使用python实现一个简单的数字货币交易回测系统

Posted 2021-09-01 linchaolong

目录

• BaseStrategy（策略基类）
• Broker（经纪人）
• 示例代码
• 数据可视化

BaseStrategy（策略基类）

BaseStrategy，交易策略基类

回调函数

• on_start：策略开始运行
• on_stop：策略运行结束
• next_bar：回测收到新的K线时调用

其他函数

• record：记录自定义数据
• output_record：输出数据记录文件

Broker（经纪人）

Broker 经纪人，负责处理处理撮合交易订单等功能.

交易相关函数

• pos：当前仓位
• cancel_all：取消所有订单
• buy：做多
• sell：平多
• short：做空
• cover：平空
• create_stop_order：创建止盈止损订单

示例代码

行情数据这里用的是币安上爬取的分钟数据，关于数据的爬取和整理可以看这里。

    from common.time_utils import timestamp_to_datetime

    # 读取分钟数据
    df = pd.read_csv('ETHUSDT-1m.csv', converters={
        'Open time': timestamp_to_datetime,
        'Close time': timestamp_to_datetime
    })

    # 数据清洗
    df.rename(columns={
        'Open time': 'open_time',
        'Close time': 'close_time',
        'Open': 'open',
        'High': 'high',
        'Low': 'low',
        'Close': 'close',
        'Volume': 'volume',
    }, inplace=True)

    # 截取指定时间范围的数据
    df = df[df['open_time'] >= '2021-05-01']
    df = df[df['close_time'] <= '2021-06-01']

    df.reset_index(inplace=True, drop=True)
    # print(df)

    broker = Broker()
    broker.set_symbol('ETHUSDT')
    broker.set_strategy(TripleFilterTradeSystemStrategy) # 设置策略类
    broker.set_leverage(1.0)  # 杠杆比例
    broker.set_cash(3600)  # 1初始资金.
    broker.set_commission(7 / 10000)  # 手续费
    broker.set_backtest_data(df)  # 数据.
    broker.run()
    broker.calculate().to_csv('triple_filter_trade_system_backtest.csv', index=False)
    broker.output_record('triple_filter_trade_system_record.csv')

    # 参数优化， 穷举法， 遗传算法。
    # broker.optimize_strategy(long_period=[i for i in range(30, 60, 5)], short_period=[i for i in range(5, 30, 1)])

数据可视化

策略中通过record方法记录了行情数据和交易信号，数据如下：

使用bokeh将K线数据、交易信号和相关技术指标可视化

from math import pi
from bokeh.plotting import figure
import numpy as np
from common.indicator import EMA
import pandas as pd
import talib
from bokeh.layouts import column
from bokeh.io import output_file, show, save
from bokeh.models import ColumnDataSource, HoverTool, RangeTool, CDSView, BooleanFilter, DataRange1d, LinearAxis, \\
    Range1d, CustomJS
from datetime import datetime
from common.time_utils import timestamp_to_datetime


class Signal:

    def __init__(self, data, marker='inverted_triangle', color='#10B479'):
        """
        信号
        :param data: 信号集数据
        :param marker: 标记类型
        :param color: 颜色
        """
        self.data = data
        self.marker = marker
        self.color = color

    @staticmethod
    def signal_below_price(data, price, func):
        """
        计算信号标记位置使其位于价格底部

        :param data:    信号数据
        :param price:   价格数据
        :param func:    目标信号判断条件方法
        :return: 位于价格底部的信号集合数据
        """
        signal = []
        for date, value in data.iteritems():
            if func(value):
                # signal.append(price[date] * 0.99)
                signal.append(price[date])
            else:
                signal.append(np.nan)
        return signal

    @staticmethod
    def signal_above_price(data, price, func):
        """
        计算信号标记位置使其位于价格顶部

        :param data:    信号数据
        :param price:   价格数据
        :param func:    目标信号判断条件方法
        :return: 位于价格顶部的信号集合数据
        """
        signal = []
        for date, value in data.iteritems():
            if func(value):
                # signal.append(price[date] * 1.01)
                signal.append(price[date])
            else:
                signal.append(np.nan)
        return signal


def make_range_tool(date, close, x_range=None, source=None):
    """
    时间范围选择工具
    :param date:
    :param close:
    :param x_range:
    :param source:
    """
    select = figure(title="", plot_height=100, plot_width=1500, x_axis_type="datetime", y_axis_type=None, tools="", toolbar_location=None, background_fill_color="#efefef")

    range_tool = RangeTool(x_range=x_range)
    range_tool.overlay.fill_color = "navy"
    range_tool.overlay.fill_alpha = 0.2

    select.line(date, close)
    select.ygrid.grid_line_color = None
    select.add_tools(range_tool)
    select.toolbar.active_multi = range_tool

    return select


def make_force_index_plot(source, color='#7922AD', x_range=None):
    """
    强力指数
    :param source: 列格式：date, efi, colors_efi
    :param color:
    :param x_range:
    :param source:
    """
    TOOLS = "crosshair,pan,wheel_zoom,box_zoom,reset,save"

    p = figure(x_axis_type="datetime", title="Force Index", plot_width=1500, plot_height=240, tools=TOOLS, toolbar_location='right', x_range=x_range)

    line = p.line(x='date', y='efi', line_width=1, color=color, source=source)

    # 悬浮提示
    hover_tool = HoverTool(

        tooltips="""
            <div">
                <div><b>EFI：</b><span style="font-size: 10px; color: @colors_efi;">@efi{0,0}</span></div>
                <div><b>Date：</b>@date{%F %T}</div>
                <div><b>Y：</b>$y{0.000}</div>
            </div>
        """,

        formatters={
            '@date': 'datetime',  # use 'datetime' formatter for 'date' field
        },

        # display a tooltip whenever the cursor is vertically in line with a glyph
        # "mouse":only when the mouse is directly over a glyph
        # "vline":	whenever the a vertical line from the mouse position intersects a glyph
        # "hline":	whenever the a horizontal line from the mouse position intersects a glyph
        mode='vline',

        # 是否显示箭头
        show_arrow=True,

        # line_policy='nearest',

        renderers=[line]
    )
    p.add_tools(hover_tool)

    return p


def make_candlestick_plot(df, period=None, signals=None, title='', filename=None, ema=(5, 10, 20), ema_color=('#C09A1C', '#7922AD', '#167BE1'), source=None):
    """
    蜡烛图
    :param df: 数据集，格式：date, open, high, low, close, volume
    :param period: 时间周期，默认，单位毫秒
    :param signals: 信号集
    :param title: 标题
    :param filename: 文件名
    :param ema: 移动平均线
    :param ema_color: 移动平均线颜色
    :param source: 数据源
    """

    inc = df.close > df.open
    dec = df.open > df.close

    w = period * 0.5 if period else 24 * 60 * 60 * 1000  # half day in ms

    # TOOLS = "pan,wheel_zoom,box_zoom,reset,save"
    TOOLS = "crosshair,pan,wheel_zoom,box_zoom,reset,save"

    # 蜡烛图
    p_candlestick = figure(x_axis_type="datetime", plot_width=1500, plot_height=450, tools=TOOLS, title=title, x_range=(df.date.iloc[0], df.date.iloc[-1]))
    # p_candlestick.sizing_mode = 'stretch_both'  # 全屏
    p_candlestick.xaxis.major_label_orientation = pi / 4  # x轴标题倾斜
    p_candlestick.grid.grid_line_alpha = 0.3

    # 绘制影线
    p_candlestick.segment(x0='date', y0='high', x1='date', y1='low', color="black", source=source)

    # 绘制柱体
    p_candlestick.vbar(df.date[inc], w, df.open[inc], df.close[inc], fill_color="#10B479", line_color="black")
    p_candlestick.vbar(df.date[dec], w, df.open[dec], df.close[dec], fill_color="#DD253E", line_color="black")

    # 绘制EMA
    ema_lines = []

    if ema:
        for index in range(0, len(ema)):

            key = 'ema%s' % ema[index]

            if key not in df:
                df[key] = EMA(df.close, ema[index])

            line = p_candlestick.line(x='date', y=key, line_color=ema_color[index], legend_label='EMA%s' % (ema[index]), source=source)

            ema_lines.append(line)

    # 绘制信号
    if signals:
        for signal in signals:
            p_candlestick.scatter(df.date, signal.data, marker=signal.marker, size=20, color=signal.color, alpha=0.6)

    if filename:
        output_file(filename, title=title, mode='inline')

    # show(p_candlestick)

    # 图例
    p_candlestick.legend.location = "top_left"
    p_candlestick.legend.border_line_alpha = 0
    p_candlestick.legend.background_fill_alpha = 0
    p_candlestick.legend.click_policy = "hide"

    return p_candlestick, ema_lines


def plot_middle_period(path, symbol):
    """
    绘制中周期图标
    :param path:    数据表路径，列格式：date, open, high, low, close, volume
    :param symbol：  交易对名称
    """
    df = pd.read_csv(path, parse_dates=True, index_col=0)

    df["date"] = df.index

    up_color = '#10B479'
    down_color = '#DD253E'

    # 计算涨幅
    pre_close = df.close.shift(1)
    df['increase'] = (df.close - pre_close) / pre_close * 100

    # EMA
    df['ema5'] = EMA(df.close, 5)
    df['ema10'] = EMA(df.close, 10)
    df['ema20'] = EMA(df.close, 20)

    # 开高低收价格颜色
    df['colors_open'] = np.where((df.open - pre_close) > 0, up_color, down_color)
    df['colors_high'] = np.where((df.high - pre_close) > 0, up_color, down_color)
    df['colors_low'] = np.where((df.low - pre_close) > 0, up_color, down_color)
    df['colors_close'] = np.where((df.close - pre_close) > 0, up_color, down_color)

    # 均线颜色
    df['colors_ema5'] = np.where((df['ema5'].diff()) > 0, up_color, down_color)
    df['colors_ema10'] = np.where((df['ema10'].diff()) > 0, up_color, down_color)
    df['colors_ema20'] = np.where((df['ema20'].diff()) > 0, up_color, down_color)

    # EFI颜色
    df['colors_efi'] = np.where(df['efi'] > 0, up_color, down_color)

    # 做多信号
    long_signal = Signal(Signal.signal_below_price(df['signals'], df['low'], lambda signals: 'buy' in signals), 'triangle', '#10B479')

    # 做空信号
    short_signal = Signal(Signal.signal_above_price(df['signals'], df['high'], lambda signals: 'short' in signals), 'inverted_triangle', '#DD253E')

    filename = 'triple_filter_trade_system_middle.html'

    title = '%s 三重滤网交易系统' % symbol

    # 数据源
    source = ColumnDataSource(df)

    # ETH/USDT 5分钟K线图
    p_candlestick, ema_lines = make_candlestick_plot(df, period=5 * 60 * 1000, signals=[long_signal, short_signal], title=title, source=source)

    # 强力指数图
    p_efi = make_force_index_plot(source, x_range=p_candlestick.x_range)

    # 悬浮提示
    hover_tool = HoverTool(

        tooltips="""
        <div">
            <div><b>Open：</b><span style="font-size: 10px; color: @colors_open;">@open{0.000}</span></div>
            <div><b>High：</b><span style="font-size: 10px; color: @colors_high;">@high{0.000}</span></div>
            <div><b>Low：</b><span style="font-size: 10px; color: @colors_low;">@low{0.000}</span></div>
            <div><b>Close：</b><span style="font-size: 10px; color: @colors_close;">@close{0.000}</span></div>
            <div><b>Increase：</b><span style="font-size: 10px; color: @colors_close;">@increase{0.00}%</span></div>
            <div><b>Volume：</b><span style="font-size: 10px; color: @colors_close;">@volume{0,0}</span></div>
            <div><b>EMA5：</b><span style="font-size: 10px; color: @colors_ema5;">@ema5{0.000}</span></div>
            <div><b>EMA10：</b><span style="font-size: 10px; color: @colors_ema10;">@ema10{0.000}</span></div>
            <div><b>EMA20：</b><span style="font-size: 10px; color: @colors_ema20;">@ema20{0.000}</span></div>
            <div><b>EFI：</b><span style="font-size: 10px; color: @colors_efi;">@efi{0,0}</span></div>
            <div><b>Date：</b>@date{%F %T}</div>
            <div><b>Y：</b>$y{0.000}</div>
        </div>
        """,

        formatters={
            '@date': 'datetime',
        },

        mode='vline',

        # 是否显示箭头
        show_arrow=True,

        renderers=[ema_lines[0]],

        # point_policy='snap_to_data',
    )
    p_candlestick.add_tools(hover_tool)

    range_tool = make_range_tool(df.date, df.close, x_range=p_candlestick.x_range, source=source)

    layout = column(range_tool, p_candlestick, p_efi)

    output_file(filename, title=title, mode='inline')

    show(layout)

项目地址：https://github.com/linchaolong/SimpleQuant