基于参考点查找指数下一个高低点的代码问题排查

需求说明 #

编写两个方法，根据指定交易对的标记价格，基于参考点查找下一组4根K线中的最高高点或最低低点。

示例说明 #

基础参考数组 #

Min Array - [2, 3.5, 5, 8.5, 9]
Max Array - [20, 17, 9, 7, 5]

输入输出示例 #

输入：标记价格 3.6

预期输出
最低低点 - 3.5
最高高点 - 5

输入：标记价格 5.1

预期输出
最低低点 - 5
最高高点 - 7

输入：标记价格 3.4

预期输出
最低低点 - 2
最高高点 - 5

当前问题 #

方法仅能单向移动索引，例如标记价格跌破$2.50时，索引会指向后续值，但当价格再次突破$2.50时，索引无法回退到之前的位置。

说明：ohlcv用于获取交易对的开/高/低/收数据，candle[2]为当前K线最高价，candle[3]为最低价。标记价格突破指数高点时，索引递减（引用前一组4根K线找下一个高点，数组按从高到低排序）；突破指数低点时则相反。

原代码 #

def compare_hl(ohlcv, mark, symbol):
    minarray = []
    maxarray = []

    low_index = 0
    high_index = 0
    # Extract the lows of all candles
    lows = [candle[3] for candle in ohlcv]
    highs = [candle[2] for candle in ohlcv]

    # Initialize variables to track the starting index of the previous four bars

    while low_index < len(lows):
        # Extract the lows of the previous four bars
        previous_lows = lows[low_index:low_index + 4]

        # Check if the mark price falls below the lows of the previous four bars
        if mark < min(previous_lows):
            # Update the starting index for the next set of previous four bars
            low_index += 4        
        elif mark > min(previous_lows) and low_index < -4:
            print("\nMark Price above Minimum of Previous Lows. Reverting to Previous Set.")
            low_index -= 4
            break
        else:
            # If mark price is not below the lows, move to the next set of four bars
            low_index += 4
        minarray.append(min(previous_lows))

    while high_index < len(highs):
        # Extract the lows of the previous four bars
        previous_highs = highs[high_index:high_index + 4]

        # Check if the mark price falls below the lows of the previous four bars
        if mark > max(previous_highs):
            # Update the starting index for the next set of previous four bars
            high_index += 4       
        elif mark < max(previous_highs) and high_index < -4:      
            high_index -= 4
            break
        else:
            # If mark price is not below the lows, move to the next set of four bars
            high_index += 4
        maxarray.append(max(previous_highs))

    return minarray, maxarray

def find_reference(mark_price, minarray, maxarray, lo_index, hi_index, lowref, hiref, crossbelow, crossabove):

    if mark_price > maxarray[hi_index]:
        crossabove = True
    elif mark_price < minarray[lo_index]:
        crossbelow = True

    if mark_price > maxarray[hi_index]:
        print(colors.CYAN + "Mark Max: ", str(mark_price), str(maxarray[hi_index]) + colors.END)
        print(colors.CYAN + "Mark has crossed above previous 4-bar Max.\nMark: ", str(mark_price),"\nMin: ", str(minarray[lo_index]) + colors.END)
        hi_index -= 1
        highreference = maxarray[hi_index]
        crossabove = True
        crossbelow = False
        print(colors.CYAN + "New High Reference Set: ", str(maxarray[hi_index]) + colors.END)
    elif hi_index <= -2 and mark_price < maxarray[hi_index + 1]:
        if not crossabove:
            print(colors.CYAN + "Mark has crossed below last high. Adjusting reference." + colors.END)
            hi_index += 1
            highreference = maxarray[hi_index]
            print(colors.CYAN + "New High Reference Set: ", str(maxarray[hi_index]) + colors.END)
            crossabove = False
            crossbelow = True

    if lo_index >= -len(minarray) and mark_price < minarray[lo_index]:
        print(colors.CYAN + "Mark has crossed below Min: ", str(mark_price), str(minarray[lo_index]) + colors.END)
        lo_index -= 1
        lowreference = minarray[lo_index]
        crossabove = False
        crossbelow = True
        print(colors.CYAN + "New Low Reference Set: ", str(minarray[lo_index]) + colors.END)
    elif lo_index <= -2 and mark_price > minarray[lo_index + 1]:
        if not crossbelow:
            print(colors.CYAN + "Mark has crossed above last min. Adjusting reference." + colors.END)
            lo_index += 1
            lowreference = minarray[lo_index]
            print(colors.CYAN + "New Low Reference Set: ", str(minarray[lo_index]) + colors.END)
            crossabove = True
            crossbelow = False

    lowreference = minarray[lo_index]
    highreference = maxarray[hi_index]

    return lowreference, highreference, lo_index, hi_index

问题根源 #

1. compare_hl方法逻辑错误：

   • 生成参考数组时，elif条件low_index < -4和high_index < -4永远无法触发（初始索引为0，循环中仅累加4，索引始终非负），导致无法回退。
   • 循环逻辑混乱，生成的数组并非按4根K线分组的有效极值序列，而是无意义的遍历结果。

2. find_reference方法缺陷：

   • 索引回退依赖crossabove/crossbelow状态，但状态初始化和更新逻辑不严谨，反向突破时无法触发回退。
   • 索引边界判断限制过严（如hi_index <= -2），导致部分场景下无法正常调整索引。
   • 强制覆盖lowreference/highreference的逻辑会忽略正确赋值，导致结果错误。

修复方案 #

1. 修正compare_hl方法：生成正确的分组极值数组 #

按4根K线为一组，依次计算每组极值，反转数组以匹配"从高到低"的索引需求：

def compare_hl(ohlcv):
    minarray = []
    maxarray = []
    lows = [candle[3] for candle in ohlcv]
    highs = [candle[2] for candle in ohlcv]
    
    # 按4根K线分组计算极值
    for i in range(0, len(lows), 4):
        group_lows = lows[i:i+4]
        group_highs = highs[i:i+4]
        if group_lows:
            minarray.append(min(group_lows))
            maxarray.append(max(group_highs))
    
    # 反转数组，匹配索引从高到低的需求
    minarray.reverse()
    maxarray.reverse()
    return minarray, maxarray

2. 修正find_reference方法：实现双向索引调整 #

基于价格突破方向，支持索引双向移动，简化状态判断逻辑：

def find_reference(mark_price, minarray, maxarray, lo_index, hi_index):
    # 调整高点索引
    if mark_price > maxarray[hi_index]:
        if hi_index > 0:
            hi_index -= 1
            print(f"标记价格突破当前高点，更新高点参考为: {maxarray[hi_index]}")
    elif mark_price < maxarray[hi_index + 1] and hi_index < len(maxarray) - 1:
        hi_index += 1
        print(f"标记价格回落至上一高点区间，更新高点参考为: {maxarray[hi_index]}")
    
    # 调整低点索引
    if mark_price < minarray[lo_index]:
        if lo_index > 0:
            lo_index -= 1
            print(f"标记价格突破当前低点，更新低点参考为: {minarray[lo_index]}")
    elif mark_price > minarray[lo_index + 1] and lo_index < len(minarray) - 1:
        lo_index += 1
        print(f"标记价格回升至上一低点区间，更新低点参考为: {minarray[lo_index]}")
    
    return minarray[lo_index], maxarray[hi_index], lo_index, hi_index

3. 使用示例 #

# 模拟ohlcv数据（对应示例中的Min/Max Array）
ohlcv = [
    [0,0,20,2],
    [0,0,17,3.5],
    [0,0,9,5],
    [0,0,7,8.5],
    [0,0,5,9]
]

minarray, maxarray = compare_hl(ohlcv)
lo_idx, hi_idx = 0, 0

# 测试标记价格3.6
low_ref, high_ref, lo_idx, hi_idx = find_reference(3.6, minarray, maxarray, lo_idx, hi_idx)
print(f"最低低点 - {low_ref}")  # 输出3.5
print(f"最高高点 - {high_ref}")  # 输出5

# 测试标记价格5.1
low_ref, high_ref, lo_idx, hi_idx = find_reference(5.1, minarray, maxarray, lo_idx, hi_idx)
print(f"最低低点 - {low_ref}")  # 输出5
print(f"最高高点 - {high_ref}")  # 输出7

# 测试标记价格3.4
low_ref, high_ref, lo_idx, hi_idx = find_reference(3.4, minarray, maxarray, lo_idx, hi_idx)
print(f"最低低点 - {low_ref}")  # 输出2
print(f"最高高点 - {high_ref}")  # 输出5

内容的提问来源于stack exchange，提问作者Zach S