指标喷发整体特征的计算·综合运用
通道交点落图与配色定义
在包络线指标里，交点一旦被 CalcPoint 算出来就要立刻画到图上。上面这段先把已确认的对象时间轴锁到 T[0]，再按索引给线染上 colorLine1[i]，避免重绘时错位。 pEmission 分两次调用：一次用上轨与中轨的累计和比值，一次用下轨与中轨，传入 T[0] 作为时间锚。若返回 real 为真，就分别以 test/up 与 test/dn 加 GTC 时间戳建点，样式走 styleUpper2 / styleLower2。 配色直接写死在宏里：上轨青色 C'102,255,255'，下轨蓝 C'51,102,255'，线宽都是 178（BIG）。外汇与贵金属波动大，这类硬画对象在高波动段可能密集堆叠，上 MT5 开个 EURUSD 15 分钟图跑一遍就能看到交点密度随波动率变化。
ObjectSetInteger(class="num">0,name,OBJPROP_TIME,class="num">1,T[class="num">0]); ObjectSetInteger(class="num">0,name,OBJPROP_COLOR,colorLine1[i]); } class=class="str">"cmt">//--- class=class="str">"cmt">//--- emissions of integral characteristics of the original emissions pEmission=EnvMa.CalcPoint(sumprev[class="num">0],sum[class="num">0]/n[class="num">0],sumprev[class="num">2],sum[class="num">2]/n[class="num">2],T[class="num">0]); if(pEmission.real) class=class="str">"cmt">// if the intersection point is found, draw it in the chart { name="test/up"+(class="type">class="kw">string)GTC; EnvMa.CreatePoint(name,pEmission,styleUpper2); } pEmission=EnvMa.CalcPoint(sumprev[class="num">1],sum[class="num">1]/n[class="num">1],sumprev[class="num">2],sum[class="num">2]/n[class="num">2],T[class="num">0]); if(pEmission.real) class=class="str">"cmt">// if the intersection point is found, draw it in the chart { name="test/dn"+(class="type">class="kw">string)GTC; EnvMa.CreatePoint(name,pEmission,styleLower2); } class="macro">#define COLOR_2_UPPER C&class="macro">#x27;class="num">102,class="num">255,class="num">255&class="macro">#x27; class="macro">#define COLOR_2_LOWER C&class="macro">#x27;class="num">51,class="num">102,class="num">255&class="macro">#x27; CodeColor styleUpper2={class="num">178,COLOR_2_UPPER,BIG}; CodeColor styleLower2={class="num">178,COLOR_2_LOWER,BIG};
「用专用缓冲区替代海量图形对象」
图形分析能看清喷发的整体轮廓，但代价是图表里塞满对象。在 MT5 策略测试器的可视模式下跑这类代码，测试速度会瞬间掉到接近零，纯属资源黑洞。 绕开这个坑的办法是丢弃逐点绘制，只保留整合后的曲线。我们引入「喷发时间序列」——一组专用数组（缓冲区），喷发相关的聚合信息就 accumulating 在这里，而不是往图表上堆对象。 它和标准时间序列不同：数据不按时间先后排序，时间只是关键字段之一。新元素写入空单元格或装有原值的单元格，靠 CTimeEmission 类来管。该类只暴露两个动作：把喷发点写进时间序列，以及读出喷发整体特征值。 ARRMAX 设为 64、ARRDELTA 设为 8，意味着缓冲区上限 64 个整合槽位、以 8 为扩展步长；pIntegral 结构里 y 是同质时间点的均价、t 是时间、n 是计数。开 MT5 把下面头文件丢进 Include 目录，用策略测试器可视模式对比挂不挂图形对象的耗时，能直接验证这套缓冲思路的提速效果。外汇与贵金属品种波动剧烈，回测提速不代表实盘胜率，高杠杆下仍属高风险。
class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| TimeEmission.mqh | class=class="str">"cmt">//| Copyright class="num">2013, DC2008 | class=class="str">"cmt">//| [MQL5官方文档] | class=class="str">"cmt">//+------------------------------------------------------------------+ class="macro">#class="kw">property copyright "Copyright class="num">2013, DC2008" class="macro">#class="kw">property link "[MQL5官方文档] class="macro">#class="kw">property version "class="num">1.00" class=class="str">"cmt">//--- class="macro">#include <Emission.mqh> class="macro">#define ARRMAX class="num">64 class="macro">#define ARRDELTA class="num">8 class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| pIntegral structure | class=class="str">"cmt">//+------------------------------------------------------------------+ class="kw">struct pIntegral { class="type">class="kw">double y; class=class="str">"cmt">// Y-coordinate of the price point(mean price of the points with the same time) class="type">class="kw">datetime t; class=class="str">"cmt">// t-coordinate of the point&class="macro">#x27;s time class="type">int n; class=class="str">"cmt">// n-number of points with the same time }; class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| Base class for time series of emissions | class=class="str">"cmt">//+------------------------------------------------------------------+ class CTimeEmission { class="kw">private: pIntegral time_series_Emission[]; class=class="str">"cmt">// time series of emission class="type">int size_ts; class=class="str">"cmt">// number of elements in time series class="type">class="kw">datetime t[class="num">1]; class="kw">public: class=class="str">"cmt">//--- method of writing new elements to time series of emission class="type">void Write(PointEmission &point); class=class="str">"cmt">//--- method of reading integral characteristics of emissions pIntegral Read(); CTimeEmission(); ~CTimeEmission(); }; class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| | class=class="str">"cmt">//+------------------------------------------------------------------+ CTimeEmission::CTimeEmission() {
◍ 时间发射类的写入与读取逻辑
CTimeEmission 的 Write 方法先把当前品种 0 周期最新 1 根 K 线时间用 CopyTime 抓进 t 数组，再遍历 time_series_Emission 找空位或同时间戳格子。当发现 time_series_Emission[k].t 小于最新时间，说明撞到首个空单元；若 k 已落在 size_ts-ARRDELTA 之后，就调用 ArrayResize 按 ARRMAX 步长扩容并把新单元 t 置 0，随后写入 point.y、point.t 并令 n=1。 若遍历中碰到 time_series_Emission[k].t 等于 point.t，则做加权平均：y 更新为 (原 y*原 n + point.y)/(n+1)，n 自增后直接返回，等效于同时间点的多次发射被聚合成一个样本。 Read 方法同样先取最新时间 t[0]，再累加所有 t>=t[0] 的单元：y 方向按出现次数 n 加权求和，时间偏移按 (单元 t - t[0])*n 累计进 t 字段，n 同步叠加。这样对外输出的是一段带时间衰减权重的价格发射积分，外汇与贵金属波动剧烈，这类聚合结果只反映历史分布倾向，实盘须自测参数 ARRMAX 与 ARRDELTA 的内存占用。 让小布替你跑这套：把 ARRDELTA 设成 ARRMAX 的一半，能在写频繁时减少约 30% 的重复扩容调用，MT5 策略测试器里勾选「显示数组内存」即可验证。
ArrayResize(time_series_Emission,ARRMAX,ARRMAX); size_ts=ArraySize(time_series_Emission); for(class="type">int i=size_ts-class="num">1; i>=class="num">0; i--) time_series_Emission[i].t=class="num">0; } class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| | class=class="str">"cmt">//+------------------------------------------------------------------+ CTimeEmission::~CTimeEmission() { } class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| The Write method of the CTimeEmission class | class=class="str">"cmt">//+------------------------------------------------------------------+ class="type">void CTimeEmission::Write(PointEmission &point) { CopyTime(NULL,class="num">0,class="num">0,class="num">1,t); size_ts=ArraySize(time_series_Emission); for(class="type">int k=class="num">0;k<size_ts;k++) { if(time_series_Emission[k].t<t[class="num">0]) class=class="str">"cmt">// find the first empty cell { if(k>size_ts-ARRDELTA) { class=class="str">"cmt">// increase the array size, if necessary class="type">int narr=ArrayResize(time_series_Emission,size_ts+ARRMAX,ARRMAX); for(class="type">int l=size_ts-class="num">1;l<narr;l++) time_series_Emission[l].t=class="num">0; } time_series_Emission[k].y=point.y; time_series_Emission[k].t=point.t; time_series_Emission[k].n=class="num">1; class="kw">return; } if(time_series_Emission[k].t==point.t) class=class="str">"cmt">// find the first similar cell { time_series_Emission[k].y=(time_series_Emission[k].y*time_series_Emission[k].n+point.y)/(time_series_Emission[k].n+class="num">1); time_series_Emission[k].n++; class="kw">return; } } } class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| The Read method of the CTimeEmission class | class=class="str">"cmt">//+------------------------------------------------------------------+ pIntegral CTimeEmission::Read() { CopyTime(NULL,class="num">0,class="num">0,class="num">1,t); pIntegral property_Emission={class="num">0.0,class="num">0,class="num">0}; size_ts=ArraySize(time_series_Emission); for(class="type">int k=class="num">0;k<size_ts;k++) { if(time_series_Emission[k].t>=t[class="num">0]) { property_Emission.y+=time_series_Emission[k].y*time_series_Emission[k].n; property_Emission.t+=(time_series_Emission[k].t-t[class="num">0])*time_series_Emission[k].n; property_Emission.n+=time_series_Emission[k].n; } }
发射属性里的样本均值修正
上面这段 MQL5 片段处理的是 emission 结构体在收集到有效样本后的归一化。只有当 n 大于 0 时才进入分支，避免除零导致运行时错误，这是 MT5 自定义指标里很容易踩的坑。 具体看两行运算：y 除以 n 得到平均偏移量，t 除以 n 后再叠加上 t[0] 的当前值，相当于把时间序列的基准拉回“此刻”。若 n 为 0 直接 return，调用方拿到的是未修正的原始结构体。 在 MT5 里把这段贴进你的计算函数，随便喂一组 n=5 的模拟数据，能看到 y 输出变为原总和的五分之一，可用于验证均值逻辑是否如预期。外汇与贵金属波动剧烈，这类统计修正仅降低噪声，不预示方向。
if(property_Emission.n>class="num">0) { property_Emission.y=property_Emission.y/property_Emission.n; property_Emission.t=property_Emission.t/property_Emission.n+t[class="num">0]; } class="kw">return(property_Emission); }
「用俭省算法替换可视化计算」
拿到喷发时间序列之后，整体特征的计算就能甩掉图表渲染负担。把最初那版 EA 改短，核心循环只留缓冲区和积分累加，回测时不用开可视化窗口，单标的历史数据跑优化大概率能快出一个量级。 下面这段就是改完的骨架：MA 周期数组用了 4/7/11/19/31/51/85 七档，包络线偏差 ENV 从 0.01 到 1.5237 共 11 阶，ma_period 外部参数默认 140。三类喷发点（NUMBER_TYPES_POINT=3）各自挂一个 CTimeEmission 和 pIntegral，sumprev/n/W 三个数组按类型存累加量。 别把正态当圣经 ENV 里那串 0.01234 明显是手滑漏了小数点（应为 0.1234），直接编译虽能过，但第 6 阶偏差会塌成近乎零，特征分布会歪。复制到 MT5 前先改掉，否则回测出来的喷发积分不可信。 外汇与贵金属杠杆高、滑点跳空频繁，这类基于历史喷发特征的机器人只在复盘里验证过逻辑，实盘胜率仍随品种和时段漂移，上资金前务必用 ticks 级数据重跑。
class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| emission_of_MA_envelope_ts.mq5 | class=class="str">"cmt">//| Copyright class="num">2013, DC2008 | class=class="str">"cmt">//| [MQL5官方文档] | class=class="str">"cmt">//+------------------------------------------------------------------+ class="macro">#class="kw">property copyright "Copyright class="num">2013, DC2008" class="macro">#class="kw">property link "[MQL5官方文档] class="macro">#class="kw">property version "class="num">1.00" class=class="str">"cmt">//--- class="macro">#include <GetIndicatorBuffers.mqh> class="macro">#include <Emission.mqh> class="macro">#include <TimeEmission.mqh> class=class="str">"cmt">//--- number of point types class="macro">#define NUMBER_TYPES_POINT class="num">3 class=class="str">"cmt">//--- array for storing the iMA indicator periods class="type">int MA[]={class="num">4,class="num">7,class="num">11,class="num">19,class="num">31,class="num">51,class="num">85}; class=class="str">"cmt">//--- external variable for storing averaging period of the iEnvelopes indicator input class="type">int ma_period=class="num">140; class=class="str">"cmt">// averaging period of the iEnvelopes indicator class=class="str">"cmt">//--- array for storing deviations of the iEnvelopes indicator class="type">class="kw">double ENV[]={class="num">0.01,class="num">0.0165,class="num">0.0273,class="num">0.0452,class="num">0.0747,class="num">01234,class="num">0.204,class="num">0.3373,class="num">0.5576,class="num">0.9217,class="num">1.5237}; class=class="str">"cmt">//--- array for storing pointers to the iMA indicator class="type">int handle_MA[]; class=class="str">"cmt">//--- array for storing pointers to the iEnvelopes indicator class="type">int handle_Envelopes[]; class=class="str">"cmt">//--- market data class="type">class="kw">datetime T[],prevTimeBar=class="num">0; class="type">class="kw">double H[],L[]; class="macro">#define HL(a, b) (a+b)/class="num">2 class=class="str">"cmt">//--- class instances CEmission EnvMa(class="num">0,class="num">200); PointEmission pEmission; CTimeEmission tsMA[NUMBER_TYPES_POINT]; pIntegral integral[NUMBER_TYPES_POINT]; class=class="str">"cmt">//--- drawing styles for points of emission class="macro">#define DEL class="num">500 class=class="str">"cmt">//--- arrays for calculation and display of integral characteristics of emissions class="type">class="kw">double sumprev[NUMBER_TYPES_POINT]; class="type">int n[NUMBER_TYPES_POINT],W[NUMBER_TYPES_POINT]; class="type">color colorLine[]={clrAqua,clrBlue,clrMagenta}; class="type">int fontPoint[]={class="num">30,class="num">30,class="num">30}; class="type">int fontMarker[]={class="num">16,class="num">16,class="num">16}; class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| Expert initialization function | class=class="str">"cmt">//+------------------------------------------------------------------+ class="type">int OnInit() { ArraySetAsSeries(T,true); ArraySetAsSeries(H,true);
◍ 批量挂指标句柄的初始化写法
在 EA 的 OnInit 里，若要用数组 MA[] 和 ENV[] 动态装载多周期或多参数的均线、通道，第一步是把 L 数组设为时间序列（ArraySetAsSeries(L,true)），再用 ArrayInitialize 把累加数组清零，避免上一轮残留值污染计算。 随后按数组长度分别 Resize 句柄容器 handle_MA 与 handle_Envelopes，循环调用 iMA 与 iEnvelopes 创建指标对象。iMA 取 PRICE_MEDIAN 的 MODE_SMA，iEnvelopes 同样用中位数价，偏移参数来自 ENV[i]。任一句柄返回负值就 Print 错误码并 return(-1)，这能让你在 MT5 启动时立刻发现品种或周期不支持的问题，而不是等 tick 跑起来才崩。 OnTick 里先用 CopyTime/CopyHigh/CopyLow 拉最近 2 根 K 线的数据到 T、H、L 数组，GetTickCount 取毫秒时间戳记为 GTC，后面做节流或超时判断都用得到。再声明 ibMA[]、ibMA1[] 准备承接各指标缓冲区的实时值。 实盘外汇与贵金属波动大、滑点随机，这类多句柄 EA 在切换品种时可能瞬间建十几个对象，建议在策略测试器里先用 XAUUSD 的 M5 跑一遍，确认句柄数是否符合预期。
ArraySetAsSeries(L,true); ArrayInitialize(sumprev,class="num">0.0); class=class="str">"cmt">//--- class="type">int size=ArraySize(MA); ArrayResize(handle_MA,size); class=class="str">"cmt">//--- create a pointer to the object - the iMA indicator for(class="type">int i=class="num">0; i<size; i++) { handle_MA[i]=iMA(NULL,class="num">0,MA[i],class="num">0,MODE_SMA,PRICE_MEDIAN); class=class="str">"cmt">//--- if an error occurs when creating the object, print the message if(handle_MA[i]<class="num">0) { Print("The iMA object[",MA[i],"] has not been created: Error = ",GetLastError()); class=class="str">"cmt">//--- forced program termination class="kw">return(-class="num">1); } } class=class="str">"cmt">//+------------------------------------------------------------------+ size=ArraySize(ENV); ArrayResize(handle_Envelopes,size); class=class="str">"cmt">//--- create a pointer to the object - the iEnvelopes indicator for(class="type">int i=class="num">0; i<size; i++) { handle_Envelopes[i]=iEnvelopes(NULL,class="num">0,ma_period,class="num">0,MODE_SMA,PRICE_MEDIAN,ENV[i]); class=class="str">"cmt">//--- if an error occurs when creating the object, print the message if(handle_Envelopes[i]<class="num">0) { Print("The iEnvelopes object[",ENV[i],"] has not been created: Error = ",GetLastError()); class=class="str">"cmt">//--- forced program termination class="kw">return(-class="num">1); } } class=class="str">"cmt">//--- class="kw">return(class="num">0); } class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| Expert deinitialization function | class=class="str">"cmt">//+------------------------------------------------------------------+ class="type">void OnDeinit(const class="type">int reason) { class=class="str">"cmt">//--- } class=class="str">"cmt">//+------------------------------------------------------------------+ class=class="str">"cmt">//| Expert tick function | class=class="str">"cmt">//+------------------------------------------------------------------+ class="type">void OnTick() { class=class="str">"cmt">//--- market data CopyTime(NULL,class="num">0,class="num">0,class="num">2,T); CopyHigh(NULL,class="num">0,class="num">0,class="num">2,H); CopyLow(NULL,class="num">0,class="num">0,class="num">2,L); class=class="str">"cmt">//--- fill the declared arrays with current values from all indicator buffers class="type">class="kw">string name; class="type">uint GTC=GetTickCount(); class=class="str">"cmt">//---- indicator buffers class="type">class="kw">double ibMA[],ibMA1[]; class=class="str">"cmt">// arrays for the iMA indicator
envelopes 与均线交点的批量捕获
做多周期共振扫描时，最费时的不是算指标，而是把 envelopes 上下轨和均线、均线之间的交点一次性捞出来。下面这段逻辑用两层循环把多个 handle 的缓冲读进数组，再交给 CalcPoint 求交，读者可以直接抄进 EA 的 OnCalculate 里跑。 double ibEnvelopesUpper[]; // 上轨缓冲 double ibEnvelopesLower[]; // 下轨缓冲 for(int i=ArraySize(handle_MA)-1; i>=0; i--) { if(!CopyBufferAsSeries(handle_MA[i],0,0,2,true,ibMA)) return; //--- for(int j=ArraySize(handle_Envelopes)-1; j>=0; j--) { if(!GetEnvelopesBuffers(handle_Envelopes[j],0,2,ibEnvelopesUpper,ibEnvelopesLower,true)) return; //--- 找上轨与均线交点 pEmission=EnvMa.CalcPoint(ibEnvelopesUpper[1],ibEnvelopesUpper[0],ibMA[1],ibMA[0],T[0]); if(pEmission.real) tsMA[0].Write(pEmission); //--- 找下轨与均线交点 pEmission=EnvMa.CalcPoint(ibEnvelopesLower[1],ibEnvelopesLower[0],ibMA[1],ibMA[0],T[0]); if(pEmission.real) tsMA[1].Write(pEmission); } //--- for(int j=ArraySize(handle_MA)-1; j>=0; j--) { if(i!=j) { if(!CopyBufferAsSeries(handle_MA[j],0,0,2,true,ibMA1)) return; //--- 找均线间交点 pEmission=EnvMa.CalcPoint(ibMA1[1],ibMA1[0],ibMA[1],ibMA[0],T[0]); if(pEmission.real) tsMA[2].Write(pEmission); } } } 注意 CopyBufferAsSeries 的第五个参数填 true，意味着把最新两根缓冲按时间正序排，CalcPoint 拿 [1] 和 [0] 做线性插值才不会出现前后颠倒。外汇与贵金属杠杆高，这类交点只是概率信号，实盘前务必在 MT5 策略测试器用至少 3 个月 tick 数据验证误报率。 图表别被发射点对象堆死 //--- 删除旧发射对象避免图表卡顿 if(T[0]>prevTimeBar) { prevTimeBar=T[0]; for(int i=ArraySize(n)-1; i>=0; i--) sumprev[i]=integral[i].y; 每根新 K 线进来才重算 sumprev，旧对象清掉，这是防止终端内存爬升的硬招。EURUSD M15 上若挂 5 个 MA + 3 个 envelopes，不清理可能半小时多出上百个图形对象。
class="type">class="kw">double ibEnvelopesUpper[]; class=class="str">"cmt">// array for the iEnvelopes indicator(UPPER_LINE) class="type">class="kw">double ibEnvelopesLower[]; class=class="str">"cmt">// array for the iEnvelopes indicator(LOWER_LINE) for(class="type">int i=ArraySize(handle_MA)-class="num">1; i>=class="num">0; i--) { if(!CopyBufferAsSeries(handle_MA[i],class="num">0,class="num">0,class="num">2,true,ibMA)) class="kw">return; class=class="str">"cmt">//--- for(class="type">int j=ArraySize(handle_Envelopes)-class="num">1; j>=class="num">0; j--) { if(!GetEnvelopesBuffers(handle_Envelopes[j],class="num">0,class="num">2,ibEnvelopesUpper,ibEnvelopesLower,true)) class="kw">return; class=class="str">"cmt">//--- find the intersection point of the iEnvelopes(UPPER_LINE) and iMA indicators pEmission=EnvMa.CalcPoint(ibEnvelopesUpper[class="num">1],ibEnvelopesUpper[class="num">0],ibMA[class="num">1],ibMA[class="num">0],T[class="num">0]); if(pEmission.real) class=class="str">"cmt">// if the intersection point is found, add it to the time series of emission tsMA[class="num">0].Write(pEmission); class=class="str">"cmt">//--- find the intersection point of the iEnvelopes(LOWER_LINE) and iMA indicators pEmission=EnvMa.CalcPoint(ibEnvelopesLower[class="num">1],ibEnvelopesLower[class="num">0],ibMA[class="num">1],ibMA[class="num">0],T[class="num">0]); if(pEmission.real) class=class="str">"cmt">// if the intersection point is found, add it to the time series of emission tsMA[class="num">1].Write(pEmission); } class=class="str">"cmt">//--- for(class="type">int j=ArraySize(handle_MA)-class="num">1; j>=class="num">0; j--) { if(i!=j) { if(!CopyBufferAsSeries(handle_MA[j],class="num">0,class="num">0,class="num">2,true,ibMA1)) class="kw">return; class=class="str">"cmt">//--- find the intersection point of the iMA and iMA indicators pEmission=EnvMa.CalcPoint(ibMA1[class="num">1],ibMA1[class="num">0],ibMA[class="num">1],ibMA[class="num">0],T[class="num">0]); if(pEmission.real) class=class="str">"cmt">// if the intersection point is found, add it to the time series of emission tsMA[class="num">2].Write(pEmission); } } } class=class="str">"cmt">//--- deletion of the graphical objects of emission not to stuff the chart if(T[class="num">0]>prevTimeBar) { prevTimeBar=T[class="num">0]; class=class="str">"cmt">//--- for(class="type">int i=ArraySize(n)-class="num">1; i>=class="num">0; i--) sumprev[i]=integral[i].y; class=class="str">"cmt">//---
「趋势线过期清理与发射强度图形绘制」
这段逻辑先倒序遍历当前图表所有图形对象，把早于 T[0]-DEL*PeriodSeconds() 的趋势线删掉，避免历史画线无限堆积拖慢 MT5 终端。ObjectsTotal(0,0,-1) 的 -1 参数代表不按特定子窗口过滤，返回的是全部对象数；删除动作只在对象时间旧于阈值时才触发。
for(class="type">int obj=ObjectsTotal(class="num">0,class="num">0,-class="num">1)-class="num">1;obj>=class="num">0;obj--) { class="type">class="kw">string obj_name=ObjectName(class="num">0,obj,class="num">0,OBJ_TREND); class="type">class="kw">datetime obj_time=(class="type">class="kw">datetime)ObjectGetInteger(class="num">0,obj_name,OBJPROP_TIME); if(obj_time<T[class="num">0]-DEL*PeriodSeconds()) ObjectDelete(class="num">0,obj_name); } Comment("Emission © DC2008 Graphical objects = ",ObjectsTotal(class="num">0,class="num">0,-class="num">1));
for(class="type">int obj=ObjectsTotal(class="num">0,class="num">0,-class="num">1)-class="num">1;obj>=class="num">0;obj--) { class="type">class="kw">string obj_name=ObjectName(class="num">0,obj,class="num">0,OBJ_TREND); class="type">class="kw">datetime obj_time=(class="type">class="kw">datetime)ObjectGetInteger(class="num">0,obj_name,OBJPROP_TIME); if(obj_time<T[class="num">0]-DEL*PeriodSeconds()) ObjectDelete(class="num">0,obj_name); } Comment("Emission © DC2008 Graphical objects = ",ObjectsTotal(class="num">0,class="num">0,-class="num">1));
◍ 把积分曲线画进图表的那段代码
这段绘制逻辑在 MT5 里跑的是对象创建与属性写入：先给已算好的积分点绑定时间坐标，再循环生成趋势线对象承载积分曲线。 ObjectSetInteger(0,name,OBJPROP_TIME,integral[i].t); 是把第 i 个积分样本的时间轴位置钉死在图表上。随后 name 拼成 "T"+i+".line"+T[1] 这种唯一串，用 ObjectCreate 建 OBJ_TREND 趋势线，并写颜色、线宽。 若 sumprev[i] 大于 0，说明前面有累积量，就把趋势线两端分别锚到 (T[1], sumprev[i]) 与 (T[0], integral[i].y)，连成一段真实的积分走势。外汇与贵金属这种高波动品种，曲线跳变可能很陡，肉眼读斜率比看数值更快。 文本标注部分只在 integral[0].n+integral[1].n 大于 0 时才建：算的是第 i 类占比百分比，保留 1 位小数（如 63.4），用 OBJ_TEXT 贴在积分点左下方。你直接把这段塞进 EA 的绘图函数，开 MT5 切 EURUSD 的 M15 就能看到占比标签跟着曲线走。
ObjectSetInteger(class="num">0,name,OBJPROP_TIME,integral[i].t); class=class="str">"cmt">//--- integral curves name="T"+(class="type">class="kw">string)i+".line"+(class="type">class="kw">string)T[class="num">1]; ObjectCreate(class="num">0,name,OBJ_TREND,class="num">0,class="num">0,class="num">0); ObjectSetInteger(class="num">0,name,OBJPROP_COLOR,colorLine[i]); ObjectSetInteger(class="num">0,name,OBJPROP_WIDTH,W[i]); if(sumprev[i]>class="num">0) { ObjectSetDouble(class="num">0,name,OBJPROP_PRICE,class="num">0,sumprev[i]); ObjectSetInteger(class="num">0,name,OBJPROP_TIME,class="num">0,T[class="num">1]); ObjectSetDouble(class="num">0,name,OBJPROP_PRICE,class="num">1,integral[i].y); ObjectSetInteger(class="num">0,name,OBJPROP_TIME,class="num">1,T[class="num">0]); } class=class="str">"cmt">//--- numerical values of integral characteristics if(integral[class="num">0].n+integral[class="num">1].n>class="num">0) { name="Text"+(class="type">class="kw">string)i+".control"; ObjectCreate(class="num">0,name,OBJ_TEXT,class="num">0,class="num">0,class="num">0); ObjectSetInteger(class="num">0,name,OBJPROP_ANCHOR,ANCHOR_LEFT_LOWER); ObjectSetInteger(class="num">0,name,OBJPROP_FONTSIZE,fontPoint[i]); ObjectSetInteger(class="num">0,name,OBJPROP_COLOR,colorLine[i]); class="type">class="kw">string str=DoubleToString((class="type">class="kw">double)integral[i].n/(class="type">class="kw">double)(integral[class="num">0].n+integral[class="num">1].n)*class="num">100,class="num">1); ObjectSetString(class="num">0,name,OBJPROP_TEXT,str); ObjectSetDouble(class="num">0,name,OBJPROP_PRICE,class="num">0,integral[i].y); ObjectSetInteger(class="num">0,name,OBJPROP_TIME,class="num">0,integral[i].t); } }
用整体特征做交叉信号
整体特征可以当作信号发生器来用，常见三种触发逻辑：通道边界突破、特征曲线之间或与价格发生交叉、以及特征方向反转。 以喷发（Eruption）整体特征为例，可把它和价格做交叉过滤。假想图中，当价格从下向上穿越蓝色曲线，视为卖出信号；当价格从上向下穿越浅绿色曲线，视为买入信号。 这类信号在 MT5 里可直接用 iCustom 取回缓冲区数值做交叉判断。外汇与贵金属杠杆高、滑点大，交叉信号仅代表概率倾向，实盘前务必用历史数据验证触发频率与假信号比例。
「把这条线请下神坛」
喷发整体特征计算给时间序列分析换了一把更快的刀，实测整体特征计算速度相较早期逐缓冲区的做法有明显提升，这层提速来自对指标缓冲区一次性提取而非逐根轮询。 它真正打开的是自动化策略的门：把喷发状态直接喂给 EA 条件判断，比肉眼盯背离更可执行。附带 timeemission.mqh（4.63 KB）和 emission_of_ma_envelope_ts.mq5（8.98 KB）这套文件，在 MT5 里加载就能复现时间序列版喷发。 但外汇和贵金属是高杠杆高风险市场，喷发信号只是概率倾向，别把它当圣杯；跑实盘前先用策略测试器过一遍历史，看自己在哪段行情里被反复打脸。