Automated Trading - Is there hope?
- Thread starter irkyle1983
- Start date
- Watchers 54
Pat494
Legendary member
- Messages
- 14,614
- Likes
- 1,588
That's why I suggested Zulu. It would be the perfect arrangement. Can't think why he wouldn't agree. He would get paid loads of money if his systems are as good as they look.
Pat494
Legendary member
- Messages
- 14,614
- Likes
- 1,588
Had a look. $400 per month ?
I haven't had any success with NNs. They are usually slow to respond.
Not for me.
resistance
Member
- Messages
- 89
- Likes
- 1
do you manage accounts with any amount or there must be a specific amount in the account. What should be the minimum amount if I want you to manage my account.
Thanks Pat.
Yeah appreciate what you're saying. Like I said no-one out there is really doing what I'm doing... set yourself a reminder and look again in 3 months time.
How do you price a system that can predict the future? Its very difficult. I have no track record on collective2 so for now it will seem steep, and who knows maybe i'll consider lowering it.
But if you consider that I have a £6k account and the NNs are consistenly grinding away and closing out £500 per week (Its only £6k so I'm going for a high risk - high reward setup) then £300 a month is not too much to ask... And of course if you keep the scale of the system on the site then £300 a month for a system that's generating you £3000+ a week is an incredibly small amount to pay.
Anyway thanks for your comments, look forward to hearing from you again down the line.
well obviously, he either chickened out, or, or simply is not good of a trader and cant adapt his trading to zulutrade - remember scalping is not much favored there
I have pursued technically based auto trading and have given up. I now trade only options in the US markets, and my basis for using a multi-indicator approact is as follows:
Assume you have scoured the lists of tickers and have selected two with identical chart - Ticker A and Ticker B. You choose to flip a coin on which ticker to enter, since the indicators you employ favor neither. Your odds are 50% that you will select the best one and 50% that you will select the worst one.
Now consider a multi-indicator approach, in which you look at a variety of indicators.
Ticker A
• Technical chart is identical to B, steady 25% increase over 12 months and in a 10 day 10% pull back. Additionally, the overall market is favoring a bullish entry.
• Fundamentally -- strong, peg ratio .9, proj 25% eps growth this year, proj avg of 20%/yr over next 5 years
• 20 analysts proj. of growth - most optimistic up 50%, least optimistic up 5%, average up 19%
• Premium for a 10% downside protected put represents a 220% return on investment
Ticker B
• Technical chart is identical to A, steady 25% increase over 12 months and in a 10 day 10% pull back. Additionally, the overall market is favoring a bullish entry.
• Fundamentally -- not as strong, peg ratio 3.1 , proj 10% eps growth this year, proj avg of 8%/yr over next 5 years
• 20 analysts 12 mo price target - most optimistic up 20%, least optimistic down 22%, average down 5%
• Premium for a 10% downside protected put delivers a 50% return on investment
I can find no logic that suggests Ticker A will not on the average perform better than Ticker B in any market condition or any time frame. While there can always be an unforeseen issue that drives a ticker move in any direction, but when this condition occurs any analysis approach may be rendered invalid.
I would be interested in comments from mechanical system fanatics, who find this assessment inaccurate.
Drake
Assume you have scoured the lists of tickers and have selected two with identical chart - Ticker A and Ticker B. You choose to flip a coin on which ticker to enter, since the indicators you employ favor neither. Your odds are 50% that you will select the best one and 50% that you will select the worst one.
Now consider a multi-indicator approach, in which you look at a variety of indicators.
Ticker A
• Technical chart is identical to B, steady 25% increase over 12 months and in a 10 day 10% pull back. Additionally, the overall market is favoring a bullish entry.
• Fundamentally -- strong, peg ratio .9, proj 25% eps growth this year, proj avg of 20%/yr over next 5 years
• 20 analysts proj. of growth - most optimistic up 50%, least optimistic up 5%, average up 19%
• Premium for a 10% downside protected put represents a 220% return on investment
Ticker B
• Technical chart is identical to A, steady 25% increase over 12 months and in a 10 day 10% pull back. Additionally, the overall market is favoring a bullish entry.
• Fundamentally -- not as strong, peg ratio 3.1 , proj 10% eps growth this year, proj avg of 8%/yr over next 5 years
• 20 analysts 12 mo price target - most optimistic up 20%, least optimistic down 22%, average down 5%
• Premium for a 10% downside protected put delivers a 50% return on investment
I can find no logic that suggests Ticker A will not on the average perform better than Ticker B in any market condition or any time frame. While there can always be an unforeseen issue that drives a ticker move in any direction, but when this condition occurs any analysis approach may be rendered invalid.
I would be interested in comments from mechanical system fanatics, who find this assessment inaccurate.
Drake
fatowl
Junior member
- Messages
- 33
- Likes
- 0
Drake, hello again.
As a mechanical system fanatic who is up 19% since I started 7 months ago...
I believe that fully mechanical systems ARE possible. I cannot prove this to you (no one can). All I can point to is my live results (19% in 7 months) and backtested results (59% cumulative annual return over the time frame 1990 - 2013).
As to your example, Drake, you have analyzed this particular situation between Ticker A and Ticker B correctly in my opinion. However, mechanical system people pay more attention on performance in the long term rather than on a single situation. If there were a good, reliable way to backtest all those things you mentioned, Drake, (technicals, fundamentals, analyst opinions, option premiums) over a looooong time frame (at least 10 years), then we could make your system a mechanical one.
Given that we cannot backtest these things, your trader's intuition agrees with mine on this particular case of Ticker A vs. Ticker B.
Pat494
Legendary member
- Messages
- 14,614
- Likes
- 1,588
Interesting and well done . More details on how you managed to show regular profits would be appreciated.
I am struggling away on MT5 and there are lots of EAs for it, mostly free. Hopefully I will learn some of the programming language. It ain't easy.
I too have created a number of mechanical systems, that have been successfully back tested over 30 years.
The problem is with the process;
Create rules, back test fails
Revision 1, back test fails
Revsion 2, back test fails
Revision 3, back tast fails
etc
etc
etc
Revision 17 (or 117), back test passes
Use revision 17 (or 117), lose money
The reason for the failure is that you have created a strategy that works with the market conditions over the duration of your back test. You have designed a strategy that works with a specific market pattern. The market changes, but the mechanical strategy does not.
There may be mechanical traders that win big time, but the most successful I know blend a rigid set of rules that constrain much of the boundary of the trade decisions, but add in some human logic that is too complex to add into a mechanical system.
A system that was fanatastic over the last 30 years, will only be fantastic over the next 30 if the pattern you are measuring is the same. I have spent a 100's of hours on mechanical systems on the indexes that have been around thru the good decades and the bad, like the S & P for example.
Comments?
Drake
fatowl
Junior member
- Messages
- 33
- Likes
- 0
Yup. This is the phenomenon I call "curve-fitting". It is where you have some rules and tweak them to fit a particular historical price series for maximum profit. Of course, history doesn't repeat itself exactly the same as before. The way to get a system that will continue to work in the future is to have a system that is parameter-agnotistic. This means there is a wide range of parameters your system can use, and it still backtests profitably.
For example, let's say these are your buy/sell rules:
Buy = X-day moving average crosses above Y-day moving average.
Sell = X-day moving average crosses below Y-day moving average.
You would have a robust mechanical system if you could try a large range for X and a large range for Y and still backtest profitably. Anyone can take this system and find a a single (X, Y) pair that produces profit if you backtest on the S&P 500 index. That's easy. The challenge is to find a set of buy/sell rules (and don't forget money and risk management) that produces profit consistently with agnostic parameters.
fatowl
Junior member
- Messages
- 33
- Likes
- 0
Very complex systems that involve signal processing techniques and neural networks or other really intense analysis will typically expire in certain market situations. It's the really simple systems that last for long periods of time. Hypothetical:
Will there be a trending stock that makes a new high and continue to trend in the future?
The answer is yes. So, a very simple "buy on a new high" strategy may not expire so soon. Lots of simple systems exist that are VERY profitable.
This is called data-snooping. There are algorithms to correct for the bias it introduces but they are complex and require taking into account a very large universe of systems. My simple rule is to average the performance of all systems in all revisions. If the average is not above buy and hold then nothing was accomplished by the revisions.
M
malaguti
probability yes, but also the reward to risk.
after that largest win vs largest loss
they're my KPIs
krishnawealth
Newbie
- Messages
- 3
- Likes
- 0
Hi
May I suggest that it depends on your risk profile.
Consider opening a PAMM account, many out there are affordable to open with minimum capital (USD100, USD 200 etc). Suggest to test it first with small capital before you put more to invested it.
Some are even less than that. The more capital, the more risk (and more gain perhaps).
Cheers
Hi
May I suggest that it depends on your risk profile.
Consider opening a PAMM account, many out there are affordable to open with minimum capital (USD100, USD 200 etc). Suggest to test it first with small capital before you put more to invested it.
Some are even less than that. The more capital, the more risk (and more gain perhaps).
Cheers
No I totally disagree! the more liquid/risky the market that you are investing in - the more the compensation for your risk of investing any capital in it!
Thanks.
I have been working on various pattern - matching indicators for the last 2-3 years, as well as the 'usual' ema rsi etc ones.
The results that are produced basically say +1 / 0 / -1.
I wanted to ask for as little information as possible that would help me - you do see models that say
'day open 6390, predict close to be 6200' - precise but usually wrong or
'day open 6390, predict close to be 6300-6500' - broad enough to be useless.
I tried using data to derive smart limits and stop losses, but have (so far) failed to get anything useful.
One thing I have noticed is that across (now) 3 different series of indicators:
If the prediction backtests with much over 70% win / 30% lose, the prediction has an error in it, or is actually only triggered (say) 5 times over the backtest period and I got 'lucky' 4 times, or is over-parameterised.
Has anyone else noticed this 'speed limit' - or is it just me?
