求助 LSTM训练集的准确度过低是怎么回事

我编写了一个LSTM程序用来预测股票未来30个tick数据。但是训练集跑出的结果里，准确度只有0.02左右。

import numpy as np
import pandas as pd

X_list = []
for i in range(10):
    temp = pd.read_excel(r'C:\Users\86186\Desktop\附件1：10支票_tick_data.xlsx',sheet_name = i)
    X_list.append(temp)

length = 0
local_index = 0
for i in range(len(X_list)):
    local_length = len(X_list[i])
    if local_length>length:
        length = local_length
        local_index = i

lengthest_tick = X_list[local_index]
X_list.remove(lengthest_tick)
title_list = ['平安银行','万科A','中兴通讯','中国长城','华侨城A','TCL科技','中联重科','申万宏源','美的集团','潍柴动力']
local_title = title_list[local_index]
title_list.remove(local_title)
title_list.append(local_title)#将最长的表名移至最后

tick_name = []
for i in lengthes