简体   繁体   English

当我只需要指标在收盘价时,如何停止指标在每个刻度上的计算?

[英]How to stop Indicator from calculating on every tick, when I only need the indicator at bar close?

This is an indicator I downloaded from the internet but I have made some modifications. 这是我从互联网上下载的指标,但是我做了一些修改。 I noticed the indicator calculates the linear regression line and the upper and lower bands on EVERY tick. 我注意到该指标计算线性回归线以及每个刻度线的上下频带。

I find that to be resource wasteful as I only need the line to be calculated at the close of each bar; 我发现这很浪费资源,因为我只需要在每个小节的末尾计算行; ie when bar 0 ends and a new bar 0 is formed. 即,当小节0结束并形成新的小节0时。

It should not calculate anything while bar 0 is still incomplete. 条0仍不完整时,它不应计算任何内容。

How I do make the necessary changes? 我如何进行必要的更改?

Thanks!! 谢谢!!

//+------------------------------------------------------------------+
//|                                       Linear Regression Line.mq4 |
//|                                                      MQL Service |
//|                                           scripts@mqlservice.com |
//+------------------------------------------------------------------+
#property copyright "MQL Service"
#property link      "www.mqlservice.com"

#property indicator_chart_window
#property indicator_buffers   3

#property indicator_color1    White
#property indicator_width1    2

#property indicator_color2    Orange
#property indicator_width2    2

#property indicator_color3    Orange
#property indicator_width3    2

//---- input parameters
extern int LRLPeriod = 20;
extern int Number_SD = 2;

//---- buffers
double LRLBuffer[], LRLBuffer_Upper[], LRLBuffer_Lower[];

//int shift = 0;
int n = 0;
double sumx = 0;
double sumy = 0;
double sumxy = 0;
double sumx2 = 0;
double sumy2 = 0;
double yint = 0;
double r = 0;
double m = 0;

//+------------------------------------------------------------------+
//|                    INITIALIZATION FUNCTION                       |
//+------------------------------------------------------------------+
int init()
{
//---- indicators
SetIndexStyle(0, DRAW_LINE);
SetIndexBuffer(0, LRLBuffer);
SetIndexStyle(1, DRAW_LINE);
SetIndexBuffer(1, LRLBuffer_Upper);
SetIndexStyle(2, DRAW_LINE);
SetIndexBuffer(2, LRLBuffer_Lower);

IndicatorDigits(Digits);
if (LRLPeriod < 2) 
    LRLPeriod = 2;

IndicatorShortName("Linear Regression Line ("+LRLPeriod+")");
SetIndexDrawBegin(0, LRLPeriod+2);
IndicatorDigits(MarketInfo(Symbol(), MODE_DIGITS)+4);

return(0);
}

//+------------------------------------------------------------------+
//|                   DEINITIALIZATION FUNCTION                      |
//+------------------------------------------------------------------+
int deinit()
{
return(0);
}

//+------------------------------------------------------------------+
//|                      ITERATION FUNCTION                          |
//+------------------------------------------------------------------+
int start()
{
int limit, j, Counted_bars;
int counted_bars = IndicatorCounted();

if (counted_bars < 0) 
    counted_bars = 0;

if (counted_bars > 0) 
    counted_bars--;

limit = Bars - counted_bars;

for (int shift=limit-1; shift >= 0; shift--)
{
    sumx = 0;
    sumy = 0;
    sumxy = 0;
    sumx2 = 0;
    sumy2 = 0;
    for (n = 0; n <= LRLPeriod-1; n++)
    { 
        sumx = sumx + n;
        sumy = sumy + Close[shift + n];
        sumxy = sumxy + n * Close[shift + n];
        sumx2 = sumx2 + n * n;
        sumy2 = sumy2 + Close[shift + n] * Close[shift + n]; 
    }
    double temp = LRLPeriod * sumx2 - sumx * sumx;
    if (temp == 0)
        temp = .0000001;
//      m = (LRLPeriod * sumxy - sumx * sumy) / (LRLPeriod * sumx2 - sumx * sumx); 
    m = (LRLPeriod * sumxy - sumx * sumy) / temp; 

    temp = LRLPeriod;
    if (temp == 0)
        temp = .0000001;
    yint = (sumy + m * sumx) / temp; // was LRLPeriod (obviously)

    temp = MathSqrt((LRLPeriod * sumx2 - sumx * sumx) * (LRLPeriod * sumy2 - sumy * sumy));
    if (temp == 0)
        temp = .0000001;
    r = (LRLPeriod * sumxy - sumx * sumy) / temp;


    LRLBuffer[shift] = yint - m * LRLPeriod;

    //Print (" "+shift+" "+LRLBuffer[shift]);
}

//----------Added Upper and Lower Bands--------------//

   int    nBARs   = 0;  
   double LRLBuffer_CPY[];                        
   ArraySetAsSeries(LRLBuffer_CPY,True);                   

   j  = Bars - Counted_bars - 1;         
   while( j >  0 )                              
   {      
    ArrayCopy( LRLBuffer_CPY, LRLBuffer, 0, j, WHOLE_ARRAY );

    double StDev  = iStdDevOnArray( LRLBuffer_CPY, nBARs, LRLPeriod, 0, MODE_SMA, 0 );  

   LRLBuffer_Upper[j] = LRLBuffer[j] + (Number_SD * StDev);                                    
   LRLBuffer_Lower[j] = LRLBuffer[j] - (Number_SD * StDev);                                     

   j--;
  }



return(0);
}
//+------------------------------------------------------------------+

The easiest way is to use standard OnCalculate(***) function and run the main cycle only if(rates_total>prev_calculated) . 最简单的方法是使用标准的OnCalculate(***)函数并仅if(rates_total>prev_calculated)运行主循环。

Also, in the main cycle try to have for(int shift=limit-1;shift>0;shift--){ (be attentive), and by the way, are you sure that you need shift=limit-1 not just shift=limit ? 另外,在主循环中,尝试使for(int shift=limit-1;shift>0;shift--){ (注意),顺便说一句,您确定需要shift=limit-1而不只是shift=limit

New - MQL4 syntax OnCalculate() will not help achieve the goal, - MQL4语法OnCalculate()将不利于实现这一目标,
rather use this: 而是使用这个:

The OnCalculate() syntax is available, but the code-block actually remains outside-of-your control, when it gets launched and how many progressive steps will it be allowed to process. 可以使用OnCalculate()语法,但是代码块实际上在您的控制之外,在启动时以及允许其处理多少个逐步步骤的情况下,都OnCalculate()您控制。

The solution is to use the "old"-syntax and add "soft"-locking. 解决方案是使用“旧”语法并添加“软”锁定。

Using static variables helps keep interim values stored between calls and a just-HOT-end recalculations thus need not re-iterate from the whole depths, if shifting registers are used to just update from the just-frozen [1] Bar. 如果使用移位寄存器只是从刚冻结的[1]小节进行更新,则使用static变量有助于将临时值存储在调用之间,并且刚好进行热端重新计算,因此无需从整个深度重新进行迭代。 The rest of the Bar [0] duration it does .NOP/JIT/RET and your code does not devastate fragility of a single-solo-thread performance, shared ( Yes! SHARED!! ) by ALL Custom Indicators ( a truly devilish anti-pattern in the recent New-MQL4 updates ... ): Bar [0]持续时间的其余部分都执行.NOP / JIT / RET,并且您的代码不会破坏由所有自定义指标共享(是!共享!)的单线程性能的脆弱性(真正的恶魔般的反-New New-MQL4更新中的模式...):

int start(){
    static aCurrentTIME  = EMPTY;
    if (   aCurrentTIME == Time[0] ) return 0;   // .NOP/JIT/RET --^ 
           aCurrentTIME  = Time[0];              // .MOV/LOCK
 // -------------------------------------------- // .CALC:
    int limit = Bars - counted_bars;
    ...
    /* update just the "HOT"-end has just got into OHLCVT[1]-cells */
 // -------------------------------------------- // .FIN
}

depending on the case, you also could: 根据情况,您还可以:

datetime calculatedBarTime;

void OnCalculate()
{
    if (  calculatedBarTime != Time[0] ) 
    {
          onBar();
    }
}

void onBar()
{
     calculatedBarTime  = Time[0];
  // on bar logic....
}

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM