Computing technical indicators from K-line data

1. Get the data first (free endpoint)

`/v1/kline` is a free endpoint (no Key). Returns forward-adjusted daily/weekly/monthly bars:

import requests
import pandas as pd

r = requests.get("https://api.ashareapi.com/v1/kline",
                 params={"code": "sh600667", "period": "day", "count": 60},
                 timeout=30)
bars = r.json()["data"]              # list[dict]
df = pd.DataFrame(bars)
print(df[["date", "open", "last", "high", "low", "volume"]].head())

• Forward-adjusted: adjusted for dividends/splits (no gap on ex-dividend days). To get raw prices, adjust back yourself (not covered here)
• `last` is the close, not the live price — computing indicators off the close is standard

2. MA (simple moving average, the most basic)

MA(N) = mean of last N closes. It is the "smoothing filter" for trend and an input to other indicators.

def ma(close, n):
    return close.rolling(n).mean()

df["ma5"]  = ma(df["last"], 5)
df["ma10"] = ma(df["last"], 10)
df["ma20"] = ma(df["last"], 20)
df["ma60"] = ma(df["last"], 60)
print(df[["date", "last", "ma5", "ma20"]].tail())

• Golden cross: MA5 crosses above MA20 (short turns stronger)
• Death cross: MA5 crosses below MA20 (weaker)
• Bull alignment: MA5 > MA10 > MA20 (uptrend); bear alignment the reverse

3. MACD (trend + momentum, most used)

MACD = difference of fast/slow EMAs (DIF) smoothed again (DEA), standard 12/26/9.

def macd(close, fast=12, slow=26, sig=9):
    ema_f = close.ewm(span=fast).mean()
    ema_s = close.ewm(span=slow).mean()
    dif = ema_f - ema_s                     # fast - slow
    dea = dif.ewm(span=sig).mean()          # EMA of DIF
    hist = (dif - dea) * 2                  # bar (common ×2)
    return dif, dea, hist

df["dif"], df["dea"], df["macd"] = macd(df["last"])

• Golden cross: DIF crosses above DEA (momentum turns positive); death cross: below
• Bar: red growing = bullish momentum, green growing = bearish
• Divergence: price new high but MACD not (top divergence, warning); price new low but MACD not (bottom divergence)

4. RSI (overbought / oversold)

RSI(N) measures recent up/down strength, used to spot overbought/oversold. Classic 14.

def rsi(close, n=14):
    delta = close.diff()
    gain = delta.clip(lower=0).rolling(n).mean()
    loss = (-delta.clip(upper=0)).rolling(n).mean()
    rs = gain / loss
    return 100 - 100 / (1 + rs)

df["rsi14"] = rsi(df["last"])

• > 70 usually overbought (watch for pullback); < 30 oversold
• Divergence applies too: price new high but RSI not = top divergence
• Not absolute: in strong trends RSI can stay overbought a long time (not necessarily a pullback)

5. KDJ (stochastic, short-term favorite)

KDJ uses high/low/close relative position; sensitive but noisy. Classic 9,3,3.

def kdj(high, low, close, n=9):
    low_n = low.rolling(n).min()
    high_n = high.rolling(n).max()
    rsv = (close - low_n) / (high_n - low_n) * 100
    k = rsv.ewm(alpha=1/3).mean()
    d = k.ewm(alpha=1/3).mean()
    j = 3 * k - 2 * d
    return k, d, j

df["k"], df["d"], df["j"] = kdj(df["high"], df["low"], df["last"])

• K crosses above D = golden cross (short-term stronger); below = death cross
• > 80 overbought / < 20 oversold, more sensitive than RSI
• J value can exceed 0/100 (extremes), used to detect saturation

6. BOLL (Bollinger Bands, volatility range)

BOLL = mid-band MA20 ± 2×std, describing the "normal range" of price.

def boll(close, n=20, k=2):
    mid = close.rolling(n).mean()
    std = close.rolling(n).std()
    upper = mid + k * std
    lower = mid - k * std
    return upper, mid, lower

df["boll_up"], df["boll_mid"], df["boll_low"] = boll(df["last"])

• Squeeze (band narrows) = volatility about to expand (pre-breakout); expansion = volatility up
• Touch upper band = strong (can ride along in a trend); touch lower band = weak
• Break below lower then reclaim = possible oversold bounce; break above upper = possible acceleration (but in a strong trend price rides the band, not necessarily falling back)

7. Combining the five (more is not better)

• Trend + momentum: MA (direction) + MACD (momentum) — the most common combination
• Overbought/oversold: RSI / KDJ (pick one; both short-term, noisy)
• Volatility range: BOLL (gives MA a "where it should be" band)
• Discipline: indicators are confirmation aids, not standalone buy/sell. Two or more agreeing + your own logic is more reliable (this guide only covers data retrieval and computation, not investment advice)

8. Common misreads (important)

• ① Adjustment basis: the API returns forward-adjusted — computing indicators off it is correct (keeps continuity). But compare against raw price when reconciling with your broker
• ② Parameters are configurable: MACD(12,26,9) / RSI(14) / KDJ(9,3,3) / BOLL(20,2) are "common", not "only". Changing them = a different indicator, possibly different conclusion
• ③ Technical indicators lag: they are all historical averages/smoothing — they reflect the past. In strong trends they can saturate; do not mechanically trade on crosses