Are technical indicators of any use?
Author: mike5
Creation Date: 1/19/2009 7:10 PM
profile picture

mike5

#1
I'm not a programmer but I've been using WL for about 18 months now. My first endeavor with posting a question to the community was 3 weeks ago. Since then I've been reading old postings trying to learn from others questions. Almost 3 years ago I saw this comment from DrKoch:

"You can find lot of explanations like the following in the literature:

Over the past 30 years we have programmed just about every technical indicator, time frame, event, or trading theory; on just about every commodity contract, and on most of the active stocks trading on the NYSE or NASDAQ. In all of those years, we have only found one or two things that can accurately predict what the markets will do in the future and give us an edge with our trading. So much so that now all of our research is based on those two things first and foremost.

(from Hank, www.programtrading.com)

There are several books that researched TA Indicaors. While some claim they found something, it is mostly errors in their procedures if you look closely:

* no account for risk
* no account for survivorship bias
* bad testing methodology (InSample/OutOfSample, etc)

Finally I did similar work for a few month. Used finest software and advanced statistical methods to find the most spurious "edge" from TA indicators.

Result: Nothing.

[Edited by DrKoch on 2/22/2006 2:44 AM]"

I greatly respect the views of the people who post on here, if for no other reason than they are light years ahead of me. So I have to ask myself, Is that basically telling me that I should give up trying to create a mechanical trading system unless I get into Neural Networks or AI?
profile picture

EMan99

#2
The problem with indicators are they are lagging, they are telling you what you already know the price direction, it is moving up, down or sideways.
profile picture

brucev

#3
Hi Mike,
I don't want to rain on your parade, but if you are having difficulty creating trading systems that work, then AI and ANN certainly won't help you. Consider those tools to be a "step 2" - that is, move onto them when you are creating successful systems.
Cheers,
BruceV
profile picture

DrKoch

#4
QUOTE:
should give up trying to create a mechanical trading system unless I get into Neural Networks or AI

Short answer: Yes
--------------------------------------------------------------------------------------
Long answer:
Let's compare trading to roulette.

There are two types of people when it comes to roulette:

1. The "scientific type" who knows that odds are 18/37 (or slightly less than 50%) to win a long sequence of roulette games. So these people will never put a chip on the roulette table.

2. "Gamblers". They play the game because they want to play. No amount of "rational" discussion can stop them. Some of them develop hefty "systems" write books or practice other sorts of pseudo science to explain why they (and everybody else) should get rich with roulette.

Starting a discussion like this usually starts a religious war between these two groups.
BTW: I think trading based on TA Indicators is no better than playing roulette - you have a chance slightly less than 50%.

-----------------------------------------------------------------------------------------

Open the stage for "The Predictors". (Its a book title)
These guys were very knowledgeable in electronics, math, physics and computer science. It took them several years to find the formulas which describe roulette with a slightly tilted drum, to develop hardware and software which could predict the outcome of a roulette game (inside a casino, during a life game without noticing anybody the hidden computers in their shoes.
They found an edge in roulette (but failed to make money out of it).

I found out that a similar thing exists in trading. If you're good enough in programming, math, physics, statistics, general scientific thinking, economics, ...

... then you may be able to develop a method which is able to make consistent profits in the long run...

And as I said before (sorry, cone): You'll need tools "better" than WL to walk this path...

profile picture

Cone

#5
Let the war begin!
profile picture

warren631

#6
According to Mike5, DrKoch said "we have only found one or two things that can accurately predict what the markets will do in the future and give us an edge with our trading".

Well? What were they?
profile picture

Roland

#7

Let the war, no, let the games begin.

I have to agree more with René’s approach. He is right, you do have to dig deep to find worthwhile gems that can help in designing valuable trading scripts.

However, the game is really biased. There is too much cheating! “They” cheat before an IPO, during and after; they will even cheat after hours. They will front run their clients, overcharge them and lie in their face when selling their worst fund simply for their own higher commissions. They will cook the books, create off balance sheet entities to hide debt and even pre-dated option grants in a big way. They will vote them selves outlandish severance pay, design their own golden parachute and will go as far as having upfront compensation for a non competitive clause after their death. They will look at their pocket first even if they have to bend the rules more than a little. They’ll mount Ponzi schemes, ephemeral markets for securities (selling tradable securities in the billions) and then closing the market.

I have not seen a single area of the whole financial industry where there was no cheating. And it is in this environment that we have to design profitable trading scripts.

The only solution is to adapt to this kind of market should you still want to play the game. And this would mean develop trading strategies that can take into account the fact that the stock you selected might go down and protect yourself before anything else. Never trust what a CEO might or might not say. Even Lehman’s CEO was predicting “no problems” two days before declaring bankruptcy. RefCo, less than a month after its IPO went out of business. Parmalat, going under for 18 billion, with a single monthly account statement serving as cornerstone for the empire. I don’t need to enumerate all the schemes, you were there too.

Learn from the past, you have paid your dues.

profile picture

mike5

#8
To Brucev, I'm not considering trying to get more complicated with my strategies. I just wanted to know if I was wasting my time, using WL 5, because there aren't any technical indicators that really work. I don't want to curve fit something to past data, only to throw my money away in the present. If those of you who are much more knowledgeable and experienced than I, can't find profitable trading systems through technical indicators, then I can spend my time elsewhere.
profile picture

brucev

#9
Hi Mike,

I don't really subscribe to the theory that technical indicators "don't work". Same goes for fundamentals. Same goes for quant work, or portfolio theory, or whatever your poison is. I have even seen working trading systems built from financial astrology (there are even tools like Wealth-Lab to do this - where you can design indicators based on planetary movements or geomagnetic radiation - see www.wave59.com if you don't believe me!). Also, there are a number of published "phases of the moon" style trading systems from the likes of Dr. Jeffery Katz, which have withstood great criticism in the academic world. Not that I'm suggesting that you should follow financial astrology, I am simply saying there are a great many paths, and, in the long run, if you decide technicals aren't for you, then thats cool!

Technical indicators do what they do, and its your job to design strategies that can try and exploit their strengths and avoid their weaknesses. Every indicator has both. Simply using the same indicator for every task is just like using the same tool in the toolbox for every task. It just makes no sense at all. But saying that a particular tool doesn't work because you can't use it to do everything also makes no sense to me. Despite my many years of research, I am yet to see a "sensible" test of an indicator - ie. one that tries to seperate the "correct" usage of the indicator from "all" usage of the indicator. In this sense, I agree with Dr.Koch that much of the "book" research being published uses poor methodologies.

However, from what I see in the WL forums, there appear to be a few people who are developing successful systems using technicals. So you either need to question your definition of "successful" systems, or change it!

If you want to pursue developing technical trading systems, then testing every published script and data mining every indicator is totally pointless, and worse than that, it saps your time away. If you want to use technical indicators, take the time to learn about specific indicators (it doesn't really matter which ones), learn their strengths and weaknesses, and then build up your own skills by trying to determine how to capture their "elusive" returns. There are no shortcuts. A friend of mine is a karate teacher, and one of his sayings is that the difference between a black belt and a white belt is that the black belt didn't quit. However, you do need education to succeed, and most of the books that seem to get published in this area are worse than useless. If you decide you want to develop technical trading systems, start looking at very few indicators, and start reading academic research. You may be suprised to know that there are a great number of people who have skills that you may not have, who are happily publishing the information you need to progress. I agree with DrKoch's comments about the abilities needed to develop trading systems. But there are no "renaissance men" any more - the world is a bit too complex for that now. Just because you don't have skills in a specific discipline, don't rule yourself out. Either start taking classes and building up your knowledge, or start reading published academic research which helps solve problems you can't.

Cheers, and good trading,
BruceV
profile picture

Roland

#10

Hi Bruce,

I agree with you, trading strategies that want to fit every stock in the same box should not work, theoretically. The problem should be more complex than using a one size fits all. And as you said, many use flawed methodologies. The game is not a 50/50 heads or tails game; long term it is biased to the upside. But in times like these when your portfolio value is only going down, some can express doubts.

But even with what I said in my previous post about all the cheating, I think as you do that someone can still design profitable scripts. First and foremost, protection has to be the rule (meaning using stop losses). Incremental betting should be put in place up to a relatively low percentage of the portfolio. Scenarios requiring 100% equity in a single position, even by incremental betting, should not be considered as they can often be the final frontier before portfolio oblivion. The prime directive here is to survive the game. And in this game, to reach the black belt trader level, you need capital just to play the game.

Position sizing procedures should adapt to an unknown future, reducing positions in times of downside turmoil and then when advancing increment with a protective stop. It is not in predicting the future that you can win - just look around at all that have tried – it is by adapting to a changing environment and adjusting your bets accordingly.

Time slicing positions, as most of the scripts here do, can only tend to an overall average performance; a lot of work to do no better than the Buy & Hold.

When both the price and the quantity on hand evolve at compounded rates, one can easily outperform the markets. Research should put more emphasis on position sizing in uncertain trading environments.

Bruce, I’ve read over 120 of the research papers you mention over the last six months in hope of finding someone, somewhere to contradict what I was doing. But all stopped before or at what I call figure one (see my paper ) unable to extend further as if all anyone could do was achieve average market performance. It is as if equations written 40 years ago were the ultimate achievement of humankind. I disagree on this; we can do better, much better.

profile picture

futurewag

#11
I agree, what is meant by "Don't work"? Indicators are a tool to aid the decision making process. Taking both fundamental and technical analysis into account will help, and as was stated earlier, we're just along for the ride. Imagine you're sailing the seas, you have waves, direction, and weather. If you have tools and knowledge you'll survive, otherwise your sailing into the unknown. I still loose my butt from being impulsive... take your hand off of the mouse Scott.

Scott
profile picture

mike5

#12
Thanks for all of the input. I've been out of town and I decided to just plain stay away from the markets until I got back. I appreciate the willingness of all of you to offer suggestions and insight. Trading can be a very difficult business at times, as you all know, and I've been getting my butt kicked lately. So I took some time off to get my head in the right place. I'm back ...

Thanks again, Mike
profile picture

futurewag

#13
Mike,
Go to Stockcharts.com and learn what you can there. The main message is to keep it simple. Choose to work with Moving Averages, directional indicators like the ADX, and an oscillator like RSI. Money management is also key. Cut your loses at 7% or 8%, hanging on hoping to get it back will likely fail, and you'll end up down 30%. In a down market ETF's can be useful because there are funds that short the indexs. Today you need to know how to trade, if anything the trend is still down. At best the market is moving sideways, a time when the RSI would be useful. I never programed before November and spent most of December writing scripts in WL4. I finally got to the point where the drag and drop was more of an inconvenience than writing out what I wanted. Because I use a x64 system I've just started to learn WL5. I'm kind of obsessive so 18 hour days over and over isn't unusual for me. My son was trying to get me to do something else, and I complain about his video games. Fine one to talk, anyhow the point is, you can win in this market if you work hard.

Good luck,
Scott
profile picture

DrKoch

#14
QUOTE:
Choose to work with Moving Averages, directional indicators like the ADX, and an oscillator like RSI.

A useless advice. None of these will give you an edge.
QUOTE:
Money management is also key.

True. But without a positive expected profit no money management will help.
QUOTE:
Cut your loses at 7% or 8%

It is easy to show that with most trading strategies a stop loss will lower expected profit. Certainly all books will tell you the contrary, but in fact most stop losses hurt performance.
QUOTE:
a time when the RSI would be useful

unproven. The RSI worked roughly until 1998 in the US stock market. No edge (usable predictability) since then.

profile picture

brucev

#15
From DrKoch:
"It is easy to show that with most trading strategies a stop loss will lower expected profit. Certainly all books will tell you the contrary, but in fact most stop losses hurt performance."

Having tested vast combinations of stop loss methodologies on many, many systems, I can confirm that I have yet to find a system which can be significantly improved using stops.

Cheers,
Bruce
profile picture

futurewag

#16
I'm not pretending to have the knowledge of trading or programming that you do DrKoch, but if I were overwhelmed with the current market conditions and losing money I'd go to cash, and then spend time learning the basics. Until you either have the time watch the market closely or an automated system that works protecting your principle with stops helps. Yes I agree when building and testing systems I get better results without stops, but I don't have the confidence in myself that my system would work, nor do I have a level of understanding with design that I desire. It sounds to me that Mike hasn't the experience and needs a starting place. I've read a great deal of what you've written over the years and consider you to be one of the experts. Frankly much of the math is over my head, this is one area I need to study, conceptually I do ok. Currently I use my discretion to trade not a WL system. I just don't know enough, but I'm working at it.


So now that you've taken the edge off <g>, what would you suggest as a path of learning?

As I was telling my Mom last night "technical indicators help convince us that our guesses are right".
profile picture

brucev

#17
Hi,

although your reply was addressed to Rene, I will comment anyway.

Futurewag, you are still missing the point.

"Until you either have the time watch the market closely or an automated system that works protecting your principle with stops helps." - NO IT DOESN'T

"but I don't have the confidence in myself that my system would work, nor do I have a level of understanding with design that I desire. " - THEN YOU KNOW WHAT TO DO NEXT

Regards,
Bruce
profile picture

futurewag

#18
Bruce, if I have $10 and trade it and the market takes 50% before I get out of the trade I have $5.00. If I had put a stoploss on it, gone off to work, come home to find that Citi bank had tanked, I'd only be out 80 cents. Of course maybe I was stopped out and Citi turns and goes up, but if I had the habit of placing poorly planned trades and losing, I'd suggest stops. Fair?

My prefrence would be to learn from what your writing. Provide me with your logic or refer to something I can study to better understand. I have run scripts written by others that in some cases use short durations, 1 to 2 days, and lower percentages to produce good profits. I did go into some of the scripts and add stoplosses and produced the result that you've shouted out. If your final comment confirms that I need study, yes I do. If you'd like to offer what has worked for you, or suggest some scripts that you've found successful, I'd appriciate the input.
Thanks,
Scott
profile picture

Cone

#19
If you use 100% sizing for a trade, then for sure you need a stop since one trade can wipe you out. However, no reasonable person puts everything on the line for 1 trade. For every person that does it and actually makes a fortune, there are thousands more who will just be wiped out.

The point is that those of us who have been doing this for a while have tried to improve backtest performance by implementing stops losses. In a great majority (nearly all) cases, where reasonable sizing is used, adding the stop losses (and even profit targets) will reduce performance.
profile picture

brucev

#20
Scott,

if your stock goes down 50%, but you were stopped out after a 20% fall, then you are indeed better off. On an individual trade basis, the conclusions are clear.

However, you are assuming that the "beneficial" effect of the stop also works at a portfolio-level. If you have a positive expectation strategy, then this assumption is false. I recently published a practitioners paper to this effect with JASSA - I suspect you can get a copy on the web if you search about for it. Its not up on my uni research page yet, probably be up there in a few weeks. In any event, it will confirm this conclusion. It also demonstrates the technique to use to determine whether the stops are helping or not, so you can do it with your own trading strategy (assuming you have access to a stats package like SPSS).

Regards,
Bruce
profile picture

warren631

#21
How about this script based on classic technical indicators (developed for my 401K):

78.95% profitable trades (sorry not 100% yet).
Long only. Daily. S&P, ETF's, QQQQ, large mutual funds.
$598,000 profit vs. B&H -$590,000 loss.

Download screen shots of my WL at:

http://www.4shared.com/dir/12439182/5d6ff779/sharing.html
profile picture

futurewag

#22
Thanks Bruce and Warren, I'll look into both of the suggestions. I'm not familiar with JASSA or SPSS but I'm sure I can find info on the web.

Bruce, I held money in 401k's and currently in an IRA for years. I would buy and hold. I didn't start trading until mind Janurary of last year. Only 3 months ago did I start working with WL4 and since then I've been introduced to a lot new information. Before WL4 almost every book I'd read recommended stops either based on ATR or at 7 or 8%. I don't use stops. I found that to often the big players would drive the price down take me out and then the price would advance leaving me behind. I ended 08' -4.3%.

Yes Cone, I did play with stops and targets and found the same thing. What I'm doing now is looking at scripts others have written and trying to understand the code language and structure. I can write codes based on indicators with cross overs and such without a problem, and even some with descent results. I was starting to get into symbol rotation when I finally got a copy of WL5, and now, its back to the basics with .NET.

Thanks,
Scott

profile picture

futurewag

#23
Warren, you have what looks like a stochastic, and also MACD Float. What is the MACD float? How does it differ from a typical MACD?

Scott
profile picture

futurewag

#24
Bruce, I found the SPSS information, its been over 20 years since I took stat's, I was an art major. I did very well at math however, I just haven't applied it, only now do I see the need.

What is JASSA? I wasn't able to find anything.

Scott
profile picture

brucev

#25
Hi Scott,

JASSA is a Journal of Applied Finance. Type "JASSA" into google for the link.

Cheers,
Bruce
profile picture

futurewag

#26
Thanks Bruce,
I did look at your web link. I plan to read your work, it appears to be rather advanced, but I'll take a crack at it. You put neuro networks to considerable use, I haven't purchased the software and it isn't part of Fidelitys WL4 Pro. Nor do I know if it is intended to be implimented in WL5.

Thanks for your input with respect to this forum,
Scott

BTW I did try JASSA before asking for clarity and found various odd links, nothing that related to finance however.

profile picture

mikael

#27
I'll add my 5 cents worth. It strikes me that there are a number of entry/exit signal combinations which have historically, however you measure that, delivered a better than 50/50 ratio. The best in my arsenal, with a reasonable standard error, has a 60:40 win/loss ratio, but I still look for the holy grail. This example appears only every couple of months, and when it does it has a gain/loss ratio of 12:7. This isn't what some people claim but traders that I have lunch with reckon it's pretty good. What I will say about this pattern and the others I've found is that they fly in the face of all the literature on the approach it uses. As far as I can tell there are zealots out there who BELIEVE in whatever system. Here we get to do our own research and find out for ourselves, backtesting, backtesting, backtesting.
profile picture

mikesblack

#28
In my opinion, I believe that price or volume using standard indicators are useful in that they highlight a particulair feature of the data , but they don't forecast very well in and of themselves( They might, but I can't find any). That said, comparisons or combinations of these data transforms with some indicator or indicators that aren't generally available or realized might have relevance as leaders.

That's the interesting, but hard part. RSIs and moving averages are there for the masses to crunch, but not everyone will measure those obscure variables. It's most likely these variables are quite simple and or intuitive, but only after the discovery. (haven't tested the sunspot theory, but I suspect it may be a red herring.)


The fun part is in thinking about what might be good leading indicators and then testing their validity. Mass human behavior is complex, but in certain situations can also be remarkably simple. Comparing technical analysis in terms of roulette or any other game with fixed odds however is fundamentally wrong. . * Ludic Falicy "the misuse of games to model real-life situations". from by Nassim Taleb -"The Black Swan".

Can I tell what these indicators may be, or if I have found anything? Only my wife might know, but she couldn't care less, provided I don't lose us money.


Good luck, and please keep your discovery to yourself or else lose or diminish its forecasting ability.

profile picture

DartboardTrader

#29
A few bloggers pointed out large increases in liquidity provider volume, and lower overall volume..
mikesblack, have your price + volume indicators been working lately?
profile picture

zwoop

#30
I dont get the discussion regarding no stops. So lets say i alocate x% om my bankroll for a trade and then after that i either loose everything or hold on to the possition 4ever? You would obv get filled 4life right away so what am i missing here?