i used to spit on technical analysts, being a loyal CFA and MBA and all, but the more i read about this stuff, the more it grows on me. fundamental analysis is pretty arrogant in the sense of ‘i KNOW more than the market, and damn the mob, my view will prevail’. while technical analysis seems more humble in the sense of ‘i don’t have a freakin clue why this thing just went up, but i’m buyin’ cos the chart says so’.
I don’t practice TA in any depth, but I definitely think it can work in certain assets or security types, and I think it also can work in certain market conditions (now being one of them). Bad TA is what Benoit Mandelbrot called “financial astrology” and boils down to just pattern recognition, but good TA is connected to theories of behavior of market actors. I think now is an environment where TA is likely to work because perceptions have changed the fundamentals which in turn change perceptions: a feedback loop George Soros might call Reflexivity. When that happens, fundamentals become very difficult to establish and so investment choices are more about fear and greed than fundamental value.
I could definitely see logical technical analysis working in an environment like this. I remember sitting in on a talk from one of the Maverick Capital guys, and he was like “I may not make my final trade decision on technicals, but I absolutely want to know what technical analysis would said to do because there are definitely people trading on that and it could move the share price in the short term.” Also, anecdotally, I think there are a lot of people looking at technicals right now. We recently had a sell side strategist in and he said that their “Head of Technical Analysis” (if you would have asked me a year ago I would have guessed this job no longer existed) was getting tons of calls.
i admit i use/d TA. but I also created a monte carlo simulation with lognornal dist and played around with the std dev in excel and charted several scenarios… now i question my charting ability… i suggest anyyone that uses TA do the same.
Would even the most devout fundamental analyst issue a stock recommendation without looking at a chart of the stock price first?
i use both and both have worked wonders. nothing wrong with working on a model for a week then having some chart analysis to bolster a buy or sell reco. Most people who hates charts cant pick a stock at all. Trust me, everyone looks at charts. and if you think only fundies matter, look at bill miller last year. damn dude cut his assets by 75%
You can’t use TA on random data. The whole point of TA is that the behavior of traders is moving the prices. ConvertArb Wrote: ------------------------------------------------------- > i admit i use/d TA. but I also created a monte > carlo simulation with lognornal dist and played > around with the std dev in excel and charted > several scenarios… now i question my charting > ability… i suggest anyyone that uses TA do > the same.
^^^^ I think the point is that if you have randomly generated data and try to do TA on it, you might convince yourself it works, even though the underlying process is in fact random.
i don’t think there will be the the same number of technical signals from random data. technical moves depend on behavioral movements. they are predictable to some extent. thats why people make money on them.
do you have your CMT, rohufish?
no - i’m done with taking exams for now. anyone here have a CMT? do you feel the depth of the knowledge from that program helps you make money in the markets? or is a single good book on TA enough to get 80% of the gains you’d get anyway?
rohufish Wrote: ------------------------------------------------------- > i don’t think there will be the the same number of > technical signals from random data. > > technical moves depend on behavioral movements. > they are predictable to some extent. thats why > people make money on them. i didn’t think so either… but there was… you should try it. below is sample (estimated annualized vol at 250 trading days and did not enter in daily RFR for stock price…) Column A enter “Trial” and number all the way down 1 to 5000 Column B enter “Normal RV” & formula =NORMINV(C2,mean,sigma/(250^0.5)) Column C enter “Rand” & formula=RAND() Column D enter “% daily change” and formula =(B2-0.5) Column E enter “Stock Price” and I started with $50 then change by daily change thus forumla =(1+D2)*E1 Now Label Mean and put in value of 0.50 & sigma will be your annual std dev (I play around with this one put in 0.30) chart the output from Stock Price and hit F9 to generate a new chart
I bought all the books for all three levels of the CMT, however I’ve decided that the designation is totally unnecessary and never took an exam despite planning for all of them. The problem is that the designation requires you to know all aspects of TA, “most” of which don’t work since they are unnecessarily complex. The CMT is good for an almost encyclopedic knowledge of all known TA, but is totally unnecessary for the main goal: to make money. When it all comes down to it, many indicators and analysis techniques are derivatives of price action and it seems that the parts of TA that work are based on simple things such as: trendlines, support, resistance, volume, and a few very simple patterns. When you’re using all types of fancy indicators, it obscures the simplicity of price action. You focus so much on the structure (indicators) that you created that you might totally miss what’s actually happening. Much of the material that you’ll learn for the CMT and TA in general are often “descriptive” and NOT “predictive” in nature, these things are often confused with each other. Heck, I was talking with a PhD in a quant field the other day who was using research on power laws describing the severity of wars with the frequency in which they happen to make the case that we’re due for another big war. I told him he was using descriptive statistics incorrectly in a predictive manner. Here’s an example of why I think there is such a huge subjective component that obscures simple price action. Consider moving average crossovers in which you look at two moving averages - there is a very large amount for subjective input. For example, should you use: - simple vs. exponential vs. weighted? (how much should the decay or weighting factor be?) - how many days should the “fast”/“slow” moving average be? 10/20? 5/25? 13/33? a/b? It’s taking something that’s totally arbitrary and imbuing it with a false amount of precision. Furthermore, how do you choose these parameters for: - different timescales (surely 10/20, 5/25, 13/33, a/b above have different meaning when applied to periods of 1 min, 1 hr, 1 day, 1 week, 1 month, etc.) - different asset classes (the ideas of TA may work, but those specific parameters might not mean the same thing for lean pork bellies and google) - bear market vs. bull market People who use the fancier indicators of TA unquestioningly may not consider these things and just act simplistically such as: buy(sell) when the fast average crosses above(below) the slow average. This may work for a little while, but what happens when you change the circumstances? Those who are initially successful using a ridiculously complicated set of technical indicators might, at the core, really be successful due to simple things: - price action - sound risk management - emotional control Anyways, that’s my rough take on why the CMT is unnecessary - though I do believe in the power of TA.
Nice post sublimity.
Anyways, I think Pring’s “Technical Analysis Explained” is really good (it’s in the high hundreds of pages).
I have Murphy’s book. What do you think of it?
i am just reading up the ‘chartschool’ section of the stockcharts.com website. its john murphy’s website. very good coverage of the basic stuff sublimity refers to. i agree, the more complex you make it, the more you infer false meaning into something that is ultimately very simple or meaningless. to work, this stuff needs to be acted upon by thousands of market actors. only handful of them will act on some fancy schmancy new indicator. the vast majority of them will just look at price action, basic patterns, etc. it makes sense that the basic stuff is what really matters.
I like that Murphy book too, it’s considered a classic after all. I was mainly addressing rohufish’s remark about “a single good book on TA enough to get 80% of the gains you’d get anyway”, for that, I think Pring’s is nicer and more comprehensive.
It seems you have to balance optimality (simplicity that works across a lot of circumstances) with maximality (getting every last cent in a specific circumstance but with risk that it may totally fail).
Good thoughts, sublimity, but if you believe in the power of TA, as you put it, then clearly mastering the subject through a recognized program, like the MTA’s CMT, shoulod make sense. The questions you raised on the predicament of choosing parameters for the MA crossover may really indicate that doing so is not trivial, not that it is nonesense. In all fairness, the same set of issues may be cast on fundamental analysis… how many financial ratios must one know? How deep and hard should you look into accounts receivables to figure out any cash problems, or how much analysis should you expend in the statement of cash flows to glean any useful data? Doesn’t all of that obscure the simplicity of fundamental analysis as a look at sales growth and profitability?