预测股票价格 模型_建立有马模型来预测股票价格

预测股票价格 模型

前言 (Preface)

If you are reading this, it’s most likely because you love to solve puzzles. I’m a very competitive person by nature. The Mt. Everest of puzzles, in my opinion, is trying to find excess returns through active trading in the stock market. This blog is my first of many posts of an attempt to — hopefully — summit the intimidating Mt. Everest of algorithmic trading and emerge profitably.

如果您正在阅读本文，则最有可能是因为您喜欢解决难题。 我天生就是一个很有竞争力的人。 山。 我认为，珠穆朗玛的难题正试图通过股票市场上的活跃交易来寻找超额收益。 这个博客是我尝试(希望)登顶这座令人生畏的山峰的众多文章中的第一篇。 算法交易的珠穆朗玛峰并盈利。

First off, I must atone for my sins. I am a failed daytrader. I attempted day trading over a summer during a hiatus from work. Everything I read told me to have a plan before I entered a trade, and when I did enter a trade to stick to my trading plan despite whatever emotion I may be feeling. I was confident that this would be no problem. I have a pretty good grasp on all my vices — don’t we all? If you’re human, then the answer is no. No, you don’t have your vices under control. That is the literal definition of a vice. I was naive and thankfully was able to quit before I did any real damage to my account. However, like a moth to a flame, I can’t leave a good puzzle. I am going back into the fray, but this time I have a plan. I plan on building a purely quantitative system that removes the worst part of any trading plan (yourself) from the equation.

首先，我必须赎罪。 我的交易员失败了。 我在工作中断期间试图在整个夏天进行日间交易。 我阅读的所有内容都告诉我在进入交易之前要有一个计划，而当我进入交易时要坚持我的交易计划，尽管我可能会感到任何情绪。 我相信这不会有问题。 我对所有的恶习都掌握得很好–不是所有人吗？ 如果您是人类，那么答案是否定的。 不，您不受控制。 那是恶习的字面定义。 我很天真，幸好我能够在对帐户造成任何实际损失之前退出。 但是，就像飞蛾扑火一样，我不能留下一个好谜。 我将重返战场，但是这次我有一个计划。 我计划建立一个纯粹的量化系统，以消除等式中任何交易计划(您自己)的最糟糕部分。

Before you go any farther, let me say this blog post is simply about ARIMA models, and this preface is just a teaser for the full algorithmic trading system that will come in the future. If the intro did not scare you off, let’s begin with a simple ARIMA model that helps us predict tomorrow’s daily closing price of whatever stock you choose.

在继续之前，请允许我说这篇博客文章只是关于ARIMA模型的，并且此序言只是对将来将要使用的完整算法交易系统的预告。 如果介绍没有吓倒您，让我们从一个简单的ARIMA模型开始，该模型可以帮助我们预测所选股票的明天每日收盘价。

让我们编码 (Let’s Code)

ARIMA is an acronym that stands for AutoRegressive Integrated Moving Average. Throughout this blog, I will break down the steps necessary to implement a successful ARIMA model.

ARIMA是首字母缩写词，代表自动回归综合移动平均线。 在整个博客中，我将分解实现成功ARIMA模型的必要步骤。

步骤1：取得资料 (Step 1: Get the data)

For this example, we will use the SPY ETF. The SPY is an ETF that mimics the S&P500, which is a basket of stocks weighted by market cap. Many ETFs mimic the S&P 500, but this is the most common and most liquid fund.

在此示例中，我们将使用SPY ETF。 SPY是模仿S＆P500的ETF，S＆P500是按市值加权的一篮子股票。 许多ETF模仿标准普尔500指数，但这是最常见和最具流动性的基金。

To use this, we will use the yfinance API. If you do not have this API installed, a simple pip install in your Jupyter Notebook should do. If you wish to learn more about this API, check out the PyPI page for more details.

要使用此功能，我们将使用yfinance API。 如果您尚未安装此API，则应该在Jupyter Notebook中进行简单的pip安装。 如果您想了解更多有关此API的信息， 请查看PyPI页面以获取更多详细信息 。

!pip install yfinance

Now, let’s pull the entire daily history of the ticker “SPY” into a neat data frame.

现在，让我们将股票代码“ SPY”的整个每日历史记录拉入一个整洁的数据框中。

import yfinance as yf
import pandas as pdspy = yf.Ticker("SPY")# get stock info
spy.info# get historical market data as df
hist = spy.history(period="max")# Save df as CSV
hist.to_csv('SPY.csv')# Read back in as dataframe
spy = pd.read_csv('SPY.csv')# Convert Date column to datetime
spy['Date'] = pd.to_datetime(spy['Date'])

If you print out “spy.info” you will get a dictionary of a lot of extra data on the stock. All we are currently worried about for this model is the closing price. The object “hist” is a dataframe, and we send it to CSV to have a local copy. You can run this every day after the market closes and get the most up to date information if you want.

如果您打印出“ spy.info”，您将获得有关大量库存额外数据的词典。 我们目前对此模型担心的只是收盘价。 对象“ hist”是一个数据框，我们将其发送到CSV以获取本地副本。 您可以在市场收盘后每天执行此操作，并根据需要获取最新信息。