{"id":52627,"date":"2020-07-15T17:19:00","date_gmt":"2020-07-15T21:19:00","guid":{"rendered":"https:\/\/ibkrcampus.com\/?p=52627"},"modified":"2022-11-21T09:45:53","modified_gmt":"2022-11-21T14:45:53","slug":"how-to-be-a-quant-trader-quantconnect-robot-wealth","status":"publish","type":"post","link":"https:\/\/www.interactivebrokers.com\/campus\/ibkr-quant-news\/how-to-be-a-quant-trader-quantconnect-robot-wealth\/","title":{"rendered":"How To Be a Quant Trader \u2013 Experiments with QuantConnect"},"content":{"rendered":"\n

Code along Robot Wealth as they present an analysis of the SPY returns process using the QuantConnect research platform.<\/p>\n\n\n\n

Excerpt<\/strong><\/em><\/p>\n\n\n\n

Example Research With QuantConnect Code<\/p>\n\n\n\n

Using the QuantConnect ecosystem in a typical quant workflow.<\/p>\n\n\n\n

Note: This code is meant to be used within QuantConnect research<\/a> environment<\/em><\/p>\n\n\n\n

\n# Import dependecies
\nimport numpy as np
\nimport seaborn as sns
\nimport matplotlib.pyplot as plt
\nplt.style.use(‘ggplot’) #There is a positive correlation between chart pretiness and risk-adjusted returns
\nplt.rcParams[‘figure.figsize’] = [10, 7]

\n# QuantBook Analysis Tool \n# Load SPY historical data

\nqb = QuantBook()
\nspy = qb.AddEquity(“SPY”)
\nhistory = qb.History(qb.Securities.Keys, 5000, Resolution.Daily) #5000 days of SPY daily data

\n# Drop pandas level
\nhistory = history.reset_index().drop(‘symbol’,axis=1)

\n# Calculate SPY returns and fillna
\nhistory[‘returns’] = (history[‘close’].pct_change() * 100).fillna(0)\n<\/p><\/p>\n\n\n\n

1. Analysing the return distribution<\/strong><\/p>\n\n\n\n

Now that we have SPY daily returns let\u2019s quickly see what we\u2019re dealing with.<\/p>\n\n\n\n

\nhistory[‘returns’].describe()\n\n<\/p>\n\n\n\n

\ncount 5000.000000
\nmean 0.030071
\nstd 1.235997
\nmin -11.638806
\n25% -0.443536
\n50% 0.061797
\n75% 0.573180
\nmax 11.360371
\nName: returns, dtype: float64\n<\/p>\n\n\n\n

Let\u2019s look at the extreme values of returns ie max and min<\/p>\n\n\n\n

\nhistory[history[‘returns’] == min(history[‘returns’])]\n<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

\nhistory[history[‘returns’] == max(history[‘returns’])]\n<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

The recent corona drawdown is the biggest single-day market drop in history, and we have the biggest up move in 2008.<\/p>\n\n\n\n

Let\u2019s look at the distribution of daily returns for the SPY<\/p>\n\n\n\n

\nsns.distplot(history[‘returns’],label=’Distribution of SPY returns’)\nplt.legend()\n\n\n<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

2. Comparing to a normal distribution<\/strong><\/p>\n\n\n\n

Let\u2019s first create some random data  and plot their distribution<\/p>\n\n\n\n

\nrandom = np.random.normal(scale=1.23,size=500000)
\nsns.distplot(random,label=’Returns sampled from normal distribution’,color=’blue’)
\nplt.legend()
\nrandom_series = pd.Series(random)\n\n\n<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

There it is, a beautiful well behaved normal distribution, Let\u2019s see how this compares to our SPY returns distribution.<\/p>\n\n\n\n

\nsns.distplot(history[‘returns’],label=’Distribution of SPY returns’)\nsns.distplot(random,label=’Returns sampled from normal distribution’)\nplt.legend()\n\n\n\n<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

Now the high kurtosis of the SPY returns becomes even more apparent.<\/p>\n\n\n\n

So far we\u2019ve learned that:<\/p>\n\n\n\n