Thread: Automated Trading

Disagree Phil - Hi by the way, I can recall watching back when I used xtick nice to see you are still around.

OK, so in argument tell me this - how can you programme a professional to know that it's not wise to trade CAD if it's at dodgy area's like parity?

How do you tell the expert to consider a japanese curiosity meeting might be the wrong time to go short?

How do you tell it to depart cable longs alone when there's the first bank run in over 100 years due to sub-par woes?

How can you tell it if NFP comes in powerful there might be a weak rally but if it comes in weak the dollar will crash?

Discretionary autotrading does not mean that you were not great enough to make it automated - it means you were great enough to take sensible decisions a dumbass computer wouldn't be able to be programmed to perform regardless of how good a programmer you are.

Originally Posted by ;

Disagree Phil - Hi incidentally, I can recall seeing your name years back when I used xtick fine to see you're still around.

OK, so in argument tell me this - how do you programme an expert to know that it is not wise to trade CAD if it is at dodgy area's like parity?

How do you inform the expert to think about that a japanese interest meeting may be the wrong time to go short?

How do you let it depart cable longs alone when there is the first bank run in over 100 years due to sub prime woes?

How do you tell it that if NFP comes in powerful there may be a poor rally but if it comes in feeble the dollar will crash?

Discretionary autotrading does not indicate that you weren't good enough to make it automated - so you're good enough to make sensible decisions that a dumbass pc could not be programmed to perform no matter how good a programmer you're.

There are automated methods which don't take care of NFP, subprime, Northern Rock, Greenspan's book or anything else. They do what they are programmed and they are automated. I can guarantee you that I understand this to be true.

I have egies that require some discretion and evaluation of market requirements. To be honest, they are more profitable compared to automated ones.

My sole point is that in case a egy can't be trusted and left to its own devices than it is not truly an automated system and ought not to be egorized as such. Any discretion taints the automation process and can not be backtested. You can not measure and test discretion. It is not a repeatable experiment capable of generating results that are verifiable. I'm not trying to aruge with anybody, I just believe anything less than full automation is by default.

I haven't thought about xtick in a long time...a year of my life which I will never get back.

The egy I use is automated - it passes auto, exits auto handles auto and resets itself auto. The discretion is in which pairs I put it on and when I dont put it upon

in this sense then do your automated pros run on each pair? Including silver and gold? Including pairs such as gbpnzd? - if they dont then you're doing exactly the same thing - determining which pairs to place it on and when. - What's the difference?

Originally Posted by ;

that the egy I use is automated - it enters auto, exits auto handles auto and resets itself auto. The discretion is where pairs I put it on and when I put it upon

inside this sense then do your automated pros run on every pair? Including silver and gold? Including pairs such as gbpnzd? - you're doing exactly the same thing, if they dont then - determining which pairs to put it on and when. - What's the difference?

I stated my point before that I felt automation did not require intervention. When the egy goes to a pair it remains there. I do not need to worry about things such as parity. Believing isn't automation and that I really don't understand how one would backtest their opinion. I don't have any desire to debate semantics.

The choice of a specific pair to exchange is a kind of optimization, not automation. I do not get gold, silver (in currency quotes), or gbpnzd quotes. If that makes me a trader them I'm comfortable with that. You win.

Lol, it's not a fight Phil that there are various levels of automation apart from the view you hold.

As for backtesting, nicely the past doesn't equal the future. I used to be an avid backtester - no longer, most backtesting is a complete waste of time IMO (note the word'many')

lets agree to disagree eh

Group Hug?

Originally Posted by ;

lol, it's not a battle Phil, only that there are various levels of automation aside from the somewhat blinkered view you hold.

As for backtesting, well the past does not equal the future. I used to be an enthusiastic backtester - no longer, most backtesting is a complete waste of time IMO (notice the word'many')

lets agree to disagree eh

Group Hug?

I believe you could leave everything to a automated trading system to trade on your behalf. Everything you have to do would be to revise your code and parameters in time to time to be certain it trades according to the current market moves.

Could I join the group hug?

Originally Posted by ;

as for backtesting, well the past does not equal the future. I used to be an avid backtester - no more, most backtesting is a complete waste of time IMO (notice the word'most')

Most backtesters already know that. A good measure of backtesting is really a circumstance where your egy walks ahead at an annualized rate of 50 percent of their in-sample results. Anything considerably higher than that and I would be skeptical. Anything and I'd be worried to not trade that egy. Evaluations 24/7 on a host run and would not consider it squandering time. In my experience, keeping up with automation is a lot harder than discretion but still simpler than waking up at 0200 for your European open.

Originally Posted by ;

Many backtesters already know that. A fantastic measure of backtesting is a situation where your egy walks in an annualized rate of 50 percent of their in-sample results. Anything would be very skeptical. Anything and I would be worried to not trade that egy. Tests 24/7 on a server conduct and would not think about it squandering time. In my experience, keeping up with automation is a lot more difficult than discretion but still simpler than waking up for your European start.

well said.

Originally Posted by ;

I personally believe you can leave everything to an automated trading platform to trade for your benefit. Everything you have to do is to revise parameters and your code in time to time to make sure it trades based on the current market moves.

Could I join the group hug?

I must agree with phil within this discussion - automation and discretion are different in only one major aspect: discretion entails decisions that must not be programmed due to their relationship to market conditions. If this app needs it's inputs changed the comments are actually describing a program that is poor. A thorough algorithm can simply select what to do and recognize changes in volatility..etc very readily. All the time that I hear about this EA that selects EAs as if it's some kind of miracle - in truth, having one EA is equal to hundreds of little ones. That's not the problem in any way, if you've got proficient programmers or it shouldn't be.

The problem is that human traders themselves are not too convincing. Making probability statements about the market is hard, which is why not many people are rich on currencies. If you attempt to sit down and come up with something authentic and trade-able about the market, you won't do it quite quickly. Also, the way that you gauge market activity is quite visual i.e traders utilize the chart and also the background real-world understanding when studying the past few bars. Visual activity such as this is extremely fast in human beings while slow in machines because you need to mathematically define what your eyes are watching for them to reach exactly the same effect - have the exact same access to market conditions which you do.

You will know what I mean if you choose a very simple graph with dots on it: the human can select a dot and instantly see ones the closest dot for this, whereas the system needs to iterate through the points and compute *all* the distances between the dots to eventually let you know which one is closest. They are not exactly the exact same algorithm.

I personally would never invest in a programmed system unless it's
a)Completely independent of me(that's the whole point after all)
b)Logically proven to operate.

Obviously, I have never seen one. The AI chart students we discussed earlier on this thread would serve as an indior/alert kind of item.

Originally Posted by ;

In my experience, keeping up with automation is a lot more difficult than discretion but still simpler than waking up at 0200 for the European start.

Very very good point philmcgrew, I'm walking example of what late nights and midst of the night wake-ups to put rankings can do for your own body. After I shifted to fx because of being down under, the result, I did this for over two years? Getting so run-down, tired and tired my trading endured, and in the end, my health, getting cancer at 30 (not the only reason of course). If ever I needed a wake-up call that display sitting day and night was not the way to do this was it, and my quest for auto trading (or my and soultraders variant of it ) was ignited once again.

Oh and the thoughts on backtesting, bang on, total waste of time unless you are completely auto trading, how can you back evaluation history with current day prejudice? I do quite superficial testing at most, better to gain market knowledge and principles, apply a dash of logic, a sprinkly of invention and a nice coating of fortune (hard ingredient to find) and then just grip and proceed.