Who does JAPAN LIGHTS shine on? - page 11

New comment
 
RomanS писал(а) >>

But Gentlemen... it's not a grail!!!! I'm not considering this version of the Expert Advisor (as a trading one), I was disappointed :( but I keep moving in this direction together with people who have shown interest in this direction... By the way, even found a fellow from my town Voronezh :))) and by accident))

And judging by the screenshots - very much "grail".

Could you elaborate on the cause of frustration? The problem is technical or algorithmic?

 
PapaYozh >> :

And judging by the screenshots, it's very much a "grail".

Could you please clarify the reason for your frustration? Is the problem technical or algorithmic?

ForexTools 07.10.2009 23:12


oops.... I tried it at home, just to clear my conscience, even though I knew the result beforehand.

Usually when you write such a code and it first test gives a nice growth, it is difficult to part with it (the result). Several years ago I wrote several Expert Advisors of that kind per day - just tried variants and watched how they behave, as well as entry/exit systems and trawls of all sorts. The result was always the same: the optimizer had a nice growth only under certain specific parameters, but the slightest movement aside not just smoothly reduced the profitability of trading, but dramatically changed the picture.

Here's a look at what I had https://forum.mql4.com/ru/17524/page2#122774 and what with 95% probability your code expects.

Sorry mate - but it's Russian roulette. There is no system here, it's just a case of all the stars and planets have become successful. decent results are only obtained after the optimizer, and this is the first sign of fitting.



Z.I.I do not write in a branch, if you want - you can copy this my answer

 

If the picture from forex-k is on page 9 for your EA, then look for an error in the code.

By the way, in your picture on page 4

a) I do not understand anything

I posted a screenshot on the 4th page, a signal to buy

b) there are more than 2 signals (small colored "bubbles"), and only 2 ... Yes, they are profitable.

And about the ratio of initial!!! TP/SL does not make sense to talk about when trailing comes into play. The actual TP (closing the trade in profit) is much smaller than the "set" one, and the SL is unchanged. (But And this has been discussed many times).

In short ... look for an error in the code :(, in particular "No, though it analyzes only one bar (the previous one), but the decision is made, even if the bar is not analyzed". To check the adequacy of the "tick generator" of the tester and the "consistency (and fluffiness) of the ticks" of the real, you can only compare trades.

 
SergNF >> :

If the picture from forex-k is on page 9 for your EA, then look for an error in the code.

By the way, in your picture on page 4

a) I do not understand anything

I posted a screenshot on the 4th page, a signal to buy

b) there are more than 2 signals (small colored "bubbles"), and only 2 ... Yes, they are profitable.

And about the ratio of initial!!! TP/SL does not make sense to talk about when trailing comes into play. The actual TP (closing the trade in profit) is much smaller than the "set" one, and the SL is unchanged. (But And this has been discussed many times).

In short ... look for an error in the code :(, in particular "No, though it analyzes only one bar (the previous one), but the decision is made, even if the bar is not analyzed". To check the adequacy of the "tick generator" of the tester and the "consistency (and fuzzy) of ticks" of a real trade, you can only compare the trades.

The forex-k picture on page 9 is not my EA.

There is no error, the Expert Advisor works exactly as I designed it.

If someone does not understand the principle of work once again!!!!

Condition for buying:

1. we analyse the previous candlestick, it must meet the condition: the difference between the high/low of more than 45 points and the lower shadow must be at least 2 times more than the upper one, the colour of the colour is not important

2. the low of the previous candlestick must be smaller than the low of the last 4 bars, i.e. it must form some kind of extremum.

3. the current price should be higher than the open price, and higher than the price of the previous bar

ALL!!!!!

further on, open buy, set a stop at the low of the candle, which is usually 25-45 points. That is why the system does not plummet much, we put profit 4 times more, i.e. if the stop is 25 points, then the take-off is 100 points. (very reasonable) I personally start trawling after 7 bars and with the condition

SL = Low[iLowest( SYMBOL,NULL,MODE_LOW, bars,1)]- spead-Point;
             if ( BID- SL > MarketInfo( SYMBOL,MODE_STOPLEVEL)*Point  && SL>OrderStopLoss() && SL>OrderOpenPrice())
              {OrderModify(OrderTicket(),OrderOpenPrice(), SL,OrderTakeProfit(),0);}
           }

I.e. we shift a stop to Breakeven immediately (if it is possible).

That is the whole system, sell backwards!

This system is very sensitive to the parameter "the difference between high/low above 45 points". These 45 points just gave the optimizer (although 35 and 50 points are also very good!!!), so I think the grail is still very far from it, now I'm working on how to get rid of this parameter, ie, to determine the threshold.i.e. to determine the threshold of the candle itself, because for example the average difference between hai and low on eurobucks for the last year at M30 is 25p. and for example for the last quarter is 19p. shorter work on it still and work on....

 
RomanS писал(а) >>

This system is very sensitive to the "high/low difference greater than 45 pips" parameter.

Again, in my opinion, it is better to move from absolute values to relative values (i.e. % of ....). After all, volatility changes periodically, doesn't it?

And engaging in constant over-optimisation of the input parameters in absolute values is most likely the fitting.

+ it would be nice to introduce at least one more parameter-filter (like in which cases it is not recommended to open at all).

 
RomanS >> :

This system is very sensitive to the parameter "the difference between high/low above 45 points". These 45 points just gave the optimizer (although 35 and 50 points are also very good!!!), so I think the grail is still very far from it, now I'm working on how to get rid of this parameter, ie, to determine the threshold.i.e. to let the EA itself determine the candle threshold, because for example the average difference between the hai and low on the eurobucks for the last year at M30 is 25 pips. And for example over the past quarter is 19 pips.

Try ATR - it has the same exact measure. At the same time some adaptivity will appear.

 
RomanS >> :

There is no error, the EA works exactly as I designed it.

....

It has to meet the condition: the difference between high/low is greater than 45 pips.

.....

further on we open buy, put a stop on the low of that candle usually it is 25-45 points, that is why the system does not plummet much, we put profit 4 times bigger, i.e. if the stop is 25 points then the take is 100 points. (very reasonable), I personally start trawling after 7 bars and with the condition

....

That's the whole system, sell backwards!!!

This system is very sensitive to the parameter "difference between high/low over 45p". This 45p is exactly what the optimizer gave out ....

i.e. if you take a stop of 25p, your take is 100p (very reasonable).

a 100p take on the m1 chart?!!! - don't be ridiculous... on D1? - i may believe it. 7 bars? during the day on the eu - maybe. but at night on a near-flat? ....

If the trading system that analyses the bars contains specific numbers in pips, it means that the system works only with that pair, only with that timeframe, only with that period that it was optimized for (or just happened to hit the right values).

 
ForexTools писал(а) >>

a 100p take on the M1 chart?!!! - don't be ridiculous... on a D1 chart? - i may believe it. 7 bars? during the day on the eu - maybe. but at night on a near-flat? ....

Why M1? I was talking about half hour movements.

 
PapaYozh писал(а) >>

Why M1? I was talking about the half-hour movements.

I think they meant M*.

 
RomanS писал(а) >>

This system is very sensitive to the "high/low difference greater than 45 pips" parameter. The optimizer has given me 45 pips (although 35 and 50 pips is very good, too!!!), so I think it is still very far from the grail.

45p is almost half a cent. It's quite possible that half a cent isn't such a meaningless number.

1...456789101112131415161718
New comment