

I generally agree with this outlook, but as Warren Buffett coined “You only find out who is swimming naked when the tide goes out.” It’s been a while since we’ve had a correction, it might have given a lot of people extended cover for bad behavior that we aren’t aware of. I kind of wonder what this correction might turn up, and what kind of contagion it might have on the broader market. I specifically wonder about all these non-liquid, leveraged and boutique ETFs that have flooded the market in recent years that have yet to be tested with a 10%-20% market correction.

35

[quote]
Mathiaus: “…you did not prove to us, you did not even show us, that your market timing systems are very successful. This could be done with, say, 10 OOS signals, which overall generated convincing market outperformance or at least DD reduction. I read nothing like that from you.”
[/quote] If you read my second post in this thread (on page 2), you would see that I did exactly what you suggested. I showed the results of my Relative Value portfolio from IntelligentValue.com, which has been live and updated weekly (aka, OOS) since March 2009. That portfolio has 110 closed transactions and used a combination of market timing to bring the portfolio drawdown to an average of only about 8% (outside of one week in 2009). It has a live (OOS) return of 89.5% over six years based on quantitative selection of value stocks.

[quote]
Debbie (shsunbarot): “I think if Chris wants us to believe his market timer is not data-mined, he needs so share the details of the proprietary portion of it. Then we can each decide if it is data-mined. Since this is not the timer he uses on his web site for paid subscribers, there should be no problem sharing this with us.”
[/quote] I did share the settings except for the one indicator which I use on my site. I will keep those settings to myself. If you eliminate that one indicator, there is nothing proprietary about it. You can use the 10-month moving average and the standard MACD (12,26,9) applied on a monthly basis. It has 80% winners over 55 years and in my backtest it produced a return 250% greater than a buy-and-hold of the S&P 500. Testing it with a multitude of start-dates, offset by months and years, shows that it is very robust. It doesn’t work on a weekly or daily basis. Unfortunately, setting monthly (or other) time-frames on p123 is not available. The 200-day MA doesn’t work as a substitute for 10 months either. It would require a double-smoothed average (the 200 MA of a 200 day MA), which also isn’t available. It’s also not the approach I use on my site to avoid market-related risk. For me the signals are too late. I exited the market on my site when my proprietary system gave a ‘Market Down’ signal in July. That example I gave is of extreme simplicity and was only for demonstration purposes to make a point about the possibilities of market timing.

[quote]
Bill (strader): I may be wrong, but I don’t think Chris is claiming 100% accuracy going forward.
[/quote] You are correct, strader. I made no claims of a forward outcome or any prediction whatsoever. To start a discussion about market timing, I was simply saying, “This long-term timing system just gave a bear market signal. Let’s discuss.” I had no idea I was going to get implications of an SEC investigation all the rest of it from Marc Gerstein.

[quote]
Guenter (CyberJoe): “First, I would like to say “thank you” to Chris for sharing his findings with everyone.”
[/quote] Thank you and others who supported me and a paying member’s (I’ve been a member since 2004) right to post a topic that may be helpful to other members, even if it offends a certain individual.

Guenter (CyberJoe): I do not appreciate the hostile comments made by some of us. While there might be some genuine concerns about the thread title, it is quite ridiculous to accuse Chris of

  • Lack of integrity
  • Plagiarism (using the 200-day moving average and monthly MACD)
  • Not selling at each high and buying at each low over the last 55 years
  • Showing us nothing
    [/quote] Thank you, Guenter, for that support. I’m glad to know I wasn’t just imagining that those were hostile comments.

[quote]
James (JRinne): I say attack away, if you want, but do it with information, facts, numbers and sim results— we will probably learn something. However, arguments along the lines of “this is true because I’m so knowledgeable” are causing some to stop posting, I think.
[/quote] Mark Gerstein: I think he’s referring to you.

[quote]
Marc Gerstein: “Actually, I’m all for market timing, and have even written favorably on it on Seeking Alpha several years ago. My aim is not to get anyone to conform to anything but to expand their use of the arsenal sitting right at their finger tips.”
[/quote] Marc, I’ll give back some of the disrespect you dumped on me and I’ll quote you: “Not impressed.” And “WTF.” Why could you not have posted something similarly reasonable from the start instead of personally attacking me? I always thought you were an asset to p123, but after this experience and reading your other attacks on paying customers you disagree with, I have changed my mind.

Also, I don’t understand how are you trying to get people to “expand their use of the arsenal” on p123 when you make statements saying, and I quote, 'MARKET TIMING DOESN’T WORK" What gets you ‘riled up’ is apparently anything you don’t agree with. It gives me “a deep sense of revulsion” to deal with cyber-bullies, especially ones who work for the company to which we are all paying good money. It’s very unprofessional and you should be ashamed of yourself. Unfortunately for you, this thread will be here forever and everyone will be able to see how you treat people with whom you disagree.

[quote]
sglinski: “Chris, I… appreciate your contributions here, on market timing or other topics, and hope you will continue to express your views.”
Guenter (CyberJoe): Honestly, I am quite surprised that Chris still replies.
[/quote] Thank you for your support, guys, but this will be my last post on P123… ever. To make a post that attempts to give a different point of view and inspire ideas, then have an employee of P123 ridicule me, make over the top ‘challenges’ requiring “legally enforceable guarantees” and the posting of a bond, and imply that I am surreptitiously deceiving people is too much. I’m going silent and won’t be reading any more of this thread or making any posts to the forums.

Goodbye and good luck to all.

36

[quote]
I did share the settings except for the one indicator which I use on my site. I will keep those settings to myself. If you eliminate that one indicator, there is nothing proprietary about it.
[/quote] Looks like CCI 50

37

[quote]

Ah, the oldest bulletin board in the trick. Distort what someone says and then rant about it. You go Chris! :slight_smile:

And yes, I am going to speak up and try to give sensible support to users, and if this inhibits others who prefer to advocate reckless practices like data mining, that’s a good thing. While anyone is free to express any opinion they wish, nobody has a right to expect things to be unopposed. I’m fine explaining and if someone argues against it, justifying what I say. If there are who others aren’t, well I think there’s an important message there about what they’re trying to say, a message that’s especially important when the things being discussed aren’t just for fun (see, e.g. espn.com, reddit.com) but can cause people to sustain genuine financial damage. This is the sort of company where I believe ultimate customer satisfaction is and will continue to be tied to the dollars-and-cents results they achieve.

38

Chris made his call: Bear Market based on his interpretation of his indicators. Time will tell.
Carter Worth of Cornerstone Macro made same call on CNBC: http://www.cnbc.com/2015/09/29/we-are-in-a-bear-market-carter-worth.html
Technical analysis guys at Credit Suisse look to 1738 too. http://www.businessinsider.com/credit-suisse-on-sp-500-top-2015-9
Evercore ISI’s Rich Ross is in on it: http://www.zerohedge.com/news/2015-09-30/even-evercore-isi-now-using-r-word-and-what-happens-next

Be advised: “If the economy goes into recession, S&P declines average -31%. If the economy remains in expansion, S&P corrections average -15%.” Evercore ISI.

I don’t see enough indications to indicate an imminent US recession… yet. Georg Vrba has done a lot of work on this and I’d like to hear from him, but based on my models, a recession is 3m+ away though regional activity indicators are scaring me. S&P predicts a V-shaped recovery in earnings growth next year because of the crappy numbers this year of which this quarter should be the worst.

We may be in a market correction of 10-19% here based on China growth worries, earnings looking terrible in the aggregate from energy, the Fed sowing confusion with negative dots, etc. This is a pricey market after 6 years of expansion with belief in central bank omnipotence. Markets move in cycles and we are closer to a top than we are to the previous trough. My portfolios are essentially majority beta hedged here until I see indications that a positive Sharpe Ratio environment is re-established.

Good luck to everyone.

39

Shaun,

Many people make public calls but most have an out of sample track record of no greater than 50% (coin toss) so I generally disregard them. In addition many of them have biases that motivate them to make certain calls which are not disclosed. Have you seen an out of sample track record for calls from Carter Worth, Credit Suisse, or Evercore and if so can you please elaborate? Chris purported track record would make James Simon’s jealous so it is reasonable to interpret his claims with skepticism.

Scott

40

Scott,

I generally dislike technical analysis and dismiss it as it always seems to be a story not rigorous and rule-based. I had wanted to test the strategy but he did not disclose the third indicator. Looking at Chris’ charts, there did seem to be some merit to the combination indicating a less robust market ahead.

The market is going through a rough spot and I noticed that some others were making similar calls. It amazes me these guys get listened to and get paid a couple hundred grand! They have to make differentiated calls to sell in the media so I am always suspect.

I did not like the hyperbole that accompanied his initial post but that is just my opinion.