Comments Automated Trading Championship 2006

1
Не пойму причем тут интервью с торгами ?
Как это может повлять на результаты ?
2006.12.09 16:49
 
183
usdeur писал(а):
не сюда
183
2006.12.09 10:08
226

Wow,

Truely sorry to see you in this state, last time I looked you were on page 1!
Is there a silver lining in that if you reverse this system does it make consistent pips?

;)

2006.12.09 01:24
 
1

Вау! Молодец, Rich! Главное, отказывайся от интервью, и первое место твоё! Если что, я могу дать интервью вместо тебя, мне уже всё равно :)

113
2006.12.08 23:47
1

Rich держись сейчас евро в коррекцию пойдет. На дневном графике - молот, на часовом - пробиты прямоугольник вниз. С таким балансом первое место почти гарантировано. Zonker - тебе не конкурент, его сейчас как ветром сдует.

68
2006.12.08 22:26
1
молоток!
украина вперед!
так держать)))
2006.12.08 20:05
 
126

Prediction:


The EUR/USD will REVERSE and head DOWNWARD by end of next week or early unless it breaks thru 1.3364 with strenght, but if it falls below 1.3251 it will be a free fall downward.

   just my two cents.
2006.12.08 18:45
 
126
bullish wrote:

This is a contest about who can create the best EA. And an EA needs to be free of human interference.
If we run Zonker's EA for a year and compare it with some other EA on here (e.g Hendrick's EA). Zonker's EA will fail very badly.

Personally I dont even consider his EA an EA. He predicted that the EURO will go up, it wasn't just him who predicted this, almost everyone was predicting that. Zonker was smart enough to build his EA around this supposition. But is his EA any good, I dont think so.

Zonker, no offence just my view. But again, you've out foxed many, and Kudos to you my friend :).


BTW, Hendrick's EA is also optimized over a period of 3 months. I have tested his EA for the Euro and GBP over the last 2 years and it tanked like a rock. Maybe for the next competition, the organizer can change the rules as follow:

1) constant lots for everyne i.e. the winnings will be count as pips rather the skew data of $$$
2) all EAs should be backtested and demonstrated that they are profitable over at least one year. This way none can optimize their EAs for the competition period.

This will give the best EA over a wide market conditions. I am sure that there are lots of other rules the organizers can add to make our lives miserable.

Trong
154
2006.12.08 17:43
135
jlpi wrote:
It is sad to see that this single bad trade has destroyed 2.5 months of good trades, specially when you consider that even if you still are on the plus side for the pips (I think +229 / -204) the balance is negative as the loss has been taken with a more than average size of lots.
Of course these bad trades are inevitable, but if in your backtest it didn't occur more than once on 650 trades, don't you think that it was over optimized as it came here just after 25 trades, or is it just bad luck?
Maybe a point to consider as well is the TP place. It seems that the minimum now is 6 pips away but maybe it is really Maybe 10 could be better. Of course it may drop the Winning ratio but I guess that an objective of 1/650 is not realistic and an idea would be to keep a ratio of 1/25 and improve the TP part.

Thanks. Yes, all your points are good, and we will be looking at various ways to improve the EA. As it is, its TP is computed with respect to the history, and it can vary quite a lot, in principle, although in these 27 trades it didn't. But what you say corresponds well to variant backtesting we have done, which show that the resulting profit go up when the TP is held higher, although at the same time the drawdown and the number of loss trades increase.

That particular bad trade I would count as "bad luck", primarily caused by the "fake rise" on Monday morning. I also tracked the FXDD chart at the time (mid morning for me), which started one hour earlier with a drop instead and gradual climb, and that Surfer's history evaluation on that chart left out the signal that the Competition Surfer acted upon. However, those kinds of "fake transients" (similar to spikes I suppose) are something that the EA should be able to digest without getting itchy.

In general I don't think this invalidates the analytical basis in the EA, but rather calls for improvement in its decision logic; or perhaps just turn the "squelsh knobs" a bit more. We'll see; I have learnt a lot during the competition, and will certainly participate in the next one, if it comes up.

Meanwhile we'll be going live, with the squelsh knobs turned high, and then we'll be able to lose real money :-)

135
2006.12.08 13:07
135
It is sad to see that this single bad trade has destroyed 2.5 months of good trades, specially when you consider that even if you still are on the plus side for the pips (I think +229 / -204) the balance is negative as the loss has been taken with a more than average size of lots.
Of course these bad trades are inevitable, but if in your backtest it didn't occur more than once on 650 trades, don't you think that it was over optimized as it came here just after 25 trades, or is it just bad luck?
Maybe a point to consider as well is the TP place. It seems that the minimum now is 6 pips away but maybe it is really Maybe 10 could be better. Of course it may drop the Winning ratio but I guess that an objective of 1/650 is not realistic and an idea would be to keep a ratio of 1/25 and improve the TP part.

230
2006.12.08 11:32
Pages: 
Total: 1565