Contrarian Investing -- What Do You Think?

Once upon a time, way back when I started screening at Market Guide using a very primitive version of what Portfolio123 eventually came to be, my most successful screen, by far, was one I labeled “Contrarian Opportunities.” Lately, though, to say it’s gone cold would be a polite understatement. At this point, I wouldn’t even consider running it against a larger universe thanPRussell2000 and even that might be overkill; perhaps a smaller-cap universe that pushes the lower boundaries of liquidity might work. But even at that, the latest test results seem hardly worth the effort.

So what about the whole field of contrarian investing? Is it done and gone? Or need the method be refreshed. At the very least, the screen, designed before Portfolio123 came into being, has no ranking system or Quick Rank – so that alone argues for some updating. Still, the unranked lists produced only by screening/buy rules used to work and now they don’t.

Thoughts? Ideas?

For convenience, here are the rules:

Pr4W%Chg < -15 And Pr4W%Chg < Pr4W%ChgInd
EPS3YCGr% > EPS3YCGr%Ind And EPS5YCGr% > EPS5YCGr%Ind
OpMgn%5YAvg > OpMgn%5YAvgInd
ROI%5YAvg > ROI%5YAvgInd
ROE%5YAvg > ROE%5YAvgInd
Close(0)>1 // eliminate penny stocks – redundant if used with a universe that aims at liquidity and which should be improved anyway

The goal of this set of rules was to find companies with above par fundamentals accompanied by subpar recent performance.

For the record – This is NOT a test or anything like that. I had not thought about this screen in ages and have no thoughts right now as to how it might be improved or why it went sour. I’m posting this thread: (i) because another thread about a slide in forum discussion indicates a need to get things going again, and (ii) contrarian investing is one of the topics offered for conference call discussion among a group of advisors with whom I talk regularly, and it stoked my interest.

In the book “The New Market Wizards” by Jack Schwager Gil Blake explains his pullback method for Mutual Funds (before ETFs). It still works. Certain ETFs work better than others. For those who have not found all of the strategies they want to use going forward I recommend looking at this strategy with diversified ETFs or using this strategy as a small part of a large portfolio. GDX is an ETF where this works well–when it is in a general uptrend.

Larry Connors in the Book “Short Term Strategies That Work” has numerous pullback strategies for short term pullbacks in stocks that are in upward trend.

Pr4W%Chg < -15 And Pr4W%Chg < Pr4W%ChgInd looks like a rule for a pullback with the other rules indicating an an upward trend.

I have not tried the screen. But I would be surprised if it did not work. There is certainly a lot of support for the idea.

BTW, Steve (and I am sure others) has a pullback strategy for the SPY ETF. I have not look closely at this SA but I believe the idea is sound

I almost finished reading “Dark Pools: The rise of A.I. trading machines and the looming threat to Wall Street” and learned that A.I. is squeezing more and more alpha out of the market step by step.

These algo’s dominate in the very short term, think nanoseconds to minutes. Almost dominate humans in regular daytrading and are stepping in to take alpha from time horizons extending to weeks. As such I would not be surprised that alpha from pullback strategies and other strategies from the book “Short Term Strategies That Work” from 2008 are slowly disappearing for human traders.

Just my 2 cents, but I would welcome any systems that prove the contrary as I and many other P123 users have no access to advanced A.I. tactics and are still trading themselves. But, I am convinced that the battle for alpha against trading machines is getting harder and harder, especially in liquid stocks and strategies that want to extract alpha within a reporting period.

A good strategy is K.I.S.S.

Universe: S&P 500
Screen for stocks with Low Beta and dividends substantially higher than the average of the S&P 500.
Keep trading turnover low. Trading every week does not help.
Use a simple ranking system.

Marc, I have not had much luck in finding good companies by comparing the company to its industry brethren. One reason, I think, for this is that the industry itself might be in decline (like some retailers are now - I just got whacked by Macy’s) and the second is that a good company might be bucketed in the wrong industry.
Having said that, I would add a rule like Pr26WRel%ChgInd>0 to find industries who are advancing relative to the SP500 over the last 26 weeks but for shorter term, use Pr4WRel%ChgInd<0.
My second issue of companies being mis-bucketed by Industry, I have no input for.
On using contrarian strategies in general, you might also want to use some sentiment factors since they gauge the feelings of analysts. Maybe average recommendation. I had once used the opposite of AvgRec (use higher is better) and got fairly consistent increased performance than if I had used what the analysts recommended (lower better). I had read in a paper somewhere that poorly recommended stocks many times out did highly recommended ones over the following year period. (Maybe an example of everyone piling into it and just driving the price up). So I would be contrarian to the analysts too in some way. I know that decreases the size of the universe since some companies are not covered.

I’ve noticed that, too. I’ve moved on from Ind and now tend to use FRank more (it didn’t exist in the pre-portfoli123 screeners I used) but I’ve often found better results with a universe sort (though there have been some exceptions).

I wonder if it has more to do with just-plain show-me-the-money rather than being a goody goody. Use of #Industry will give genuinely “better” companies without regard to how the industry is doing, and the guru speak I learned early on advocated pursuing these. And it’s a good idea if you really NEED to spread around industry-wise. But if you don’t care about “good” ad just want movers and are willing to accept a tilt that leans away from cold industries, then FRank #All seems to be the way to go.

That definitely is a well-accepted narrative nowadays. But I wonder of its bark is worse than its bite.No amount of A.I. technology can the world process information and develop credible assumptions about the future more quickly. Have you been in the virtual strategy-design seminar? If not, grab the Topic 6 (Momentum) pdf and check the impact of jackrabbit versus patient momentum strategies.

I get a sense most of the A.I. algo proponents are tech and fintech bloggers who need cool topics about which to write and more important, vendors looking to sell the data. I’m not seeing much by users actually demonstrating the alpa that they are earning.

A variant of dogs-of. I’ll take a shot at that.

Here is what the sim looks like. Anybody making a 20% return should be happy. There is no market timing in this sim.


And this is what it looks like if you only select the best 3 stocks. Recently it could only find one stock - MO.


1 Like

Marc, what about the idea of going the opposite direction of analysts opinions/recommendations? Any experience with that? I feel very nervous about swimming against that tide but…

Definitely worth looking into. I’ve found horrible results by trying to go WITH ratings, so incorporating a contrary approach into a model cold be interesting.

Hedgehog,
I am in general agreement. And you convinced me to read “Dark Pools:…”

I was thinking it is not just the algos though. Who thinks they have an edge on predicting BREXIT? Did anyone buy anything that might be affected by how the currencies trade today?

At a minimum the big players have better insight into the polls in UK than I do. Maybe they know which polls are better. I am sure someone has done a meta-analysis. I don’t know: maybe they called one of those UK bookies and talked to him personally. Maybe they paid him to get a detailed discussion of his thoughts.

But here is my question. Do you think there is a big player–maybe Soros, maybe Goldman Sacs, maybe someone I have never heard of–doing an exit poll or a phone poll of those who have already voted as I write this?

My guess is that if it involves enough money for someone it is being done.

I do know that big firms pay for their own weather predictions when buying futures on commodities that might be affected by the weather: e.g., corn or wheat.

Whatever, anyone thinks on this: there is no reason to think I have an edge.

I like low beta because it is contrarian to the whole CAPM model. Buying low beta and shorting high beta seems to give steady, if unspectacular, returns.

Also, I was surprised to see that stocks with low momentum (as defined by the P123 ranking module) have done better since 2009 then high momentum stocks. I still cannot figure that one out. So low-mo.

A couple of reactions.

First, I suggest testing other measures of momentum as well. I created the p123 ranking model and momentum is not my strongest suit, so don’t let that model ber the last word on the topic.

Second, we know momentum is really about persistence. If the factors that drove s stock in period A continue to drive it in period B, then we know the stock shlould act in a way that makles it apperar AS IF the movement in A caused the movement in B. One of the reasons I’ve berern excpressing nervousness aboiuyt momentum lsately is my lack of conviction regarding persistence going forward.

Now here’s a thought. We all crave measures of risk. But it seems to me the fiunance filed got too lazy and honed in too tightly on volatility/standard deviation which really is not what risk is about (Would you fire and/or sue an R.I.A. who delivers too many excess gains as having violated your moderate-risk requirements?) but nothing more than a conveninet shorthand to open the door for quantiative solutions.

Very early in finance, we learn expected value. We articulate X number of scenarios and for each, we assume a potetiakl rfeturn and aprobability. Then we average each return weightecd by its probability. Maybe we can stretch that and think of risk notg as volatiulity per se but slightly differently; as deviation from expectation. If we then assume deviation (risk) is negatively correlated wityh persistence, then we’d also be saying that risk is negsatively correlated with ongoing momemntum. So . . .

Is utpossible to use momentm indications to decvelop some sort of a risk indicator; an expectations-based thing (not unlike the way volatility is used in the options market)?

This isn’t a polished idea; I concocted it as I typed. So it’s really a matter of thinking out loud.

A couple of reactions.

First, I suggest testing other measures of momentum as well. I created the p123 ranking model and momentum is not my strongest suit, so don’t let that model ber the last word on the topic.

Second, we know momentum is really about persistence. If the factors that drove s stock in period A continue to drive it in period B, then we know the stock shlould act in a way that makles it apperar AS IF the movement in A caused the movement in B. One of the reasons I’ve berern excpressing nervousness aboiuyt momentum lsately is my lack of conviction regarding persistence going forward.

Now here’s a thought. We all crave measures of risk. But it seems to me the fiunance filed got too lazy and honed in too tightly on volatility/standard deviation which really is not what risk is about (Would you fire and/or sue an R.I.A. who delivers too many excess gains as having violated your moderate-risk requirements?) but nothing more than a conveninet shorthand to open the door for quantiative solutions.

Very early in finance, we learn expected value. We articulate X number of scenarios and for each, we assume a potetiakl rfeturn and aprobability. Then we average each return weightecd by its probability. Maybe we can stretch that and think of risk notg as volatiulity per se but slightly differently; as deviation from expectation. If we then assume deviation (risk) is negatively correlated wityh persistence, then we’d also be saying that risk is negsatively correlated with ongoing momemntum. So . . .

Is utpossible to use momentm indications to decvelop some sort of a risk indicator; an expectations-based thing (not unlike the way volatility is used in the options market)?

This isn’t a polished idea; I concocted it as I typed. So it’s really a matter of thinking out loud.

Georg

Nice idea. I’ve been fiddling around with it. I don’t go as far down as you in terms of holdings but I’ve seen seeing some nice results. Interestingly, I like what I see better with 3 month rebsalnciong than weith 1 wk or 4 wks.

Not sure, though, that it’s contrarian per se. Your apprach does use the betting-agasitmn-beta anomaly as a universe qualifier.

Rather than Dogs of the SP500 or something like that, perhpas it’s Dogs of the Low-Beta Anamoly. Cool . . .

…so we are back at BESTOGA :slight_smile:

Marc, on momentum, I looked at not only the P123 module but other factors like Pr26W%Chg and it too has the opposite performance of pre 2007 versus post 2010. I created some other custom formulas and kept getting the same results.
I understand there is a reasonable logical explanation for why it should work. I just cannot find the right factor or formula that works since 2009.
I dont want to turn this into a Momentum discussion. Maybe I will start another thread on that.