Optimization function and routine

pyrenko_optimization.py

# Max index of train data
train_ind = round(stocks_close.shape[0] * 0.7)

# Function for optimization
def evaluate_renko(brick, history, column_name):
    renko_obj = pyrenko.renko()
    renko_obj.set_brick_size(brick_size = brick, auto = False)
    renko_obj.build_history(prices = history)
    return renko_obj.evaluate()[column_name]

def scoring_stocks(method = 'optimization'):
    # Create result dataframe
    stock_result = pd.DataFrame(columns = ['brick_size', 'balance', 'price_ratio', 'score'], 
                            index = stocks_close.columns)

    # Optimization and evaluation of each stock
    for s in stock_result.index:
        # Getting bounds by ATR
        atr = talib.ATR(high = np.double(stocks_high[s][:train_ind]),
                       low = np.double(stocks_low[s][:train_ind]),
                       close = np.double(stocks_close[s][:train_ind]),
                       timeperiod = 14)
        atr = atr[np.isnan(atr) == False]
    
        # Brick size maximization
        opt_bs = 0.0
        if method == 'optimization':
            opt_bs = opt.fminbound(lambda x: -evaluate_renko(brick = x, 
                                                            history = stocks_close[s][:train_ind], column_name = 'score'), 
                                  np.min(atr), np.max(atr), disp=0)
    
        # Test data
        data = stocks_close[s][train_ind:]
    
        # Build renko chart and evaluation
        renko_obj = pyrenko.renko()
        if method == 'optimization':
            renko_obj.set_brick_size(brick_size = opt_bs, auto = False)
        else:
            renko_obj.set_brick_size(brick_size = atr[-1], auto = False)
        renko_obj.build_history(prices = data)
        renko_res = renko_obj.evaluate()
    
        # Save result to dataframe
        if method == 'optimization':
            stock_result.loc[s]['brick_size'] = opt_bs
        else:
            stock_result.loc[s]['brick_size'] = atr[-1]
        stock_result.loc[s]['balance', 'price_ratio', 'score'] = renko_res
    
    return stock_result

atr_result = scoring_stocks(method = 'atr')
opt_result = scoring_stocks(method = 'optimization')