Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
The best book I have ever read is by David Aronson called Evidence Based Technical Analysis (EBTA). Essentially he debunks the majority of technical analysis and chart pattern trading as being worse than bad and provides a plethora of evidence to back it up.
Classic chart patterns, Fibonacci and Elliott Waves are blown up right in front of you eyes and as a 'non-believer' in the first place, this just cemented what I have known for years.
The shining light of this book is that he does define methods for data mining that are evidence based and could provide you with positive results if you follow them. At the end of the day, you can either keep on believing technical analysis works, or you can read his book and find out how to potentially create something that does work.
--------------------------------------------------------
- Trade what you see. Invest in what you believe -
--------------------------------------------------------
EBTA is more of a great concept, maybe because my "free" kindle edition is hard to read, but on the other hand, the message is pretty ,much delivered in the first paragraph or so - ie. almost all technical stuff people use is idiotic.
My personal opinion is that his stuff, despite being more intelligent, is about as flawed as the rest of the industry.
The main reason is that the individual finance industry stopped developing at a certain point to allow commercial products a space to peddle their fish. Commercial backtesting, for example, relies on a trade listing. From a development point of view, the trade listing is a reasonable intermediate step in prototype development but not a good place to stop. Among the issues arising from this, is that the trade listing generally calculates returns on a fixed investment amount which is wrong - returns should be expressed as natural logarithms.
Aronson's book does cover many of the aspects to do with data mining and back testing yourself. It does get heavy in places and takes a few reads, but once you wrap your head around what he is trying to get across, it is fascinating - but best of all, his testing can be replicated. So, I did.
As an example, I have used his method(s) to debunk some very common technical indicators. It took me a while to data mine and put together (plus this is an ongoing exercise) but essentially this is my list (so far) of indicators that I have found to have zero (most are beyond zero) statistical edge in using them. I.E. they are completely useless.
Stochastics
Price levels (round numbers etc.)
ADX
CCI
Diagonal trendlines (I couldn't believe it either...and I still have tendency to use them)
Japanese candlestick patterns
Chart patterns (non EBTA)
Fibonacci
Bollinger bands
MA Crossovers (not completely useless because I discovered a different way to apply them that could be quite profitable.).
RSI
It is a punch in the gut to the T.A. believers no matter what evidence is presented to them. They still want to believe in it and I guarantee there are many traders out there using some of the indicators I have listed.
I continue to study new ones as I find the time. Next up is traditional Elliott Waves, not Quantified Elliott Waves (which actually works for medium to long term forecasting).
--------------------------------------------------------
- Trade what you see. Invest in what you believe -
--------------------------------------------------------
I've been seriously looking at backtesting recently. Like the name suggests, it seems to be set up backwards. Instead of setting something up and then checking to see if it worked in the past, why not first look at what worked in the past and then set something up?
There are several serious issues with using academic statistics, not least of which is the typical annoying statistician personality type. Quantitative analysis has little in common with normal statistical applications. In any case, both disciplines are OK for demonstrating the idiocy of the implementation of the vast majority of technical techniques. Non Statistical Quantitative analysis can allow one to work at a high level without the baggage of a statistical guilt trip.
There are three major rational ways to analyze price in relation to time: Rate of change, average, and range.
Stochastics (from the list in the post above) is a highly dubious interpretation of range. One can legitimately label stochastics silly - (why use exponential averages, who cares about the signal line, are 20 and 80 really magic numbers) but that just suggests that one is not analyzing range properly.
A Monte-Carlo test is the best way to test a null hypothesis. You can determine if a trading system or indicator truly matches the position of raw returns as opposed to being intelligent. But this can't be done all of the time because not all of the information is available. Hence, why the data mining part is imperative.
As for the Stochastics, they were invented in the 1950s, are based on overbought / oversold which in futures trading doesn't exist, the fast or slow mechanic is terrible as it gives many false signals even when price is range bound and when price trends, traders get killed.
You can test this hypotheses on any numbers you want in the Stochastics, 20 and 80 are just the default. It is a terrible indicator that has no place on a chart.
--------------------------------------------------------
- Trade what you see. Invest in what you believe -
--------------------------------------------------------
I bought Bandy's book on Monte Carlo simulations and was quite disappointed, not by the book quality but because they seem to be a waste of time and the book wasn't cheap. Even the name is weird; I prefer Atlantic City simulation because it is tackier.
Statisticians are good at telling other people their ideas are worthless; but they are not so good at focusing that insight onto themselves.
--------------------------------------------------------
- Trade what you see. Invest in what you believe -
--------------------------------------------------------
When I looked at the setup instructions carefully for an Atlantic City simulation, I realized my work already incorporated whatever it is supposed to do.
A basic problem with statistics seems to be that there isn't an addressing methodology to catalog information. Not easy to explain simply - sort of a data rather than a mathematical issue. This problem is obvious if one looks at a quantitative finance text; there are tons of fascinating studies but they are generally too complex to be easily recallable at a later time.
For issues with a simpler example, take the standard deviation. That has great value as an example of data manipulation using Euler's number, but the standard deviation value is not very useful when applied to price movements. The information that standard deviation imparts can be derived more easily by other methods.
Other than the addressing methodology (sort of related to abstractionation - if that is a word) the main theoretical problem with statistics is time. Anything statistics does with time is questionable - standard deviation relies on taking the square root of time for example. Tell me how that ends well.