Automated Trading - Is there hope?

[jago25_98]

futures?

I want to add what I've found to this post as a reference for anyone interested and add my own question at the end in exchange.

Collective2: Variety of options, well presented, US not UK regulated. Good fees for large balances. I didn't like the fee collection method of dipping into the brokerage account itself.
ZuluTrade: Seems like the easiest to get into. I'm not clear on what the fees are. Forex only. Probably USA regulated only?
MirrorTrader: Fees are hidden in the pips so it's a sliding scale cheaper for the little guy _I think_ and not a rip off with low balances. Systems available seem to be only for forex and gold. Systems only cover one symbol each. Fees taken more securely from broker.

I'm interested in all this because I have tried to trade manually and I just can't respond quick enough.
Something I have wanted to do in the past was things like
"Buy when MACD crosses" or "When Vortex = x then sell" but I don't have the programming skills for this and only linux based servers to execute this.

So:

- where can I find a way to program in those conditional buy/sell statements based on indicators without me having to calculate the indicator?

- are there any automated providers other than Collective2 that offer systems trading futures and options rather than just forex? Forex is only part of my strategy. Ideally I would like to select a system according to my own thoughts on market conditions

- what's the fee schedule at ZuluTrade?

edit: Also I'm still a bit unconvinced about marketing trade accounts as spreadbetting with no leverage in order to not pay UK tax:
http://www.trade2win.com/boards/general-trading-chat/149556-mirror-vs-self.html

Also a very real threat to me seems the possibility of mathematical pump and dump - making a system to draw enough people in to be mathematically visible in the market and then suddenly trading against this in an alternative account. Or as another example that's easier to understand just putting a signal on your own company and then selling assets.

 

[Purple Brain]

It is extremely unlikely any commercially available signals providers or automatic trading facility for retail traders is going to serve the customers best interests. If you have a system which provides you with signals (entry & exit) of sufficient quality to provide positive expectancy over the long haul and this system includes risk & money management and - the holy grail of all trading systems - can be automated to a degree where there is no increase in risk or negative impact on P&L – why would you ever consider turning it into a commercial product? Why would you need to?

The reason brokers exist is because they don’t know which way the market is going either, but they’ll happily take a slice of all their customers’ bets who think they do. Same as with the trading side of investment banks. As I understand it, they rarely make money off their own treading efforts – the bulk of their profits coming from commissions/spreads on customers’ orders.

It also stands to reason (mine, which could well be wrong) that even the HFT algos of the finest of the finest will be constantly coming up against the constraints inherent in competing with the other firms HFT algos. They can’t all be spotting the same arb in the same way all the time with everyone winning. An opportunity cannot be seized by all.

My point is that even the big boys’ automated trading is I suspect a day-in day-out succession of tweaks and even major overhauls run by teams of quants and metaquants. They would also like to just turn on their machines and sit back and let them make money, week after week, month after month with no additional effort. I really don’t think it can happen that way.

Much less likely for the retail trader, the majority of whom I am told have a single role; that of providing brokers with commissions and the big boys with liquidity - until they run out of capital on their underfunded, over-exposed accounts.

Although there is a large amount of cynicism, probably mostly justified, on this site I do agree with those that suggest the best way to make money out of any automated systems or signals provision service is to sell that service – not buy it.

 

[andrewdb]

...
It is extremely unlikely any commercially available signals providers or automatic trading facility for retail traders is going to serve the customers best interests.

...My point is that even the big boys’ automated trading is I suspect a day-in day-out succession of tweaks and even major overhauls run by teams of quants and metaquants. They would also like to just turn on their machines and sit back and let them make money, week after week, month after month with no additional effort. I really don’t think it can happen that way.

In terms of commercially available tools - i suggest you look for the one that has been around for a long time, and has had a number of mentions here.

You could also look for non-commercial advice/tools - some people do this not to make money, but to be 'right', or just helpful. There are some very good articles here - the one on the basics of trading covers the subject in about 30 a4 sheets that many books seem to take 300 pages over.

It also depends what you mean by long haul.
My first model lasted for ~2 years. The current one has been largely in place for the last 18 months.

 

[numbertea]

That last post is quite to the point and those items must be considered in the overall trading plan. Automated trading is really a game of finding what doesn't work and eliminating those things from the system. If you dont reach the end of your investment capital and you still have methods left to trade in your system then you are on the right track. If you find that no method is consistent then you can continue until you are broke or call it quits. Depending on luck, skill, and perserverence, you may make it or you may not.

Cheers

 

[SlowlyButSurely]

Automated Trading - Is there hope?

No.

 

[random12345]

Automated Trading - Is there hope?

No.

...for you or all of us?

 

[SlowlyButSurely]

...for you or all of us?

From what I have seen personally, and from what I have been told by programmers (some excellent with years of experience) that have tried and failed, it is a never ending process of tweaking and rewriting codes/permissions/patterns etc. Whenever a successful model is found the lifetime is limited and once it ends, it can cause losses which can sometimes be severe (look at knight Capital) and you have to then find the next one.

As for EA's the only 'successful' EA I have come across is one that trades on latency and bad prices across multiple brokers on MT4/5. It is not a legitimate trading method though and is usually not allowed except by bucket shops.

In my opinion these models do not work for retail traders as the resources needed to keep the parameters constantly updated outweigh the benefits and it is much easier to learn to trade manually.

 

[random12345]

From what I have seen personally, and from what I have been told by programmers (some excellent with years of experience) that have tried and failed, it is a never ending process of tweaking and rewriting codes/permissions/patterns etc. Whenever a successful model is found the lifetime is limited and once it ends, it can cause losses which can sometimes be severe (look at knight Capital) and you have to then find the next one.

As for EA's the only 'successful' EA I have come across is one that trades on latency and bad prices across multiple brokers on MT4/5. It is not a legitimate trading method though and is usually not allowed except by bucket shops.

In my opinion these models do not work for retail traders as the resources needed to keep the parameters constantly updated outweigh the benefits and it is much easier to learn to trade manually.

I have to disagree. I am able to keep parameters updated within a single 24 hour session. More often than not only the application of a model needs to change rather than finding a whole new approach.

Well Knight Capital was a major software error, basically a bug, as opposed to any underlying functional (trading wise) element of their algos failing.

The problem with manual trading is that it is also easier to convince yourself that you have learned to do it with any measure of success. Not many people can spare the time or money to generate a relevant sample size in their lives to find out the truth, good or bad. Though automated trading is a costly endeavour in terms of time, sanity and effort, it can provide more metrics by which to understand how well your live trading is going.

 

[SlowlyButSurely]

I have to disagree. I am able to keep parameters updated within a single 24 hour session. More often than not only the application of a model needs to change rather than finding a whole new approach.

Well Knight Capital was a major software error, basically a bug, as opposed to any underlying functional (trading wise) element of their algos failing.

The problem with manual trading is that it is also easier to convince yourself that you have learned to do it with any measure of success. Not many people can spare the time or money to generate a relevant sample size in their lives to find out the truth, good or bad. Though automated trading is a costly endeavour in terms of time, sanity and effort, it can provide more metrics by which to understand how well your live trading is going.

Fair enough, I can only say congratulations then as I have known some great programmers who have worked for large institutions that struggled with getting an automated system to work consistently without being exposed to large drawdowns. If I may ask out of curiosity, which products do you trade? Also is your system an algorithm that trades off of a set of instructions and parameters or is it an EA of some kind that automates the trading but requires input by you?

 

[random12345]

Fair enough, I can only say congratulations then as I have known some great programmers who have worked for large institutions that struggled with getting an automated system to work consistently without being exposed to large drawdowns. If I may ask out of curiosity, which products do you trade? Also is your system an algorithm that trades off of a set of instructions and parameters or is it an EA of some kind that automates the trading but requires input by you?

I have worked for Bank of Tokyo (UFJ) in the past and we had a number of profitable models that did not have extreme risk profiles at all, though I do not use their stuff in my own systems, not least since generally the desk only made adjustments at Python level.

I trade only forex, I find futures much too difficult. The speculative control over mean reversion is too extreme for my liking. If you have a MR curve system then you may do better. Some futures markets have become incredibly efficient.

It is an algo with parameters. I do not use pattern recognition or neural stuff.

All automated trading should ultimately boil down to the average gross pt movement of a cross or instrument outside of consolidation levels during the day and thus how many times can this can be traded at your level of costs before overtrading becomes harmful.

Even so, if anyone has years of manual trading under their belt, I would not recommend a switch to automated trading. Development time has taken a lot out of me over the years. I have been known to converse with myself in mark up code while taking a shower in the past. I am looking to book my final profits after this next 3-5 year period as volatility is picking up and I see it staying that way for a while.

 

[InviertaBien]

You have put some valid points there - on many levels.
I have seen actually people winning from following + I myself am not making bad so far. I guess you can get money out of it, just you cannot make that much if you compare it to manual. At the end of the day is that you receive what you get.

Definitely there's more money in developing and selling EAs and signals than in using them. That's not to say that you can't find a profitable system. I think there are several commercial and even some free EA's that are profitable in the long term. The key words here are "long term", because most of these systems have long periods of draw down, some even more than a year, like many volatility or breakout EAs. I you are patient enough with these systems you can be profitable after several years, but the problem is that few people can do it.

 

[numbertea]

Definitely there's more money in developing and selling EAs and signals than in using them. That's not to say that you can't find a profitable system. I think there are several commercial and even some free EA's that are profitable in the long term. The key words here are "long term", because most of these systems have long periods of draw down, some even more than a year, like many volatility or breakout EAs. I you are patient enough with these systems you can be profitable after several years, but the problem is that few people can do it.

It does seem that this is the case. I have given up after spending a lot of time discussing with others if there is any great systems for sale. Like many things for sale in this world, the product is rarely as good as the sales pitch. So I developed my own from scratch and it has been a great adventure from the beginning. It is a never ending process and life consuming at times. However, it is profitable. As was said before, I can now truly see how a trading method really performs and statistically identify problems or advantages. I continue to add patterns and when I have time I look for better ways to stay on the correct side of the market when a reversal is missed. As time goes on my profitability is better. I've been working up to where I am on and off for about 6 years, steadily for the past 2. I have invested thousands of hours but I can truly say that it is worth it to me even if it wasn't profitable. It has forced me to learn so many things about the markets. They are awesome vehicles of human psychology!

Cheers

 

[Pat494]

I have been testing on a demo account the EAs on MT4. I am currently on page 7 and have just found out that Ride Alligator is like most of them - nothing happened !!
I put the EA on 20 forex pairs split equally between 1M, 5M, 15M and 30M for 1/2 a day.

Of the few that did anything at all - they failed on their results !!

Disappointing or what ?????

 

[Solas0077]

The more people that trade systems, the more free tools, the less the opportunity gets.

 

[D70]

I have worked for Bank of Tokyo (UFJ) in the past and we had a number of profitable models that did not have extreme risk profiles at all, though I do not use their stuff in my own systems, not least since generally the desk only made adjustments at Python level.

I trade only forex, I find futures much too difficult. The speculative control over mean reversion is too extreme for my liking. If you have a MR curve system then you may do better. Some futures markets have become incredibly efficient.

It is an algo with parameters. I do not use pattern recognition or neural stuff.

All automated trading should ultimately boil down to the average gross pt movement of a cross or instrument outside of consolidation levels during the day and thus how many times can this can be traded at your level of costs before overtrading becomes harmful.

Even so, if anyone has years of manual trading under their belt, I would not recommend a switch to automated trading. Development time has taken a lot out of me over the years. I have been known to converse with myself in mark up code while taking a shower in the past. I am looking to book my final profits after this next 3-5 year period as volatility is picking up and I see it staying that way for a while.

What does that mean?

 

[Liquid validity]

If you can't or won't code and test it yourself, forget it.
If you can't or won't pay for decent tick data, or understand and work
within the limitations of the best free data (only one source I'm aware of that is acceptable), forget it.

If you think anyone will tell you, show you, give for free or sell you anything
that has half a chance or works, forget it.
Anyone drawn to automation as it seems the easier route, forget it.
There is no easy answer, no solution for all time.

Discretionary or automated, you have to put the time in on research,
testing, SIM forwards testing, live forwards testing.
Then you can answer your own questions...

 

[Pat494]

If you can't or won't code and test it yourself, forget it.
If you can't or won't pay for decent tick data, or understand and work
within the limitations of the best free data (only one source I'm aware of that is acceptable), forget it.

If you think anyone will tell you, show you, give for free or sell you anything
that has half a chance or works, forget it.
Anyone drawn to automation as it seems the easier route, forget it.
There is no easy answer, no solution for all time.

Discretionary or automated, you have to put the time in on research,
testing, SIM forwards testing, live forwards testing.
Then you can answer your own questions...

So all in all the average Joe need not bother because he won't get the big money in a 100 years or two ?

 

[Liquid validity]

So all in all the average Joe need not bother because he won't get the big money in a 100 years or two ?

Well if the average joe is a synonym for unrealistic, inept and lazy, then yes
Nothing is easy, or has any guarantee.

That possibly sounds harsh, its still true, and its also a truth
every individual must acknowledge and be aware of.

 

[random12345]

What does that mean?

Just that I intend to retire from trading after a few more years - I've had enough of sleepless nights. I've had winning streaks of over 30 working days in a row and I STILL can't get more than 4 hours during those times. Even less during others.

The idea that you've just been found out and been lucky for the past x years never goes away on a losing day... the lows of a losing day are much lower than the highs of a winning day, no matter the magnitude of the latter vs the insignificance of the former.

In my mind, anyway.

 

[random12345]

Well if the average joe is a synonym for unrealistic, inept and lazy, then yes
Nothing is easy, or has any guarantee.

That possibly sounds harsh, its still true, and its also a truth
every individual must acknowledge and be aware of.

I wouldn't wish automation on anyone. It's like a nasty recurring disease that you spend 2 years of your life trying to cure and then it still flares up in the subsequent years.

If anyone thinks they can automate without working harder than they have ever worked and giving up all their social time and weekends for a few years then good luck to them.

I'm still amortising my lost time each time I return!