HOW TO TRADE FOR NOTHING ? - page 9

 
Vizard:


you see - you understood the crowd as an extremum ...but how Richie understood it is unclear ...i asked him ...he must have thought i was joking and sent me off to watch phonetics ...and i was asking seriously ...no jokes ...

But again about extrema... you have to identify it first - to understand that it is him... on the history it's easy to do... but in real time it's another matter... how do you identify them for example?

Knowing the rule doesn't mean following it.... Extremes are tops, troughs of trends at least on the days. The eu has now reversed and it is too early to trade...
 
Tantrik:
Knowing the rule does not mean following it.... Extremes are tops, troughs of trends at least on the days. Now the eu has reversed and it's too early to trade....


i asked you what is your definition of an extremum ? how did you get to the conclusion that it was an extremum on d1 ? that's my point ... that's 1 question

2 - and where and when (exact date and time if possible) did the reversal of the euro occur ? + explanations.... why ....

 
Vizard:


i asked what is your definition of an extremum ? how, on what basis did you come to the conclusion that the extremum is now on d1 ? that's what i mean ... that's 1 question

2 - and where and when (exact date and time if possible) did the reversal of the euro occur ? + explanations.... why ....

Bottom on the Euro 1.19 rise say 1000 pips - this could turn out to be a trend correction down 1.50 - 1.19 = 3100 pips - 30% movement.

The rise time is a month - not enough, and at least two corrections are needed - already a new trend. (you can see the exact date on the chart)

 
Vizard:

....

or not through 0 in this example ? but for the colour change...if anyone has it, post it in the csv file - i may play with it later - i'll pick it up....

i don't think that similar indicator may appear in public domain. i programmed all of the above examples in about 5 minutes, but it took me about a month to figure out how to do the calculations ...

If you ask Mathemat, he's a very good analyst, I don't think he'll miss it. If I've posted level breakdowns in the forecasts section, they were like toys from Prival ))

If i have posted some breakdowns of levels and the direction of the breakdown.

 
Prival:

I don't think that such an indicator would appear in the public domain. If I programmed all the above examples in about 5 minutes, then this one took me about a month to figure out how to do it and what to calculate...

If you ask Mathemat, he's a very good analyst, I don't think he'll miss it. If I've posted level breakdowns in the forecasts section, they were like toys from Prival ))

it's serious. he'll see why i was so good at identifying breakdowns of levels and the direction of the breakdown.

ahh... it's yours... i thought it was someone else's... if you don't mind, send it to me in a CSV file...in this format .... (separators with a semicolon...or just a comma)

"Date";INDYUK
14.07.2010;1.27238
14.07.2010 0:05:00;1.27235
14.07.2010 0:10:00;1.27234
14.07.2010 0:15:00;1.27215
14.07.2010 0:20:00;1.27219

I don't even need the index myself.... to be honest I didn't quite understand its value - and the time span on the screen is certainly small.... but in general it's nothing special at first glance ... maybe I'm wrong ...

 

just a hastily made copy.... Prival line of turkey in tsv will be ?

 
Vizard:

just a hastily made copy.... Prival the line of the turkey in the csv will be ?


I will write. but later. the input is there. we need to think about the output on what (the network) to train it (the network). I just need a test, to train the network on a section and see how it handles it.

S.S. there are more important things to do right now

 

I am reading this thread and realise that I have already read about it somewhere. Then I remembered, in a very good book Encyclopaedia of Trading Strategies, Katz and McCormick tested a model similar to this one. It was called the cycle theory. Basically, they tried to define a cycle, to identify the current phase or something, and on the assumption that this phase will continue for some time, bet on it. The result was so awful that even using simple moving averages gave better results.

Some periodic component will stand out of course. But I don't think that in the long run trades on this will even recoup the size of the commissions (Katz and McCormick, for example, have failed to do so). Using neural algorithms is not a panacea. Of course you can train them so that they will show tens of thousands of percent per annum on the history, but in practice their yield is unlikely to exceed the amount of deductions to the broker. Yes, making accurate predictions is very good, but you also understand that trading is not one or even ten accurate predictions, it is thousands of transactions, and the commissions have to be recouped.

 
CROM:
Thank you.

I learn a lot with you !


Post with decompiled indicator deleted