py-as-zhong/predictor.py

import time
import utils
import numpy as np

class Predictor:
    '''
        reference
    '''
    def __init__(self, ref_name = ["Unknown Market"], alpha = [1.0 for _ in range(99)], gamma = 0.999):
        self.loop = 0
        self.arr = []
        self.trade_mp_series = []
        self.ref_mp_series = []
        self.ref_num = len(ref_name)
        ### 定价
        self.window = 10
        ### 价格系数
        self.alpha = alpha
        # 参考
        print('定价系数:', gamma)
        self.gamma = gamma
        self.avg_spread = [None for _ in range(self.ref_num)]
    
    def processer(self):
        '''
            计算任务
        '''
        # update trade mp
        bp=self.arr[-1][utils.BP_INDEX]
        ap=self.arr[-1][utils.AP_INDEX]
        mp=(bp+ap)*0.5
        self.trade_mp_series.append(mp)
        # 更新参考盘口mp
        ref_mp = []
        for i in range(self.ref_num):
            bp = self.arr[-1][utils.LEN*(1+i)+utils.BP_INDEX]
            ap = self.arr[-1][utils.LEN*(1+i)+utils.AP_INDEX]
            mp=(bp+ap)*0.5
            ref_mp.append(mp)
        self.ref_mp_series.append(ref_mp)
        # 偏差计算
        self._update_avg_spread()

    def _update_avg_spread(self):
        '''
            更新平均偏差
        '''
        # 计算偏差1
        for i in range(self.ref_num):
            bias = self.ref_mp_series[-1][i]*self.alpha[i] - self.trade_mp_series[-1]
            # 如果是刚启动 gamma不能太大
            if self.loop < 100:
                gamma = 0.9
            else:
                gamma = self.gamma
            if self.avg_spread[i] == None:
                self.avg_spread[i] = bias
            else:
                self.avg_spread[i] = self.avg_spread[i]*gamma + bias*(1-gamma)

    def check_length(self):
        # 行情缓存
        if len(self.arr) > self.window:del(self.arr[0])
        if len(self.trade_mp_series) > self.window:del(self.trade_mp_series[0])
        if len(self.ref_mp_series) > self.window:del(self.ref_mp_series[0])

    # @utils.timeit
    def onTime(self, data):
        if isinstance(data, list):
            if len(data) > 0:
                self.loop += 1
                self.arr.append(data)
                self.processer()
                self.check_length()
            else:
                print("行情数据为空")
        else:
            print("行情数据为None")

    # @utils.timeit
    def Get_ref(self, ref_ticker):
        '''
            get ref price
        '''
        ref_mid = []
        for i in range(self.ref_num):
            ref_mid.append(
                [ref_ticker[i][0]*self.alpha[i] - self.avg_spread[i], ref_ticker[i][1]*self.alpha[i] - self.avg_spread[i]]
            )
        return ref_mid

if __name__ == "__main__":

    import pandas as pd
    import numpy as np
    import matplotlib
    matplotlib.use('TkAgg')
    import matplotlib.pyplot as plt

    arr = pd.read_csv('history/ftm_usdt_binance_usdt_swap.csv').values.tolist()

    def line_data_to_tickers(data, ref_num):
        ref_tickers = []
        for i in range(ref_num):
            ref_tickers.append([data[utils.LEN*(i+1)+utils.BP_INDEX], data[utils.LEN*(i+1)+utils.AP_INDEX]])
        return ref_tickers

    ref_num = len(arr[0])//utils.LEN - 1

    p = Predictor(ref_name=["unkwon" for _ in range(ref_num)])
    t = []
    ref_index = 1

    for data in arr:
        p.onTime(data)
        trade_mp = (data[utils.BP_INDEX] + data[utils.AP_INDEX])*0.5
        ref_price = p.Get_ref(line_data_to_tickers(data, ref_num))
        t.append([
            trade_mp,
            (ref_price[ref_index][0]+ref_price[ref_index][1])*0.5,
        ])

    t = pd.DataFrame(t,columns=['mp','ref'])

    if 1:
        plt.figure()
        plt.plot(t['mp'],'k')
        plt.plot(t['ref'],'g')
        plt.grid()
        plt.show()