如何用 Python 自动识别股票的支撑位与压力位？

from alphafeed import AlphaFeed
import pandas as pd
import numpy as np

af = AlphaFeed()  # 读取 ALPHAFEED_API_KEY

df = af.klines.get("600519.SH", period="1d", count=250,
                   adjust="forward", to_dataframe=True)
df = df.sort_values("trade_date").reset_index(drop=True)

def swing_points(df, k=5):
    highs, lows = [], []
    h, l = df["high"].to_numpy(), df["low"].to_numpy()
    for i in range(k, len(df) - k):
        if h[i] == max(h[i - k:i + k + 1]):
            highs.append((df["trade_date"].iloc[i], h[i]))
        if l[i] == min(l[i - k:i + k + 1]):
            lows.append((df["trade_date"].iloc[i], l[i]))
    return highs, lows

highs, lows = swing_points(df, k=5)
print("摆动高点数:", len(highs), " 摆动低点数:", len(lows))

def cluster_levels(prices, tol=0.02):
    """把相差在 tol(默认2%) 以内的价位聚成一档，返回(代表价, 触及次数)。"""
    prices = sorted(prices)
    clusters = []
    for p in prices:
        if clusters and abs(p - clusters[-1][0]) / clusters[-1][0] <= tol:
            arr = clusters[-1][1]; arr.append(p)
        else:
            clusters.append((p, [p]))
    return [(round(np.mean(arr), 2), len(arr)) for _, arr in clusters]

res_levels = cluster_levels([p for _, p in highs], tol=0.02)   # 压力
sup_levels = cluster_levels([p for _, p in lows], tol=0.02)    # 支撑

price = df["close"].iloc[-1]
above = sorted([(lv, n) for lv, n in res_levels if lv > price])[:3]
below = sorted([(lv, n) for lv, n in sup_levels if lv < price], reverse=True)[:3]
print(f"现价 {price:.2f}")
print("上方压力(价, 触及次数):", above)
print("下方支撑(价, 触及次数):", below)

bins = pd.cut(df["close"], bins=20)
vol_profile = df.groupby(bins, observed=True)["volume"].sum().sort_values(ascending=False)
top_zones = [round(iv.mid, 2) for iv in vol_profile.head(3).index]
print("成交密集价位(前3):", top_zones)