📊

解密开盘区间突破（ORB）日内交易策略·综合运用

（3/3）·从单一信号到投资组合，用α、β与VWAP把ORB策略的边界和局限一次说清

偏理论进阶第 3/3 篇

很多交易者把ORB当成万能钥匙，照搬论文参数却忽略时区映射和标的差异，回测漂亮实盘崩盘。把QQQ的研究结论直接套到USTEC差价合约上，杠杆和隔夜跳空会把原本微弱的优势吞掉。认清策略依赖的是市场趋势还是独特α，才谈得上仓位分配。

本章目录

VWAP与买卖执行的代码骨架
锚定会话起点与绘制 VWAP 标记
用14天波动锚定当日上轨
用14天波动算出当日下轨并标点
按止损点数反推手数的实盘函数
组合分散化不是堆数量而是看相关性
画得少，看得清

VWAP与买卖执行的代码骨架

下面这段逻辑把时间窗判定、买卖下单和VWAP计算拆成了独立函数，直接在MT5里建EA就能跑通主干。时间判定只用了小时和分钟的硬比较：当当前小时大于等于结束小时且分钟也达标时返回true，否则false。这种写法不考虑跨日，仅适合日内固定收尾的场景。卖出函数先取SYMBOL_BID，用NormalizeDouble按品种小数位规整，止损价按 bid*(1+slp) 算，slp是外部止损比例；仓位由 calclots(bid*slp) 得出，最后 trade.Sell 带止损进场。买入对称处理，用SYMBOL_ASK和 ask*(1-slp)。 VWAP函数从指定K线索引 startCandle 向下循环到1（排除当前K线），典型价取 (High+Low+Close)/3，乘成交量累加得sumPV，体积累加sumV，最终 vwap = sumPV/sumV；若sumV为0返回0.0。回测时可观察：在EURUSD的M15上，用最近50根K线算VWAP，价格偏离均值±0.3%往往对应短期均值回归概率偏高，但外汇高杠杆下仍可能继续滑出。

MQL5 / C++

if (currentHour >= endHour && currentMinute >= endMinute) class="kw">return true;
else class="kw">return false;
}
class=class="str">"cmt">//+------------------------------------------------------------------+
class=class="str">"cmt">//| Sell execution function                                        |
class=class="str">"cmt">//+------------------------------------------------------------------+    
class="type">void executeSell() {    
     class="type">class="kw">double bid = SymbolInfoDouble(_Symbol, SYMBOL_BID);
     bid = NormalizeDouble(bid,_Digits);
     class="type">class="kw">double sl = bid*(class="num">1+slp);
     sl = NormalizeDouble(sl, _Digits);
     lot = calclots(bid*slp);
     trade.Sell(lot,_Symbol,bid,sl);    
}
class=class="str">"cmt">//+------------------------------------------------------------------+
class=class="str">"cmt">//| Buy execution function                                         |
class=class="str">"cmt">//+------------------------------------------------------------------+     
class="type">void executeBuy() {
     class="type">class="kw">double ask = SymbolInfoDouble(_Symbol, SYMBOL_ASK);
     ask = NormalizeDouble(ask,_Digits);
     class="type">class="kw">double sl = ask*(class="num">1-slp);
     sl = NormalizeDouble(sl, _Digits);
     lot = calclots(ask*slp);
     trade.Buy(lot,_Symbol,ask,sl);
}
class=class="str">"cmt">//+------------------------------------------------------------------+
class=class="str">"cmt">//| Get VWAP function                                              |
class=class="str">"cmt">//+------------------------------------------------------------------+
class="type">class="kw">double getVWAP(class="type">int startCandle)
{
   class="type">class="kw">double sumPV = class="num">0.0;   class=class="str">"cmt">// Sum of(price * volume)
   class="type">long sumV = class="num">0.0;      class=class="str">"cmt">// Sum of volume
   class=class="str">"cmt">// Loop from the starting candle index down to class="num">1 (excluding current candle)
   for(class="type">int i = startCandle; i >= class="num">1; i--)
   {
      class=class="str">"cmt">// Calculate typical price: (High + Low + Close) / class="num">3
      class="type">class="kw">double high = iHigh(_Symbol, PERIOD_CURRENT, i);
      class="type">class="kw">double low = iLow(_Symbol, PERIOD_CURRENT, i);
      class="type">class="kw">double close = iClose(_Symbol, PERIOD_CURRENT, i);
      class="type">class="kw">double typicalPrice = (high + low + close) / class="num">3.0;
      class=class="str">"cmt">// Get volume and update sums
      class="type">long volume = iVolume(_Symbol, PERIOD_CURRENT, i);
      sumPV += typicalPrice * volume;
      sumV += volume;
   }
   class=class="str">"cmt">// Calculate VWAP or class="kw">return class="num">0 if no volume
   if(sumV == class="num">0)
      class="kw">return class="num">0.0;

   class="type">class="kw">double vwap = sumPV / sumV;
   class=class="str">"cmt">// Plot the dot

◍ 锚定会话起点与绘制 VWAP 标记

在 MT5 自定义指标里，把 VWAP 数值落进图表不能只靠返回 double，还得在当根 K 线时间上挂一个可视对象。下面这段把当前 Bar 时间取出来拼成唯一对象名，再用 OBJ_ARROW 画绿点，颜色、线型、宽度分别写死为 clrGreen、STYLE_DOT、1。 datetime currentBarTime = iTime(_Symbol, PERIOD_CURRENT, 0); // 取当前图表品种与周期的第0根Bar时间 string objName = "VWAP" + TimeToString(currentBarTime, TIME_MINUTES); // 用分钟级时间字符串避免重名 ObjectCreate(0, objName, OBJ_ARROW, 0, currentBarTime, vwap); // 在currentBarTime处、vwap价格上建箭头对象 ObjectSetInteger(0, objName, OBJPROP_COLOR, clrGreen); // 绿色 ObjectSetInteger(0, objName, OBJPROP_STYLE, STYLE_DOT); // 点状 ObjectSetInteger(0, objName, OBJPROP_WIDTH, 1); // 点大小为1 return vwap; // 把算好的VWAP交回调用处会话区间的界定靠 getSessionStartIndex：从 i=1 开始向后扫最多 1000 根 Bar，一旦某根本地时间 hour 等于 startHour 且 min 等于 30 就记为会话起点索引并 break。若你的品种每天 9:30 开盘，把 startHour 设为 9，就能抓到约 9:30 那根。 getBarShiftForTime 则反过来用：给定某日 day_start，把时分秒改写到目标 hour/minute，StructToTime 合成时间后丢给 iBarShift（PERIOD_M1, true），返回距现在的 M1 Bar 偏移量。配合 getUpperBand 里 iTime(_Symbol, PERIOD_D1, 0) 拿到的今日零点时间，可进一步定位 9:30 开盘价做通道上轨。外汇与贵金属杠杆高，这类时段锚定只描述概率性结构，实盘前请在策略测试器用历史数据核对偏移是否因节假日跳空而失效。

MQL5 / C++

class="type">class="kw">datetime currentBarTime = iTime(_Symbol, PERIOD_CURRENT, class="num">0);
class="type">class="kw">string objName = "VWAP" + TimeToString(currentBarTime, TIME_MINUTES);
ObjectCreate(class="num">0, objName, OBJ_ARROW, class="num">0, currentBarTime, vwap);
ObjectSetInteger(class="num">0, objName, OBJPROP_COLOR, clrGreen);    class=class="str">"cmt">// Green dot
ObjectSetInteger(class="num">0, objName, OBJPROP_STYLE, STYLE_DOT);  class=class="str">"cmt">// Dot style
ObjectSetInteger(class="num">0, objName, OBJPROP_WIDTH, class="num">1);          class=class="str">"cmt">// Size of the dot

class="kw">return vwap;
}
class=class="str">"cmt">//+------------------------------------------------------------------+
class=class="str">"cmt">//| Find the index of the candle corresponding to the session open   |
class=class="str">"cmt">//+------------------------------------------------------------------+
class="type">int getSessionStartIndex()
{
   class="type">int sessionIndex = class="num">1;
   class=class="str">"cmt">// Loop over bars until we find the session open
   for(class="type">int i = class="num">1; i <=class="num">1000; i++)
   {
      class="type">class="kw">datetime barTime = iTime(_Symbol, PERIOD_CURRENT, i);
      class="type">MqlDateTime dt;
      TimeToStruct(barTime, dt);
      
      if(dt.hour == startHour && dt.min == class="num">30)
      {
         sessionIndex = i;
         break;
      }
   }
      
   class="kw">return sessionIndex;
}
class=class="str">"cmt">//+------------------------------------------------------------------+
class=class="str">"cmt">//| Get the number of bars from now to market open                  |
class=class="str">"cmt">//+------------------------------------------------------------------+
class="type">int getBarShiftForTime(class="type">class="kw">datetime day_start, class="type">int hour, class="type">int minute) {
      class="type">MqlDateTime dt;
      TimeToStruct(day_start, dt);
      dt.hour = hour;
      dt.min = minute;
      dt.sec = class="num">0;
      class="type">class="kw">datetime target_time = StructToTime(dt);
      class="type">int shift = iBarShift(_Symbol, PERIOD_M1, target_time, true);
      class="kw">return shift;
}
class=class="str">"cmt">//+------------------------------------------------------------------+
class=class="str">"cmt">//| Get the upper Concretum band value                              |
class=class="str">"cmt">//+------------------------------------------------------------------+
class="type">class="kw">double getUpperBand() {
      class=class="str">"cmt">// Get the time of the current bar
      class="type">class="kw">datetime current_time = iTime(_Symbol, PERIOD_CURRENT, class="num">0);
      class="type">MqlDateTime current_dt;
      TimeToStruct(current_time, current_dt);
      class="type">int current_hour = current_dt.hour;
      class="type">int current_min = current_dt.min;
      
      class=class="str">"cmt">// Find today&class="macro">#x27;s opening price at class="num">9:class="num">30 AM
      class="type">class="kw">datetime today_start = iTime(_Symbol, PERIOD_D1, class="num">0);

「用14天波动锚定当日上轨」

要把「Concretum 上轨」画出来，核心是先拿到今天 9:30 那根 M1 柱的开盘价，再拿过去 14 天同一时段的相对波动做 sigma。若当天或历史数据里缺了 9:30 柱、或开盘价为 0，函数直接返回 0，避免在图表上喷无意义对象。下面这段是上轨计算的主干：先定位今日 9:30 柱，再循环回看 i=1 到 14 的日线起点，分别取历史日 9:30 开盘与当前时分收盘，算 abs(close/open-1) 累加。valid_days 不足时同样返回 0，只有拿到有效样本才用均值当 sigma，上轨即 open_930_today*(1+sigma)。

MQL5 / C++

class="type">int bar_at_930_today = getBarShiftForTime(today_start, class="num">9, class="num">30);
if (bar_at_930_today < class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// 无9:30柱则退出
class=class="str">"cmt">// 取今日9:class="num">30 M1开盘价
class=class="str">"cmt">// 取今日9:class="num">30 M1开盘价
class="type">class="kw">double open_930_today = iOpen(_Symbol, PERIOD_M1, bar_at_930_today);
if (open_930_today == class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// 价格无效则退出
class=class="str">"cmt">// 基于过去14天计算sigma
class="type">class="kw">double sum_moves = class="num">0;
class="type">int valid_days = class="num">0;
for (class="type">int i = class="num">1; i <= class="num">14; i++) {
  class="type">class="kw">datetime day_start = iTime(_Symbol, PERIOD_D1, i);
  class="type">int bar_at_930 = getBarShiftForTime(day_start, class="num">9, class="num">30);
  class="type">int bar_at_HHMM = getBarShiftForTime(day_start, current_hour, current_min);
  if (bar_at_930 < class="num">0 || bar_at_HHMM < class="num">0) class="kw">continue; class=class="str">"cmt">// 柱不存在则跳过
  class="type">class="kw">double open_930 = iOpen(_Symbol, PERIOD_M1, bar_at_930);
  class="type">class="kw">double close_HHMM = iClose(_Symbol, PERIOD_M1, bar_at_HHMM);
  if (open_930 == class="num">0) class="kw">continue; class=class="str">"cmt">// 开盘价无效则跳过
  class="type">class="kw">double move = MathAbs(close_HHMM / open_930 - class="num">1);
  sum_moves += move;
  valid_days++;
}
if (valid_days == class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// 无有效数据退出
class="type">class="kw">double sigma = sum_moves / valid_days;
class=class="str">"cmt">// 计算上轨
class="type">class="kw">double upper_band = open_930_today * (class="num">1 + sigma);
class=class="str">"cmt">// 在上轨位画蓝点
class="type">class="kw">string obj_name = "UpperBand_" + TimeToString(current_time, TIME_DATE|TIME_MINUTES|TIME_SECONDS);
ObjectCreate(class="num">0, obj_name, OBJ_ARROW, class="num">0, current_time, upper_band);
ObjectSetInteger(class="num">0, obj_name, OBJPROP_ARROWCODE, class="num">159); class=class="str">"cmt">// 圆点符号
ObjectSetInteger(class="num">0, obj_name, OBJPROP_COLOR, clrBlue);
ObjectSetInteger(class="num">0, obj_name, OBJPROP_WIDTH, class="num">2);
class="kw">return upper_band;
}

画点用 OBJ_ARROW 加 arrowcode 159（小圆点），染成 clrBlue、宽度 2，对象名拼了当前时间避免重名覆盖。外汇与贵金属属高风险品种，sigma 仅描述历史波动幅度，价格触及上轨后回落或突破均属概率事件，实盘前请在 MT5 用策略测试器跑一遍样本外日期。

MQL5 / C++

class="type">int bar_at_930_today = getBarShiftForTime(today_start, class="num">9, class="num">30);
if (bar_at_930_today < class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// Return class="num">0 if no class="num">9:class="num">30 bar exists
class="type">class="kw">double open_930_today = iOpen(_Symbol, PERIOD_M1, bar_at_930_today);
if (open_930_today == class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// No valid price

class=class="str">"cmt">// Calculate sigma based on the past class="num">14 days
class="type">class="kw">double sum_moves = class="num">0;
class="type">int valid_days = class="num">0;
for (class="type">int i = class="num">1; i <= class="num">14; i++) {
  class="type">class="kw">datetime day_start = iTime(_Symbol, PERIOD_D1, i);
  class="type">int bar_at_930 = getBarShiftForTime(day_start, class="num">9, class="num">30);
  class="type">int bar_at_HHMM = getBarShiftForTime(day_start, current_hour, current_min);
  if (bar_at_930 < class="num">0 || bar_at_HHMM < class="num">0) class="kw">continue; class=class="str">"cmt">// Skip if bars don&class="macro">#x27;t exist
  class="type">class="kw">double open_930 = iOpen(_Symbol, PERIOD_M1, bar_at_930);
  class="type">class="kw">double close_HHMM = iClose(_Symbol, PERIOD_M1, bar_at_HHMM);
  if (open_930 == class="num">0) class="kw">continue; class=class="str">"cmt">// Skip if no valid opening price
  class="type">class="kw">double move = MathAbs(close_HHMM / open_930 - class="num">1);
  sum_moves += move;
  valid_days++;
}
if (valid_days == class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// Return class="num">0 if no valid data
class="type">class="kw">double sigma = sum_moves / valid_days;

class=class="str">"cmt">// Calculate the upper band
class="type">class="kw">double upper_band = open_930_today * (class="num">1 + sigma);

class=class="str">"cmt">// Plot a blue dot at the upper band level
class="type">class="kw">string obj_name = "UpperBand_" + TimeToString(current_time, TIME_DATE|TIME_MINUTES|TIME_SECONDS);
ObjectCreate(class="num">0, obj_name, OBJ_ARROW, class="num">0, current_time, upper_band);
ObjectSetInteger(class="num">0, obj_name, OBJPROP_ARROWCODE, class="num">159); class=class="str">"cmt">// Dot symbol
ObjectSetInteger(class="num">0, obj_name, OBJPROP_COLOR, clrBlue);
ObjectSetInteger(class="num">0, obj_name, OBJPROP_WIDTH, class="num">2);

class="kw">return upper_band;
}

用14天波动算出当日下轨并标点

这段逻辑把「当天 9:30 开盘价」当作基准，再用过去 14 个交易日的同向时段波动幅度求平均，得到一条动态下轨。外汇与贵金属属高杠杆品种，这种统计带只反映历史波动概率，不预示方向。先取当前时间拆出小时和分钟，定位今天 D1 起点后找 9:30 那根 M1 bar 的开盘价；若找不到或价格为 0 直接返回 0，避免脏数据污染后续计算。

循环 i=1 到 14 遍历历史日，每天同样取 9:30 开盘与当前时刻收盘，算	收盘/开盘 - 1	的绝对值累加。valid_days 计数跳过无 bar 或开盘为 0 的日期，最终 sigma = 累加值 / valid_days；若 14 天全无效也返回 0。

下轨 = 今开 × (1 - sigma)，用 ObjectCreate 画一个 arrowcode 159 的红点，宽度 2，名字带时间戳防重名。你在 MT5 里把这段接进 OnCalculate，就能在子图看到随波动收缩或扩张的红色下轨点。

MQL5 / C++

  TimeToStruct(current_time, current_dt);
  class="type">int current_hour = current_dt.hour;
  class="type">int current_min = current_dt.min;
  
  class=class="str">"cmt">// Find today&class="macro">#x27;s opening price at class="num">9:class="num">30 AM
  class="type">class="kw">datetime today_start = iTime(_Symbol, PERIOD_D1, class="num">0);
  class="type">int bar_at_930_today = getBarShiftForTime(today_start, class="num">9, class="num">30);
  if (bar_at_930_today < class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// Return class="num">0 if no class="num">9:class="num">30 bar exists
  class="type">class="kw">double open_930_today = iOpen(_Symbol, PERIOD_M1, bar_at_930_today);
  if (open_930_today == class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// No valid price
  
  class=class="str">"cmt">// Calculate sigma based on the past class="num">14 days
  class="type">class="kw">double sum_moves = class="num">0;
  class="type">int valid_days = class="num">0;
  for (class="type">int i = class="num">1; i <= class="num">14; i++) {
    class="type">class="kw">datetime day_start = iTime(_Symbol, PERIOD_D1, i);
    class="type">int bar_at_930 = getBarShiftForTime(day_start, class="num">9, class="num">30);
    class="type">int bar_at_HHMM = getBarShiftForTime(day_start, current_hour, current_min);
    if (bar_at_930 < class="num">0 || bar_at_HHMM < class="num">0) class="kw">continue; class=class="str">"cmt">// Skip if bars don&class="macro">#x27;t exist
    class="type">class="kw">double open_930 = iOpen(_Symbol, PERIOD_M1, bar_at_930);
    class="type">class="kw">double close_HHMM = iClose(_Symbol, PERIOD_M1, bar_at_HHMM);
    if (open_930 == class="num">0) class="kw">continue; class=class="str">"cmt">// Skip if no valid opening price
    class="type">class="kw">double move = MathAbs(close_HHMM / open_930 - class="num">1);
    sum_moves += move;
    valid_days++;
  }
  if (valid_days == class="num">0) class="kw">return class="num">0; class=class="str">"cmt">// Return class="num">0 if no valid data
  class="type">class="kw">double sigma = sum_moves / valid_days;
  
  class=class="str">"cmt">// Calculate the lower band
  class="type">class="kw">double lower_band = open_930_today * (class="num">1 - sigma);
  
  class=class="str">"cmt">// Plot a red dot at the lower band level
  class="type">class="kw">string obj_name = "LowerBand_" + TimeToString(current_time, TIME_DATE|TIME_MINUTES|TIME_SECONDS);
  ObjectCreate(class="num">0, obj_name, OBJ_ARROW, class="num">0, current_time, lower_band);
  ObjectSetInteger(class="num">0, obj_name, OBJPROP_ARROWCODE, class="num">159); class=class="str">"cmt">// Dot symbol
  ObjectSetInteger(class="num">0, obj_name, OBJPROP_COLOR, clrRed);
  ObjectSetInteger(class="num">0, obj_name, OBJPROP_WIDTH, class="num">2);
  
  class="kw">return lower_band;

◍ 按止损点数反推手数的实盘函数

在 MT5 里做固定风险比例的仓位管理，核心是把「账户愿意亏多少」和「这笔单子止损占了几跳」绑在一起算。下面这个函数把风险百分比、止损点数、合约跳值一次性揉成可下单数，省得每次手动估手数。函数先取账户净值乘以 risk 百分比得到单次风险金额；再拿当前品种的 tick_size 与 tick_value，把止损点数的 slpoints 换算成「每最小步长要亏多少美元」。用风险额除以单步亏损，向下取整到 lotstep 倍数，就是理论手数。最后两行把算出的 lots 夹在品种允许的 SYMBOL_VOLUME_MIN 与 SYMBOL_VOLUME_MAX 之间，避免报「无效交易量」错误。外汇与贵金属杠杆高，错手数可能瞬间吃掉本金，开仓前务必在策略测试器用真实点差跑一遍。回测样本里该仓位模块配合趋势过滤，风险参数 risk=1 时样本斜率 α=1.6562、截距 β=-0.1183，说明手数随止损拉长而线性收缩，但实际滑点未计入，实盘概率上会偏乐观。

MQL5 / C++

class="type">class="kw">double calclots(class="type">class="kw">double slpoints) {
   class="type">class="kw">double riskAmount = AccountInfoDouble(ACCOUNT_BALANCE) * risk / class="num">100;
   class="type">class="kw">double ticksize = SymbolInfoDouble(_Symbol, SYMBOL_TRADE_TICK_SIZE);
   class="type">class="kw">double tickvalue = SymbolInfoDouble(_Symbol, SYMBOL_TRADE_TICK_VALUE);
   class="type">class="kw">double lotstep = SymbolInfoDouble(_Symbol, SYMBOL_VOLUME_STEP);
   class="type">class="kw">double moneyperlotstep = slpoints / ticksize * tickvalue * lotstep;
   class="type">class="kw">double lots = MathFloor(riskAmount / moneyperlotstep) * lotstep;
   lots = MathMin(lots, SymbolInfoDouble(_Symbol, SYMBOL_VOLUME_MAX));
   lots = MathMax(lots, SymbolInfoDouble(_Symbol, SYMBOL_VOLUME_MIN));
   class="kw">return lots;
}

「组合分散化不是堆数量而是看相关性」

三种源自股市开盘区间突破的变体策略，回测都验证了能盈利，但单拎出来最大回撤都超过15%。把它们按一致风险分配凑成一个组合后，回测最大回撤压到约10%，净值和回撤曲线能在不同时间段互相抵消——这是算法交易里分散化降低整体风险的直接证据。必须明确，分散化不是越多越好，无相关性才是核心。像所有加密资产上跑同一套策略就没意义，广义上它们高度同向。回测里的相关性也看周期，日收益和月收益算出来能差很多；遇市场结构突变，策略间相关性会扭曲，所以有人更信实盘相关性相对回测的偏移来判断优势是否衰减。拿这三种策略的月度收益算相关性矩阵，平均约0.3，低于0.5算可接受但不够理想。它们都带多空、却还是正相关，根子在同一标的资产上做相似逻辑。3000多笔样本下，回撤收益按月分布均匀，说明没极端环境捣乱，但也提示：这组合同类度高，真要扩敞口得搭外部不同策略，而不是内部再加同类。外汇和贵金属这类高杠杆品种若套用思路，先想清楚跨品种相关性是不是真低，否则分散只是账面安慰，实盘一遇联储决议或流动性断档，相关性瞬间拉满，回撤可能远超回测的10%。

画得少，看得清

前面三套开盘区间突破 EA 把信号规则、MQL5 源码和回测统计都摊开了，核心结论其实就一句：单策略在 NQ 一类期货上默认参数能跑出文献结果，但外汇盘用模拟账户数据复现，ORB1 偶尔对齐、ORB2 与 ORB3 多半对不上——评论区已有四位读者在 2025年10月到2026年2月间报告同类偏差。分散化才是原文真正落点：把三套逻辑不相关的 ORB 摞进一个组合，回测波动比单跑任一套更平，但外汇与贵金属本身杠杆高、滑点跳空频繁，组合也只是降低而非消除爆仓概率。真要动手，先把 ORB1.mq5 拖进 MT5 用期货数据验一遍，再换你的经纪商外汇 tick 看偏移多少；有空就把 Wu簡單改的那版「指定时间开仓后持 N 根 bar 平仓」逻辑移植到 ORB2 试水，比盲信回测报告实在。

让小布替你跑这套时区映射

这些诊断小布盯盘的AIGC已内置，打开对应品种页即可看到服务器时间与美东开盘的自动换算，以及VWAP和β的实时估算，把重复劳动交给小布，你专注决策。

←

解密开盘区间突破（ORB）日内交易策略·进阶篇

常见问题

原文研究用美东9:30-16:00，对应UTC+2/3的18:30-24:00，测试时须按自己经纪商时区平移，否则信号会整体错位。

QQQ是ETF不可直接交易，USTEC是带杠杆的CFD，波动与跳空特征不同，β放大后原优势可能被磨损，需重测参数。

α揭示策略相对基准的超额收益即真实优势，β衡量对大盘波动的敏感度，两者结合可判断方向性偏差与分散空间。

小布盯盘内置了品种页的VWAP、β估算与开盘区间标记，可省去手动计算，直接观察多标的分散效果。

VWAP常作为突破过滤或回踩参考，用来区分假突破与有效动量，尤其在高β资产上能压制噪声。

解密开盘区间突破（ORB）日内交易策略·综合运用

VWAP与买卖执行的代码骨架

◍ 锚定会话起点与绘制 VWAP 标记

「用14天波动锚定当日上轨」

用14天波动算出当日下轨并标点

◍ 按止损点数反推手数的实盘函数

「组合分散化不是堆数量而是看相关性」

画得少，看得清

常见问题

相关阅读