# Is there any good research on support and resistance?

Could somebody advise me on where to find good literature on the justification or motivation for using support and resistance lines - and also lines of maximums and minimums. In finance, it is sometimes difficult to find the relevant literature because even if a book does not claim to be able to make you rich, the content of the book can still be bad.

Any good literature on support/resistance would be nice.

-
I would like to see empirical evidence of support and resistance. I see it all the time, but I'm not sure if it's a mirage or not. –  James Roth Mar 2 '11 at 17:32
This question is now being discussed on meta. –  chrisaycock Mar 3 '11 at 0:25
@chrisaycock, I posted an answer there. –  Ilya Mar 3 '11 at 9:40

Contrary to popular belief, there does exist some truly high quality academic literature on this topic. The most sophisticated and well executed paper in this regard is Lo, Mamaysky, and Wang (2000). They write:

In this paper, we propose a systematic and automatic approach to technical pattern recognition using nonparametric kernel regression, and we apply this method to a large number of U.S. stocks from 1962 to 1996 to evaluate the effectiveness of technical analysis. By comparing the unconditional empirical distribution of daily stock returns to the conditional distribution—conditioned on specific technical indicators such as head-and-shoulders or double-bottoms—we find that over the 31-year sample period, several technical indicators do provide incremental information and may have some practical value.

and

We find that certain technical patterns, when applied to many stocks over many time periods, do provide incremental information, especially for Nasdaq stocks. Although this does not necessarily imply that technical analysis can be used to generate “excess” trading profits, it does raise the possibility that technical analysis can add value to the investment process.

Moreover, our methods suggest that technical analysis can be improved by using automated algorithms such as ours and that traditional patterns such as head-and-shoulders and rectangles, although sometimes effective, need not be optimal. In particular, it may be possible to determine “optimal patterns” for detecting certain types of phenomena in financial time series, for example, an optimal shape for detecting stochastic volatility or changes in regime. Moreover, patterns that are optimal for detecting statistical anomalies need not be optimal for trading profits, and vice versa. Such considerations may lead to an entirely new branch of technical analysis, one based on selecting pattern-recognition algorithms to optimize specific objective functions.

A search for this paper also reveals other potentially valuable academic articles on technical analysis and chart patterns, such as this list.

-

Below, I see a lot of support and resistance. Here's the code:

x <- cumsum(rnorm(1000))
plot(x, type="l", main="Support and Resistance")


Edit (03/03/2011) ================================================

Gortaur, I put my answer here to avoid filling up the comment area.

Your question 1) "......I was not asking for the "garbage" literature, I can find it in net by myself........"

The reason that I posted that article is because it, and a few others, are consistently referenced as "credible proof" that support and resistance, as predicted by certain groups, is a profitable strategy. As far as I can tell, "credible" comes from the fact that the author worked at FRBNY, not her results. "Proof" comes from those that accept her results, not from repeating her results. If you pick at that article long enough, the term "garbage" might just come to mind.

Don't get me wrong. Unlike others, I don't think the author was trying to mislead anyone. I just think that she didn't chew on her methods well enough.

Your question 2) ".......the fact that S&R appear in random walks and the human's eye try to catch it everywhere do not prove the fact that these notions are meaningless in real life where prices are NOT random walks and random walks are just model to simplify dynamics and don't care about the game theory part of the question (because the system is too huge)........"

If you run that (or similar) code over and over, and count the plots without "support" and/or "resistance", you might not get to 1. If it's that close to 100% of the time, how can you tell the difference between S&R that is random and S&R that isn't? And, if both the random series and the real series "bounce" off of some arbitrary lines, does it really have to mean something (more importantly, in the random case, you know it doesn't mean anything)? My point is really simple. By the time you devise a test that means something, and then run real data through it, you might find that you can't tell when random data has been passed through it. If so, then you're back to square one.

You don't have to believe me, try it yourself. As a place to start, go to .pdf page 8 in that article and look at the test (Calculating Artificial S&R Levels). Draw what is described, for a few days worth of data. Then ask yourself, could you come up with some S&R lines that will beat her test? If so, does her test mean anything? And, if you use your method of beating her test as your test, can it provide a way for you to differentiate random versus non-random results?

Edit (03/04/2011) ==========================================

"....If you just use logics - then what are you trying to say? A = {for the random data there is smth which looks like S&R}. B = {S&R can be used to analyze random data}. A->B = false -> what is looking like S&R in random data is not S&R (at least like a tool for analysis). On the other hand, you are trying to use this fact to prove that there is no S&R (like a tool for analysis) for the non-random data. I don't think that it is right....."

I'm not sure that I understand your questions, but here goes.

By using S&R levels, aren't you trying to react to something that the market is actually doing, something meaningful? For instance....a bunch of people/money want to buy at support and sell at resistance? The idea is that something meaningful and non-random is happening at areas of S&R.

The problem is, a random process will consistently generate lots and lots of S&R levels, and you can be 100% sure that those S&R levels mean absolutely nothing. Think about it, how can a random process NOT turn and go the other way? You can calculate the odds.

So, when you find an S&R level in real data, do you have a way to identify that this is a "random" versus "non-random" S&R level? How about a way to calculate the odds that this is a random versus non-random S&R level? If you can't get past this issue then the rest of your analysis will at least partially (maybe completely) be based on meaningless random events.

Obviously, a lot of people have decided that they can't differentiate random versus non-random S&R levels with price alone, so they add volume, seasonality, sun-spots, or whatever. Lots of ink has been wasted on this subject.

The bottom line is, whatever tool you develop to identify and use S&R levels, when it's all finished, run the above random "x" through it and see if your answers are any better or worse. If the performance isn't statistically different when you run random data through it, is the tool doing anything worthwhile?

-
I guess the moral of the story is that totally random stock prices appear to have a pattern. So what seems to be a trading strategy is actually just noise. –  chrisaycock Mar 3 '11 at 0:27
@chris: I wasn't trying to be a jerk by giving a short answer. It's just that that is the answer. Anyone that runs the above, or similar code over and over will find that it is rare to NOT find "support" and "resistance". They can prove it to themselves, regardless of what I say. I don't think this is off-topic, I think it is reality poking through. –  bill_080 Mar 3 '11 at 1:32
@bill_080, thank you for the example - but 1) I was not asking for the "garbage" literature, I can find it in net by myself. 2) the fact that S&R appear in random walks and the human's eye try to catch it everywhere do not prove the fact that these notions are meaningless in real life where prices are NOT random walks and random walks are just model to simplify dynamics and don't care about the game theory part of the question (because the system is too huge). –  Ilya Mar 3 '11 at 9:48
@everyone who upvoted this answer: the question is a query for references, not a query on everyone's opinion on the validity of S&R. –  Pete Mar 4 '11 at 0:02
If you just use logics - then what are you trying to say? A = {for the random data there is smth which looks like S&R}. B = {S&R can be used to analyze random data}. A->B = false -> what is looking like S&R in random data is not S&R (at least like a tool for analysis). On the other hand, you are trying to use this fact to prove that there is no S&R (like a tool for analysis) for the non-random data. I don't think that it is right. –  Ilya Mar 4 '11 at 10:30