本文重點摘要
- 文章難度：★☆☆☆☆
- 在 Google Colab 的TQuant Lab 使用教學
- 簡易測試，選擇營收最高的五間公司（任意產業），回測報酬率
- 安裝字型，讓 Pyfolio 套件能順利繪圖
前言
對於想使用 TQuant Lab 的同學們來說，除了原本的 GitHub 安裝教學，現在提供了一個更快速、簡潔的方式，讓我們可以在 Google Colab 上直接使用，無需設定虛擬環境，大大降低了使用門檻。
在 Google Colab 安裝 TQuant Lab 套件
只需簡單兩行代碼，完成環境設定
（因應近期 Google Colab 改版，需要新增 dask 降版指令）
!pip install zipline-tej !pip install pandas==1.5.3 !pip install dask==2.30.0 # dask降版
遇到彈跳視窗顯示錯誤，直接按「重新啟動工作階段」即可
最終運行完成會出現以下錯誤訊息，但無需理會，實際上已經可以正常使用
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. cudf-cu12 24.4.1 requires pandas<2.2.2dev0,>=2.0, but you have pandas 1.5.3 which is incompatible. google-colab 1.0.0 requires pandas==2.0.3, but you have pandas 1.5.3 which is incompatible.
TQuant Lab 使用教學：選擇營收最高的五家半導體公司進行回測
載入常用套件
import os os.environ['TEJAPI_BASE'] = 'https://api.tej.com.tw' os.environ['TEJAPI_KEY'] = 'Your_Key' import datetime import tejapi import pandas as pd import numpy as np
先找出所有是半導體及電腦產業的公司
from zipline.sources.TEJ_Api_Data import get_universe pool = get_universe(start = '2023-07-02', end = '2024-07-02', mkt_bd_e = 'TSE', # 已上市之股票 stktp_e = 'Common Stock', # 普通股 sub_ind_e=['M2324 Semiconductor', 'M2325 Computer and Peripheral Equipment'])
看他們的營收，並選擇前五高的，並儲存其股票代碼
import TejToolAPI start_time = pd.Timestamp('2023-07-02') end_time = pd.Timestamp('2024-07-02') data = TejToolAPI.get_history_data(start = start_time, end = end_time, ticker = pool, fin_type = 'Q', # 為累計資料，舉例來說，Q3累計：1月～9月的資料。 columns = ['主產業別_中文', '常續ROE', '營業毛利率', '營運產生現金流量', '投資產生現金流量', '負債比率', 'per_tej', '營業總收入'], transfer_to_chinese = True) data = data.drop_duplicates(subset=['股票代碼'], keep='last').reset_index(drop=True) data = data.nlargest(5, '營業總收入_Q') #找出最高的5間 data
股票代碼
日期 主產業別_中文 本益比_TEJ 營運產生現金流量_Q 負債比率_Q 營業毛利率_Q 投資產生現金流量_Q 常續ROE_Q 營業總收入_Q 6 2330 2024-07-02 M2300 電子工業 29.1046 436311108.0
36.67 53.07 -159806991.0 6.40 592644201.0 27 2382 2024-07-02 M2300 電子工業
26.0091 23733615.0 76.20
8.48 22640275.0 6.50
258939378.0 86 4938 2024-07-02 M2300 電子工業 18.0987 20958867.0 57.42 4.23 -2946059.0 1.54 250399656.0 64 3231 2024-07-02 M2300 電子工業
21.5123 3127081.0 69.48 7.20 -2693302.0 4.39 239325146.0 4 2324 2024-07-02 M2300 電子工業 17.8337 8250149.0 70.01 4.91 -2205331.0 1.69 199571114.0 tickers = data['股票代碼'].unique() start = '2023-01-01' end = '2024-07-02' os.environ['mdate'] = start + ' ' + end os.environ['ticker'] = ' '.join(tickers) + ' ' + 'IR0001' !zipline ingest -b tquant
將資料載入本地電腦
start = '2023-01-01' end = '2024-07-02' os.environ['mdate'] = start + ' ' + end os.environ['ticker'] = ' '.join(tickers) + ' ' + 'IR0001' !zipline ingest -b tquant from zipline.data import bundles bundle_data = bundles.load('tquant') from zipline.api import * from zipline.finance import commission, slippage def initialize(context): context.day = 0 context.tickers = tickers set_slippage(slippage.VolumeShareSlippage(volume_limit = 0.025, price_impact = 0.1)) set_commission(commission.Custom_TW_Commission(min_trade_cost = 20, discount = 1.0, tax = 0.003)) set_benchmark(symbol('IR0001')) set_liquidity_risk_management_rule(['全額交割股票(Full-Cash Delivery Securities)', '漲停股票(Limit Up)', '跌停股票(Limit Down)', '開盤即鎖死(Limited Whole Day)']) def handle_data(context, data): #回測第一天買進 if context.day == 0: for ticker in context.tickers: order_percent(symbol(ticker), 1 / len(tickers)) context.day += 1 import matplotlib.pyplot as plt capital_base = 1e6 # 設定初始資金 def analyze(context, results): fig = plt.figure() ax1 = fig.add_subplot(111) results['benchmark_cum'] = results.benchmark_return.add(1).cumprod() * capital_base results[['portfolio_value', 'benchmark_cum']].plot(ax = ax1, label = 'Portfolio Value($)') ax1.set_ylabel('Portfolio value (TWD)') plt.legend(loc = 'upper left') plt.gcf().set_size_inches(18, 8) plt.grid() plt.show() from zipline import run_algorithm start_date = pd.Timestamp('20230101', tz = 'utc') end_date = pd.Timestamp('20240702', tz = 'utc') # 轉換成時間序列格式 results = run_algorithm( start = start_date, end = end_date, initialize = initialize, handle_data = handle_data, analyze = analyze, bundle = 'tquant', capital_base = capital_base, ) results
使用 Pyfolio 會遇到的字型問題以及解決方式
from pyfolio.utils import extract_rets_pos_txn_from_zipline import pyfolio as pf # 從 results 資料表中取出 returns, positions & transactions returns, positions, transactions = extract_rets_pos_txn_from_zipline(results) # 從 results 資料表中取出 returns, positions & transactions benchmark_rets = results.benchmark_return # 取出 benchmark 的報酬率 # 繪製夏普比率圖 from pyfolio.plotting import plot_rolling_sharpe plot_rolling_sharpe(returns, factor_returns=benchmark_rets) WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Microsoft JhengHei WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Microsoft JhengHei WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Microsoft JhengHei WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Microsoft JhengHei WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Microsoft JhengHei
即使圖表已成功繪製，但仍可能出現字體缺失的警告訊息。為了解決這個問題，我們可以自行載入字體
!wget -O MicrosoftJhengHei.ttf https://drive.google.com/uc?id=1nMlvxPOPUGkHxYD5kuP8Ur37EmKlZAW_&export=download !wget -O ArialUnicodeMS.ttf https://drive.google.com/uc?id=1Y4O8Flv7lfrzHqOE8dkFTSctyYOpAJ0N&export=download matplotlib.font_manager.fontManager.addfont('MicrosoftJhengHei.ttf') matplotlib.rc('font', family='sans-serif') matplotlib.font_manager.fontManager.addfont('ArialUnicodeMS.ttf') matplotlib.rc('font', family='sans-serif')
再試一次後，警告訊息將不再出現
plot_rolling_sharpe(returns, factor_returns=benchmark_rets)
結論
本次介紹，在測試的部分幾乎都參考了這篇：『巴菲特選股策略』的代碼內容，如果有任何問題，都可以查看原始延伸閱讀的內容。
另外，這次的介紹讓大家不需額外安裝虛擬環境，一大堆的設定才能使用 TQuant，而是更方便的就可以開始操作，小編認為這是一個非常棒的方式，推薦大家！
【TQuant Lab回測系統】解決你的量化金融痛點
全方位提供交易回測所需工具