UFO,
In your post last night you showed a market timing tool that you were toying around with. It was a crossover system using 5 day regression oscillator and a 5 day sma.
It looked pretty good over the last couple of months so I decided to backtest it all the way to 1985 using the NDX.
The result was surprising to me.
Total return = -21%
So over the last nearly 20 years you would have made a 21% loss using this system, not to mention the commissions... Sure makes Buy and Hold look alot more attractive huh?
Boy, that sucked the big one, didn't it? I wonder what something well accepted by the industry like MACD would do using the same parameters you used for this backtest?
U.F.O.
U.F.O.
Daily MACD (12,26,9) the most popular settings used as defaults in stockcharts, bigcharts etc.
Total return since 1985 going long and short on MACD crossovers = 75%
Buy and Hold return since 1985 = 596%
Total return since 1985 going long and short on MACD crossovers = 75%
Buy and Hold return since 1985 = 596%
Makes you wonder why we even use indicators, doesn't it? I'd love to see Sentient Being's thoughts on this. He's turned into the biggest backtest junkie I've ever seen. 
Thx. st13 for the feedback!
U.F.O.
Thx. st13 for the feedback!U.F.O.
UFO et all, I played around with WealthLab's backtesting and scripting tools for a while last summer. And its amazing how hard it is to write a decent system using indicators. I've learned to appreciate, that the entry and exit is only a small part of the equation, and that money management is key. I've also spent a lot of time reading about mechanical systems over at Turtle Trading Software Forum. Keeping the losses small is the obvious part. But proper pyramiding seems important along with a host of other things.
After I finished reading Elder's books, I though wow!, this triple screen stuff seems simple, just look for a good weekly trend (rising 26 week ema, etc), and then look for stochastics (or mh) to signal a buy on the day (or at least get oversold, and then have price reverse). Throw a trailing stop on and voila, unimpressive results.
Though did you see the article that was posted on the board the other day, using a simple macd cross combined with favorable months? It had a nice track record compared to buy and hold.
Has anyone backtested the DE/AD indicators yet?
After I finished reading Elder's books, I though wow!, this triple screen stuff seems simple, just look for a good weekly trend (rising 26 week ema, etc), and then look for stochastics (or mh) to signal a buy on the day (or at least get oversold, and then have price reverse). Throw a trailing stop on and voila, unimpressive results.
Though did you see the article that was posted on the board the other day, using a simple macd cross combined with favorable months? It had a nice track record compared to buy and hold.
Has anyone backtested the DE/AD indicators yet?
SB did a preliminary a few weeks ago and I don't think he liked those results either. He said he needed to do some optimizations before giving me a final number.
U.F.O.
U.F.O.
I'd love to comment.
I'm not at all suprised that over time it does poorly on the S&P. But in general, if blindly followed, I think virtually all indicators due much worse than many here would suspect.
Most of the indicators which I test against the S&P tend to do badly even when optimized. But if you go to the QQQ I think it's a bit easier to make a system that beats buy and hold. Or find individual stocks which optimize well. One thing that makes sense to me is to locate the stocks or indexes that simply trade well as opposed to those that the indicators don't seem to work on very well.
Some stocks I can optimize and trade fairly easily but the S&P has always been more of a challange.
Gary Smith talked about the fact that the bulk of the money he made (or something to that effect) came from issues that were moving upwards in tight bands.
I think when trading any indicator or group of indicators it makes sense to determine up front under what circumstances the buys are NOT to be taken.
Now, I'm not a very great system builder. And I have virtually no programming skills. I know there are some system building fanatics who are skilled programmers. They can program into a mechanical system all the stuff I can't and thus I have to run a visual check for things they include in their program. If I could teach a system to see trend, support resistance they way my brain can visualize it when looking at a chart....I would follow my systems blindly. Because they wouldn't be blind systems.
As an example, I'm trading WM. I avoided the last two buys because they came under a down trend line. both would have led me to whipsaws. So by paying attention to price, I hung on to the 12% I made on my last trade and avoided two whipsaws in a row, thus reducing the drawdown of my trading system. I actually avoided three but the first one was pure dumb luck, I was so busy that by the time I saw the buy, two days had passed and it had failed...thus I was spared yet another whipsaw.
But it's a matter of percentages right? Probablitily. Some times you will delay the buy and see price break through the down-trend resistace line to the upside. So you miss a little upside here and there as well. But hopefully more downside than upside and improve the bottom line. If out every three buys which I opt to delay (due to proximity of recent down-trend or resistance) two of them save me from whipsaws the little profit I sacrifice on the third due to a delayed buy will be more than worth it. My losses in WM occur fairly quickly on a losing trade as my losers only run about 17 days. But my winning trades run about 75 days. So I can miss a few days waiting to see if trend breaks or holds and still have some good gains out there if trend breaks and I ultimately accept the buy.
On the other hand, if the trend is up and the buy comes along, you have a support line for guidance and are probably much better off taking a buy in an up trend than a buy in a down trend?
I think we will always see people get excited about indicators which happen to be working then depressed over them as they fall apart on them as all indicators must. The averages have to catch up to them.
Now there are those other traders out there. They don't have rigid systems but have a lot of skill and experience. They have mastered the basics and can look at indicators and see the message but look at other things and put it all together and get a "feel" for the market. System work would be nonesense to them. Too constraining.
A lot of times I think newbies try to start out trading like that...and that's why they fail and never make it. They didn't learn how to take baby steps before they tried to soar with the eagles. And I believe the number of true trading eagles out there is very limited. I don't even seek the "eagle" status. I just want to improve my lot in life with the trading and by all means avoid failure as a trader.
I'm not at all suprised that over time it does poorly on the S&P. But in general, if blindly followed, I think virtually all indicators due much worse than many here would suspect.
Most of the indicators which I test against the S&P tend to do badly even when optimized. But if you go to the QQQ I think it's a bit easier to make a system that beats buy and hold. Or find individual stocks which optimize well. One thing that makes sense to me is to locate the stocks or indexes that simply trade well as opposed to those that the indicators don't seem to work on very well.
Some stocks I can optimize and trade fairly easily but the S&P has always been more of a challange.
Gary Smith talked about the fact that the bulk of the money he made (or something to that effect) came from issues that were moving upwards in tight bands.
I think when trading any indicator or group of indicators it makes sense to determine up front under what circumstances the buys are NOT to be taken.
Now, I'm not a very great system builder. And I have virtually no programming skills. I know there are some system building fanatics who are skilled programmers. They can program into a mechanical system all the stuff I can't and thus I have to run a visual check for things they include in their program. If I could teach a system to see trend, support resistance they way my brain can visualize it when looking at a chart....I would follow my systems blindly. Because they wouldn't be blind systems.
As an example, I'm trading WM. I avoided the last two buys because they came under a down trend line. both would have led me to whipsaws. So by paying attention to price, I hung on to the 12% I made on my last trade and avoided two whipsaws in a row, thus reducing the drawdown of my trading system. I actually avoided three but the first one was pure dumb luck, I was so busy that by the time I saw the buy, two days had passed and it had failed...thus I was spared yet another whipsaw.
But it's a matter of percentages right? Probablitily. Some times you will delay the buy and see price break through the down-trend resistace line to the upside. So you miss a little upside here and there as well. But hopefully more downside than upside and improve the bottom line. If out every three buys which I opt to delay (due to proximity of recent down-trend or resistance) two of them save me from whipsaws the little profit I sacrifice on the third due to a delayed buy will be more than worth it. My losses in WM occur fairly quickly on a losing trade as my losers only run about 17 days. But my winning trades run about 75 days. So I can miss a few days waiting to see if trend breaks or holds and still have some good gains out there if trend breaks and I ultimately accept the buy.
On the other hand, if the trend is up and the buy comes along, you have a support line for guidance and are probably much better off taking a buy in an up trend than a buy in a down trend?
I think we will always see people get excited about indicators which happen to be working then depressed over them as they fall apart on them as all indicators must. The averages have to catch up to them.
Now there are those other traders out there. They don't have rigid systems but have a lot of skill and experience. They have mastered the basics and can look at indicators and see the message but look at other things and put it all together and get a "feel" for the market. System work would be nonesense to them. Too constraining.
A lot of times I think newbies try to start out trading like that...and that's why they fail and never make it. They didn't learn how to take baby steps before they tried to soar with the eagles. And I believe the number of true trading eagles out there is very limited. I don't even seek the "eagle" status. I just want to improve my lot in life with the trading and by all means avoid failure as a trader.
Good comments SB. I'd like to grow a few more feathers on my backside too. Best.
U.F.O.
Best.U.F.O.
Hi Swingtrader13,
I was wondering if your results would be a lil different if you were to test the
weekly,... or monthly indicators like the 5day-reg or MACD to the results of LT buy
and hold. I don't back test so I would known, but seems almost any daily indicator
compared to a LT date of buy and hold may fail to to exceed.
Looking forword to you next backtest!
thanks
MrDev
I was wondering if your results would be a lil different if you were to test the
weekly,... or monthly indicators like the 5day-reg or MACD to the results of LT buy
and hold. I don't back test so I would known, but seems almost any daily indicator
compared to a LT date of buy and hold may fail to to exceed.
Looking forword to you next backtest!
thanks
MrDev
QUOTE (U.F.O. @ Apr 21 2004, 09:38 PM)
SB did a preliminary a few weeks ago and I don't think he liked those results either. He said he needed to do some optimizations before giving me a final number.
U.F.O.
U.F.O.
I did not get good results with the raw indicator or optimizations playing the long side.
But I think I ran it against the S&P? Might have done better against Nasdaq? Long term results were the problem. Get in the wrong type of market and it beats you up.
Since then I've moved on as I had my own goals here for my time off and my system work. presently I'm working on building my first short systems.
this is what happens when I'm trying to do to many things at once...
Should of said I wouldn't know!
QUOTE
I don't back test so I would known,
Should of said I wouldn't know!
Dev,
I'll check out the backtest on weekly and monthly MACD tomorrow as it's time for bed now, but I'll keep you posted on what I find out.
ST13
I'll check out the backtest on weekly and monthly MACD tomorrow as it's time for bed now, but I'll keep you posted on what I find out.
ST13
When building systems to program trading, it helps to pay attention to the likelihood of success simply based on the instrument one is planning on trading. The higher the volatility of an instrument, the harder it is to profit from timing schemes. This can be measured by the ratio of the daily/monthly standard deviation.
So for example, trying to time the ndx or nasdaq puts you behind the eight-ball as the daily SD is quite high compared to the monthly SD. Lots of ups and down and jiggles. This translates into lots of drawdowns.
On the other hand, Junk Bonds have a much lower ratio and can be much more easily timed with several indicators, but the returns are much less exciting and to those who like action, especially 2x beta, it can be like watching the grass grow or paint dry.
Intermediate SD ratios can be found in indexes such as RUT or SP600. These can be reasonably timed with acceptable drawdowns for most.
ECHO
So for example, trying to time the ndx or nasdaq puts you behind the eight-ball as the daily SD is quite high compared to the monthly SD. Lots of ups and down and jiggles. This translates into lots of drawdowns.
On the other hand, Junk Bonds have a much lower ratio and can be much more easily timed with several indicators, but the returns are much less exciting and to those who like action, especially 2x beta, it can be like watching the grass grow or paint dry.
Intermediate SD ratios can be found in indexes such as RUT or SP600. These can be reasonably timed with acceptable drawdowns for most.
ECHO
Echo,
I agree and also disagree i've had much success trading the swings in the NDX,
could I have been more profitable long term, well maybe, but could i stomach the
swings ie drawdowns, I seriously doubt it. Wait let me say No Way.
Most of my trading mistakes are my own fault in trusting/not trusting what the technicals are saying.
Which could also be part of the problem for most st traders in general.
I haven't built a system for trading but trade the simple MAs that seem to work well.
It can't be that simple.....or can it?
best
MrDev
I agree and also disagree i've had much success trading the swings in the NDX,
could I have been more profitable long term, well maybe, but could i stomach the
swings ie drawdowns, I seriously doubt it. Wait let me say No Way.
Most of my trading mistakes are my own fault in trusting/not trusting what the technicals are saying.
Which could also be part of the problem for most st traders in general.
I haven't built a system for trading but trade the simple MAs that seem to work well.
It can't be that simple.....or can it?
best
MrDev
Mr. Dev,
I should have stated that my statements were directed towards building mechanical systems for intermediate term trading, say 4-10 signals a year.
For NDX, given the higher volatility, short term trading would be in order to have a better chance at succeeding with smaller drawdowns. Don't you day-trade watching 5-10 min charts for divergences and chart patterns?
Echo
I should have stated that my statements were directed towards building mechanical systems for intermediate term trading, say 4-10 signals a year.
For NDX, given the higher volatility, short term trading would be in order to have a better chance at succeeding with smaller drawdowns. Don't you day-trade watching 5-10 min charts for divergences and chart patterns?
Echo
great thread, guys.
Echo:
I am using oscillators in timing along with volatility bands and I found that the higher the volatility (actually velocity) the higher the chance to catch the bottom/top of the move and once the swing starts there is a higher chance of a good move.
I am fishing for stocks on the high Alpa rating lists such as on Barchart.com under Sectors heading. My favorites are that cruise on the outside of the BB. Thais is similar to the tight channel situation described by Gary Smith.
QUOTE
When building systems to program trading, it helps to pay attention to the likelihood of success simply based on the instrument one is planning on trading. The higher the volatility of an instrument, the harder it is to profit from timing schemes. This can be measured by the ratio of the daily/monthly standard deviation.
I am using oscillators in timing along with volatility bands and I found that the higher the volatility (actually velocity) the higher the chance to catch the bottom/top of the move and once the swing starts there is a higher chance of a good move.
I am fishing for stocks on the high Alpa rating lists such as on Barchart.com under Sectors heading. My favorites are that cruise on the outside of the BB. Thais is similar to the tight channel situation described by Gary Smith.
F & D,
Interesting. What is the average length of your buy to sell time using this particular technique in higher velocity stocks. Also, are you speaking about beta or alpha or both?
Echo
Interesting. What is the average length of your buy to sell time using this particular technique in higher velocity stocks. Also, are you speaking about beta or alpha or both?
Echo
Time duration one swing which is 2-5 days. This happens to coincide be my temperament.
After the first swing it has to be reevaluated because the initial favorable conditions may no longer be present.
Beta by itself means nothing. A high beta stock could have wide swings which are useless to me without some directional force(Alpha) behind it.
After the first swing it has to be reevaluated because the initial favorable conditions may no longer be present.
Beta by itself means nothing. A high beta stock could have wide swings which are useless to me without some directional force(Alpha) behind it.
I tested one system on the NAsdaq 100 stocks and OEX 100 stocks..for 5 years of data..
IF i find a place to upload the system results i fill post them here..
Ps And the test over the OEX sucked...But the nasdaq gave a pretty nice equity curve, with very little drawdowns..I feel that i have an overoptimized systems though..
(
IF i find a place to upload the system results i fill post them here..
Ps And the test over the OEX sucked...But the nasdaq gave a pretty nice equity curve, with very little drawdowns..I feel that i have an overoptimized systems though..
(
F&D,
I think my comments don't apply for such short term signals. Your approach has merit and works for your timeframe.
Sagitarius,
Why not optimize your system for the first 3 years of data, then see how it did in the next 2 years? Then, of course, the real test is real time.
Echo
I think my comments don't apply for such short term signals. Your approach has merit and works for your timeframe.
Sagitarius,
Why not optimize your system for the first 3 years of data, then see how it did in the next 2 years? Then, of course, the real test is real time.
Echo
OK here is the system for trading all the nasdaq 100 shares. 1% of the equity is used for each position.
Here it is with a 100 % margin
This is the same system with OEX, 1 % per position
This is the same system with OEX stocks, 100 % margin
This is the NAZ 100 system, but with 500 % margin
It is not possible at hte moment ,but who knows,with the development of SSF's in the future, it might be a nice investment vehicle.
This is how the same system performs on the QQQ ETF..
All results are hipothetical, and are not recommendations to trade any of the securities or indexes mentioned. The system is not for sale, and the purpose of this post is to show some systems that i have encountered.
Here it is with a 100 % margin
This is the same system with OEX, 1 % per position
This is the same system with OEX stocks, 100 % margin
This is the NAZ 100 system, but with 500 % margin
It is not possible at hte moment ,but who knows,with the development of SSF's in the future, it might be a nice investment vehicle.
This is how the same system performs on the QQQ ETF..
All results are hipothetical, and are not recommendations to trade any of the securities or indexes mentioned. The system is not for sale, and the purpose of this post is to show some systems that i have encountered.