在聚宽平台上编写海龟交易法则

import jqdata
import math
import numpy as np
import pandas as pd
from collections import deque

def initialize(context):
    
    set_option('use_real_price', True)
    set_order_cost(OrderCost(close_tax=0.001, open_commission=0.0003, close_commission=0.0003, min_commission=5), type='stock')

    g.security = '000060.XSHE'
    set_benchmark(g.security)
    g.in_day = 20
    g.out_day = 10
    g.today_units = 0
    g.current_units = 0
    g.N=deque(maxlen=19)
    g.current_N = 0
    g.last_buy_price = 0
    
    price = attribute_history(g.security, g.N.maxlen*2+1, '1d', ('high', 'low', 'close'))
    
    for i in range(g.N.maxlen+1, g.N.maxlen*2+1):
        li = []
        for j in range(i-19,i+1):
            a = price['high'][j]-price['low'][j]
            b = abs(price['high'][j]-price['close'][j-1])
            c = abs(price['low'][j]-price['close'][j-1])
            li.append(max(a,b,c))
        current_N = np.array(li).mean()
        g.N.append(current_N)
        
    
def before_trading_start(context):
    g.current_N = cal_N()
    g.today_units = 0

    
def handle_data(context, data):
    dt = context.current_dt
    current_price = data[g.security].price #上一分钟价格
    value = context.portfolio.total_value
    cash = context.portfolio.available_cash
    
    unit = math.floor(value * 0.01 / g.current_N)
    
        
    if g.current_units == 0:
        buy(current_price, cash, unit)
    else:
        if stop_loss(current_price):
            return
        if sell(current_price):
            return
        addin(current_price, cash, unit)
    
def cal_N():
    # if len(g.N) < g.N.maxlen:
    #     price = attribute_history(g.security, g.N.maxlen+2, '1d', ('high', 'low', 'close'))
    #     li = []
    #     for i in range(1, g.N.maxlen+2):
    #         a = price['high'][i]-price['low'][i]
    #         b = abs(price['high'][i]-price['close'][i-1])
    #         c = abs(price['low'][i]-price['close'][i-1])
    #         li.append(max(a,b,c))
    #     current_N = np.array(li).mean()
    # else:
    price = attribute_history(g.security, 2, '1d', ('high', 'low', 'close'))
    a = price['high'][1]-price['low'][1]
    b = abs(price['high'][1]-price['close'][0])
    c = abs(price['low'][1]-price['close'][0])
    current_N = (max(a,b,c) + np.array(g.N).sum())/(g.N.maxlen+1)
    g.N.append(current_N)
    return current_N
    
def buy(current_price, cash, unit):
    price = attribute_history(g.security, g.in_day, '1d', ('close',))
    if current_price > max(price['close']):
        shares = cash / current_price
        if shares >= unit:
            print("buying %d" % unit)
            o = order(g.security, unit)
            g.last_buy_price = o.price
            g.current_units += 1
            g.today_units += 1
            return True
    return False
            

def addin(current_price, cash, unit):
    if current_price >= g.last_buy_price + 0.5 * g.current_N:
        shares = cash / current_price
        if shares >= unit:
            print("adding %d" % unit)
            o = order(g.security, unit)
            g.last_buy_price = o.price
            g.current_units += 1
            g.today_units += 1
            return True
    return False
            
def sell(current_price):
    price = attribute_history(g.security, g.out_day, '1d', ('close',))
    if current_price < min(price['close']):
        print("selling")
        order_target(g.security, 0)
        g.current_units = g.today_units
        return True
    return False
        
def stop_loss(current_price):
    if current_price < g.last_buy_price - 2 * g.current_N:
        print("stop loss")
        order_target(g.security, 0)
        g.current_units = g.today_units
        return True
    return False