Ce qu’on va faire :

Charger les données

PYTHON 
import yfinance as yf
from lightweight_charts import JupyterChart

TICKER = 'EURUSD=X'
INTERVAL = '3mo'
TIME_FRAME = '1h'

df = yf.download(TICKER, period=INTERVAL, interval=TIME_FRAME, auto_adjust=True)
df = df[['Open','High','Low','Close','Volume']]
df.columns = ['open','high','low','close','volume']
df['time'] = df.index.strftime('%Y-%m-%d %H:%M:%S')
df.reset_index(drop=True, inplace=True)

Dans ce bloc, on récupère les données historiques de l’EUR/USD sur les 3 derniers mois avec des bougies d’une heure grâce à yfinance. Ensuite, on prépare le DataFrame pour lightweight-charts-python en renommant les colonnes et en ajoutant une colonne time au format compatible.

Détecter les FVG

PYTHON 
MITIGATION = 0.5

def detect_fvg(df):
    current = df
    previous2 = df.shift(2)

    mitigation_bullish = current['low'] - ((current['low'] - previous2['high'])*MITIGATION)
    mitigation_bearish = current['high'] + ((previous2['low'] - current['high'])*MITIGATION)

    future_highs = df['high'][::-1].cummax()[::-1]
    future_lows  = df['low'][::-1].cummin()[::-1]

    mitigation_not_hit_bullish = (future_lows > mitigation_bullish)
    mitigation_not_hit_bearish = (future_highs < mitigation_bearish)

    fvg_bullish_start = previous2[(previous2['high'] < current['low']) & mitigation_not_hit_bullish]
    fvg_bullish_end   = current[(current['high'].shift(2) < current['low']) & mitigation_not_hit_bullish]

    fvg_bearish_start = previous2[(previous2['low'] > current['high']) & mitigation_not_hit_bearish]
    fvg_bearish_end   = current[(current['low'].shift(2) > current['high']) & mitigation_not_hit_bearish]

    return fvg_bullish_start, fvg_bullish_end, fvg_bearish_start, fvg_bearish_end

bull_start, bull_end, bear_start, bear_end = detect_fvg(df)

Ici, on détecte les Fair Value Gaps (FVG) bullish et bearish. Le MITIGATION = 0.5 sert à filtrer les FVG partiellement comblés. Concrètement si le prix suivant comble 50 % ou plus de la zone du FVG, celle-ci n’est plus considérée comme valide et disparait du graphique.

Les FVG avec leur mitigation à 50 % en bleu

Les FVG avec leur mitigation à 50 % en bleu

Afficher le graphique

PYTHON 
df = load_data()
bull_start, bull_end, bear_start, bear_end = detect_fvg(df)
chart = JupyterChart(width=1200, height=600, toolbox=True)
chart.set(df)
draw_fvg(chart, df, bull_start, bull_end, bear_start, bear_end)
chart.load()
⚠️

Ce code est conçu pour les notebooks (Jupyter, Colab…). L’affichage interactif avec JupyterChart fonctionne directement dans la cellule, mais ne sera pas identique dans un IDE classique. Consulte l’article sur lightweight-charts pour en savoir plus.

Graphique avec les FVG

Graphique avec les FVG

Voici le code complet pour afficher le graphique interactif avec les FVG dans un notebook.

PYTHON 
import yfinance as yf
from lightweight_charts import JupyterChart

TICKER = 'EURUSD=X'
INTERVAL = '3mo'
TIME_FRAME = '1h'
MITIGATION = 0.5

def load_data():
    df = yf.download(TICKER, period=INTERVAL, interval=TIME_FRAME, auto_adjust=True)
    df = df[['Open','High','Low','Close','Volume']]
    df.columns = ['open','high','low','close','volume']
    df['time'] = df.index.strftime('%Y-%m-%d %H:%M:%S')
    df.reset_index(drop=True, inplace=True)
    return df[['time','open','high','low','close','volume']]

def detect_fvg(df):
    current = df
    previous2 = df.shift(2)
    mitigation_bullish = current['low'] - ((current['low'] - previous2['high'])*MITIGATION)
    mitigation_bearish = current['high'] + ((previous2['low'] - current['high'])*MITIGATION)
    future_highs = df['high'][::-1].cummax()[::-1]
    future_lows  = df['low'][::-1].cummin()[::-1]
    mitigation_not_hit_bullish = (future_lows > mitigation_bullish)
    mitigation_not_hit_bearish = (future_highs < mitigation_bearish)
    fvg_bullish_start = previous2[(previous2['high'] < current['low']) & mitigation_not_hit_bullish]
    fvg_bullish_end   = current[(current['high'].shift(2) < current['low']) & mitigation_not_hit_bullish]
    fvg_bearish_start = previous2[(previous2['low'] > current['high']) & mitigation_not_hit_bearish]
    fvg_bearish_end   = current[(current['low'].shift(2) > current['high']) & mitigation_not_hit_bearish]
    return fvg_bullish_start, fvg_bullish_end, fvg_bearish_start, fvg_bearish_end

def draw_fvg(chart, df, bull_start, bull_end, bear_start, bear_end):
    last_time = df['time'].iloc[-1]
    for (_, start_row), (_, end_row) in zip(bull_start.iterrows(), bull_end.iterrows()):
        chart.box(start_row['time'], start_row['high'], last_time, end_row['low'],
                  color="#00ff1175", fill_color="#00ff1136", width=2)
    for (_, start_row), (_, end_row) in zip(bear_start.iterrows(), bear_end.iterrows()):
        chart.box(start_row['time'], start_row['low'], last_time, end_row['high'],
                  color="#dd00007a", fill_color="#8f00002e", width=2)

df = load_data()
bull_start, bull_end, bear_start, bear_end = detect_fvg(df)
chart = JupyterChart(width=1200, height=600, toolbox=True)
chart.set(df)
draw_fvg(chart, df, bull_start, bull_end, bear_start, bear_end)
chart.load()

N’hésite pas à t’inscrire à la newsletter pour recevoir des insights algorithmiques exclusifs. À bientôt !