Hi all, I was searching for studies of technical trading and came across this presentation by Aswath Damodaran that lists some items that might be of interest to users here. It’s more of a broad survey, but might spark some backtests.
http://people.stern.nyu.edu/adamodar/pdfiles/invphiloh/techanal.pdf
I found that interesting. I haven’t tried price to book yet, but I did try PE in my momentum test setup and lower PE did much better than higher.
Better -
FRank("PEExclXorTTM",#All,#Asc)>75Worse -
FRank("PEExclXorTTM",#All,#Desc)>75Thanks for the pdf.
Walter
I’ve had trouble fitting technicals into my models, but have actually found mean reversion more helpful than momentum for my ranking systems.
This was a good year for the January effect I’d guess 
I’ve been following some guys who do charts on twitter and am trying to pick up some ideas, but seems really hard to frame and test many systematically - at least I’ve not been able to apply many of the ideas successfully. I think I’ve been able to to spot some chart patterns and use discretion to avoid a few bad situations, and a couple times identified some bullish cases similar to breakouts - but efforts at painting with a broad brush has not been encouraging. Definitely in early stages of learning though.
I think the simple patterns have been all but arb’ed away. I do, however, believe there’s still gold in 'em hills for those who are willing to dig deep.
A good starting place are the core principles of pattern recognition. What is a pattern? What is an outlier? How are these things quantifiable?
For example, I could try to quantify abnormal-seeming price/volume action based on deviations versus peers with similar business models, market sizes, investment characteristics, shareholder bases, etc. I could also try to explain in terms of economic cycles and market regimes. Again, I think the days of being able to profit from technical analysis in a vacuum are long gone.
That appears to be true in some of the cases I’m examining. I thought I had a good momentum indicator but its alpha faded years ago.
It seems there’s an added hurdle in the context of a multifactor p123 model. It’s one thing to find a technical edge, but quite another to have it benefit a p123 ranking system.
That said, I get the impression that there are traders who make a living trading patterns, so I’m very open to the idea and am studying to pick up possible ideas. (Perhaps I’m mistaken on the this observation - i don’t know.)
I struggle with quantifying patterns in a useful way for testing though. Still learning, however. I’ve had a few recent trades where I think considering the chart probably helped. Acting on one sad chart in particular saved me an additional 10-15% loss. Another breakout setup encouraged me to go ahead and get into a top position in a small cap model, even though I’d prioritized larger, more liquid positions. But it’s visual, not systematic. At this point I’m just tracking my limited decisions in this area to see if I help/hurt.
“Hi all, I was searching for studies of technical trading and came across this presentation by Aswath Damodaran that lists some items that might be of interest to users here. It’s more of a broad survey, but might spark some backtests.”
I respect Damodaran’s work on fundamental analysis and I have actually employed some of his techniques on occasion. An example is how he proposes to calculate cost of equity for stocks that don’t have sufficient market data, or are going through a capital restructuring or M&A. There is some priceless work - readers will have to do their own internet search however for such info because I don’t have links at hand.
He is one of the few academics that I respect. HOWEVER, he is not a technical analyst and I have to be critical of his logical arguments in this presentation. I am going to expose one example from his presentation to state my case. The problem with academics, in general, is that they have an objective in mind at the beginning of a paper and it causes a bias in their thinking and ultimate findings. Academics don’t think “I am going to spend a gazillion hours researching and writing a paper” without first understanding for what purpose they are doing so, how the conclusion will be received by his/her peers, and what the work will ultimately accomplish.
In the case of this presentation, Mr. Damodaran states his bias up in the first three words of the title (Smoke and Mirrors). So it is not surprising to see anti-TA bias within the presentation. I am going to restrict my dissection to his arguments regarding the weekend effect. He presents a nice day-of-week column chart of returns by day of the week from 1927 to 2001.
This particular day of week chart and conclusions is like making the following argument… “over the last century, Americans have drunk 2 bottles of beer per day on average. Therefore prohibition never occurred.” One can not conclude anything based on long-term statistics. This is true not only of technical analysis. but also in fundamental analysis. It is a dynamic environment and factors come and go, there is no universal truth.
Now the folly in Damodaran’s presentation of the weekend effect becomes clear in his next chart which breaks down the day of week effect in 5-year chunks. His intent was to show that the effect was “all over the map” and therefore not tradable. But if his biases were not in control of his (normally very logical) mind, he would have realized that the stock market crash of 1987 strongly shaped the day of week characteristics for at least the next decade. The crash occurred on a Monday and hence dramatically skewed the 1986-1990 Monday returns. And you can also see that traders’ strategies were impacted from that point forward. They no longer held positions over the weekend due to the possibility of a repeat crash. This impact can be seen right into the early 2000s.
The other observation that i would like to make on the chart is that the returns are off by at least an order of magnitude. Positive Monday returns of 0.05% post-1987 crash are simply wrong. I know because I did a lot of work on developing a Monday trading system back in the 1990s. For the SP500 the Monday return was on the order of 1%, not 0.05%. The effect was definitely tradeable and in the opposite direction that Damodaran had suggested based on long term stats.
In summary, I love Damodaran’s work on fundamental analysis but I suggest staying away from his attempts at scrutinizing technical analysis. This is not his forte. Just because he is a genius when it comes to fundamentals doesn’t mean he knows what he is talking about when it comes to Technical Analysis.
Steve
. . . Like!
…
When I got my MBA way back when, I took Damodaran’s valuation class. My views on investing are heavily influenced by his.
I think the best way to use TA with P123 is narrow down a set of stocks that passed through your buy rules. For example, you have a 20 stock port. Your sell rule sells off 5 stocks at rebalance. Your backtested buy rule of Rank > 95 returns 25 potential stock candidates to fill the 5 new port slots. Maybe use TA to single out the best stocks with the best chart rather than just blindly selecting the 5 highest rank stock over a marginal difference in rank from the rest.
Fundamentals and technicals support and enhance one another.
Fundamentals tell you what the stock “should have been” doing (theory).
Technicals tell you what the stock “actually has been” doing (actuality).
The future can never be more than an exercise in assumptions and probabilities. If you want to go with the flow of things, to fly with a tailwind, you help yourself when theory is aligned with actuality. If you want to play the contrarian game, misalignments between theory and actuality cue you to look deeper to see if there’s an opportunity.
