Visualizing Correlations Among Dow 30 Stocks Via NetworkX

NetworkX is a Python package for the creation, manipulation, and study of the structure, dynamics, and functions of complex networks.

Using daily adjusted close data from 20201118 to 20201218 for Dow 30 stocks, we compute correlation coefficients, apply a threshold of 0.8 to find similar stocks, and produce two types of graphs with NetworkX.

To compute correlation coefficients, we read in daily adjusted close data and use Pandas to construct a correlation coefficient matrix.

def correlation_coefficient_matrix():   
    # get symbols
    df = pd.read_pickle('df_dow30.pkl')

    # sort symbols alphabetically
    df.sort_values(by='Symbol', ascending=True, inplace=True)
    array_symbols = df['Symbol'].values
    
    # get data, put into dictionary then dataframe
    dict_sym_ac = {}  # key=symbol, value=array of adj close
    for sym in array_symbols:
        dftemp = READ DATA
        dict_sym_ac[sym] = dftemp['Adj_Close'].values
    
    # create correlation coeff df
    dfdata = pd.DataFrame.from_dict(dict_sym_ac)
    dfcc = dfdata.corr().round(3)
    dfcc.to_csv('correlation_coefficient_matrix.csv')
    dfcc.to_pickle('correlation_coefficient_matrix.pkl')
    
    return dfcc

Dow 30 Correlation Coefficient Matrix

	AAPL	AMGN	AXP	BA	CAT	CRM	CSCO	CVX	DIS	DOW	GS	HD	HON	IBM	INTC	JNJ	JPM	KO	MCD	MMM	MRK	MSFT	NKE	PG	TRV	UNH	V	VZ	WBA	WMT
AAPL	1.0	0.817	0.322	0.561	0.643	-0.803	0.837	-0.12	0.763	-0.455	0.716	-0.236	0.628	0.563	0.736	0.832	0.237	0.606	-0.409	0.006	0.367	0.662	0.798	-0.412	0.525	0.507	0.228	0.123	0.637	-0.734
AMGN	0.817	1.0	0.239	0.552	0.669	-0.76	0.704	-0.112	0.641	-0.364	0.604	-0.474	0.633	0.419	0.653	0.855	0.085	0.693	-0.59	-0.105	0.469	0.351	0.668	-0.529	0.489	0.607	0.1	0.178	0.708	-0.807
AXP	0.322	0.239	1.0	0.834	0.482	-0.621	0.639	0.696	0.171	-0.044	0.656	-0.318	0.663	0.804	0.754	0.331	0.901	0.318	-0.412	0.056	0.614	0.28	0.618	-0.474	0.065	0.657	0.649	0.778	0.733	-0.096
BA	0.561	0.552	0.834	1.0	0.718	-0.843	0.81	0.601	0.5	-0.189	0.806	-0.618	0.857	0.78	0.858	0.624	0.722	0.637	-0.651	0.064	0.716	0.245	0.814	-0.791	0.31	0.624	0.349	0.608	0.936	-0.542
CAT	0.643	0.669	0.482	0.718	1.0	-0.697	0.814	0.447	0.745	0.109	0.848	-0.501	0.881	0.77	0.664	0.752	0.49	0.914	-0.403	0.372	0.428	0.366	0.827	-0.66	0.668	0.358	0.312	0.202	0.773	-0.762
CRM	-0.803	-0.76	-0.621	-0.843	-0.697	1.0	-0.921	-0.251	-0.715	0.456	-0.84	0.428	-0.825	-0.757	-0.893	-0.798	-0.492	-0.608	0.619	-0.057	-0.622	-0.419	-0.843	0.746	-0.397	-0.573	-0.212	-0.395	-0.862	0.71
CSCO	0.837	0.704	0.639	0.81	0.814	-0.921	1.0	0.307	0.787	-0.251	0.926	-0.295	0.861	0.879	0.856	0.824	0.586	0.683	-0.433	0.235	0.509	0.613	0.905	-0.637	0.505	0.5	0.417	0.338	0.812	-0.686
CVX	-0.12	-0.112	0.696	0.601	0.447	-0.251	0.307	1.0	0.129	0.42	0.525	-0.286	0.592	0.605	0.383	0.052	0.792	0.323	-0.193	0.432	0.37	-0.054	0.371	-0.528	0.215	0.097	0.286	0.375	0.433	-0.031
DIS	0.763	0.641	0.171	0.5	0.745	-0.715	0.787	0.129	1.0	-0.188	0.767	-0.162	0.748	0.593	0.583	0.72	0.249	0.701	-0.209	0.39	0.155	0.596	0.732	-0.553	0.667	0.081	0.022	-0.193	0.539	-0.824
DOW	-0.455	-0.364	-0.044	-0.189	0.109	0.456	-0.251	0.42	-0.188	1.0	-0.031	0.105	-0.056	0.02	-0.447	-0.232	0.218	0.165	0.396	0.551	-0.31	-0.123	-0.221	0.229	0.365	-0.301	0.143	-0.033	-0.217	0.208
GS	0.716	0.604	0.656	0.806	0.848	-0.84	0.926	0.525	0.767	-0.031	1.0	-0.269	0.914	0.914	0.793	0.738	0.7	0.704	-0.372	0.463	0.462	0.535	0.904	-0.713	0.649	0.362	0.374	0.248	0.756	-0.674
HD	-0.236	-0.474	-0.318	-0.618	-0.501	0.428	-0.295	-0.286	-0.162	0.105	-0.269	1.0	-0.461	-0.144	-0.372	-0.48	-0.124	-0.584	0.844	0.321	-0.714	0.292	-0.374	0.571	-0.155	-0.504	0.084	-0.407	-0.675	0.506
HON	0.628	0.633	0.663	0.857	0.881	-0.825	0.861	0.592	0.748	-0.056	0.914	-0.461	1.0	0.853	0.801	0.716	0.652	0.789	-0.495	0.374	0.571	0.381	0.876	-0.79	0.537	0.368	0.303	0.292	0.832	-0.664
IBM	0.563	0.419	0.804	0.78	0.77	-0.757	0.879	0.605	0.593	0.02	0.914	-0.144	0.853	1.0	0.795	0.552	0.838	0.562	-0.225	0.459	0.426	0.554	0.828	-0.578	0.407	0.393	0.599	0.431	0.705	-0.405
INTC	0.736	0.653	0.754	0.858	0.664	-0.893	0.856	0.383	0.583	-0.447	0.793	-0.372	0.801	0.795	1.0	0.637	0.618	0.537	-0.527	0.007	0.61	0.461	0.835	-0.708	0.253	0.626	0.355	0.455	0.845	-0.524
JNJ	0.832	0.855	0.331	0.624	0.752	-0.798	0.824	0.052	0.72	-0.232	0.738	-0.48	0.716	0.552	0.637	1.0	0.193	0.754	-0.654	0.071	0.603	0.381	0.779	-0.548	0.585	0.496	0.133	0.21	0.737	-0.774
JPM	0.237	0.085	0.901	0.722	0.49	-0.492	0.586	0.792	0.249	0.218	0.7	-0.124	0.652	0.838	0.618	0.193	1.0	0.327	-0.131	0.365	0.364	0.394	0.581	-0.399	0.247	0.419	0.638	0.604	0.561	-0.082
KO	0.606	0.693	0.318	0.637	0.914	-0.608	0.683	0.323	0.701	0.165	0.704	-0.584	0.789	0.562	0.537	0.754	0.327	1.0	-0.431	0.302	0.364	0.28	0.701	-0.595	0.741	0.328	0.135	0.169	0.757	-0.779
MCD	-0.409	-0.59	-0.412	-0.651	-0.403	0.619	-0.433	-0.193	-0.209	0.396	-0.372	0.844	-0.495	-0.225	-0.527	-0.654	-0.131	-0.431	1.0	0.367	-0.883	0.17	-0.463	0.603	-0.079	-0.615	0.085	-0.464	-0.697	0.472
MMM	0.006	-0.105	0.056	0.064	0.372	-0.057	0.235	0.432	0.39	0.551	0.463	0.321	0.374	0.459	0.007	0.071	0.365	0.302	0.367	1.0	-0.292	0.326	0.299	-0.171	0.559	-0.459	0.088	-0.269	-0.01	-0.157
MRK	0.367	0.469	0.614	0.716	0.428	-0.622	0.509	0.37	0.155	-0.31	0.462	-0.714	0.571	0.426	0.61	0.603	0.364	0.364	-0.883	-0.292	1.0	-0.155	0.546	-0.612	-0.019	0.606	0.131	0.603	0.671	-0.276
MSFT	0.662	0.351	0.28	0.245	0.366	-0.419	0.613	-0.054	0.596	-0.123	0.535	0.292	0.381	0.554	0.461	0.381	0.394	0.28	0.17	0.326	-0.155	1.0	0.515	0.057	0.382	0.218	0.536	-0.025	0.258	-0.307
NKE	0.798	0.668	0.618	0.814	0.827	-0.843	0.905	0.371	0.732	-0.221	0.904	-0.374	0.876	0.828	0.835	0.779	0.581	0.701	-0.463	0.299	0.546	0.515	1.0	-0.646	0.533	0.467	0.386	0.299	0.798	-0.687
PG	-0.412	-0.529	-0.474	-0.791	-0.66	0.746	-0.637	-0.528	-0.553	0.229	-0.713	0.571	-0.79	-0.578	-0.708	-0.548	-0.399	-0.595	0.603	-0.171	-0.612	0.057	-0.646	1.0	-0.344	-0.227	0.146	-0.187	-0.737	0.645
TRV	0.525	0.489	0.065	0.31	0.668	-0.397	0.505	0.215	0.667	0.365	0.649	-0.155	0.537	0.407	0.253	0.585	0.247	0.741	-0.079	0.559	-0.019	0.382	0.533	-0.344	1.0	0.037	-0.011	-0.167	0.392	-0.715
UNH	0.507	0.607	0.657	0.624	0.358	-0.573	0.5	0.097	0.081	-0.301	0.362	-0.504	0.368	0.393	0.626	0.496	0.419	0.328	-0.615	-0.459	0.606	0.218	0.467	-0.227	0.037	1.0	0.478	0.795	0.705	-0.283
V	0.228	0.1	0.649	0.349	0.312	-0.212	0.417	0.286	0.022	0.143	0.374	0.084	0.303	0.599	0.355	0.133	0.638	0.135	0.085	0.088	0.131	0.536	0.386	0.146	-0.011	0.478	1.0	0.499	0.309	0.127
VZ	0.123	0.178	0.778	0.608	0.202	-0.395	0.338	0.375	-0.193	-0.033	0.248	-0.407	0.292	0.431	0.455	0.21	0.604	0.169	-0.464	-0.269	0.603	-0.025	0.299	-0.187	-0.167	0.795	0.499	1.0	0.582	0.094
WBA	0.637	0.708	0.733	0.936	0.773	-0.862	0.812	0.433	0.539	-0.217	0.756	-0.675	0.832	0.705	0.845	0.737	0.561	0.757	-0.697	-0.01	0.671	0.258	0.798	-0.737	0.392	0.705	0.309	0.582	1.0	-0.657
WMT	-0.734	-0.807	-0.096	-0.542	-0.762	0.71	-0.686	-0.031	-0.824	0.208	-0.674	0.506	-0.664	-0.405	-0.524	-0.774	-0.082	-0.779	0.472	-0.157	-0.276	-0.307	-0.687	0.645	-0.715	-0.283	0.127	0.094	-0.657	1.0

We then create a new DataFrame for use in NetworkX. Here, we have three columns: ‘from’, ‘to’, ‘corr coeff’ which are graph nodes (stock symbol), graph edges (undirected), and graph edge weights (correlation coefficient), and only include edges with correlation coefficient > 0.8.

We construct two types of graphs, circular and spring, then plot the results. As we are only interested in displaying correlation coefficient information, the choice of graph types was dictated by a desire to show similar stocks and clusters of such stocks.

import pandas as pd
import networkx
import matplotlib.pyplot as plt

if __name__ == '__main__':
    
    # compute correlation coeff matrix
    try:
        df_cc = pd.read_pickle('correlation_coefficient_matrix.pkl')
    except:
        df_cc = correlation_coefficient_matrix()
        
    
    # apply corr coeff threshold and create new df
    list_symbols = df_cc.columns.to_list()
    threshold = 0.8  # corr coeff threshold
    list_from = []
    list_to = []
    list_corr_coeff = []
    for sym_from in list_symbols:
        for sym_to in list_symbols:
            if sym_from != sym_to:
                corr_coef = df_cc.loc[sym_from, sym_to]
                if corr_coef > threshold:
                    list_from.append(sym_from)
                    list_to.append(sym_to)
                    list_corr_coeff.append(corr_coef)
                    
    # create df for constructing graph
    df_graph = pd.DataFrame({'from':list_from, 'to':list_to, 
                             'corr coeff':list_corr_coeff})
    
    G=networkx.from_pandas_edgelist(df_graph, 'from', 'to')
    
 
    # plot the network
    networkx.draw_spring(G, with_labels=True, node_color='cornflowerblue', node_size=600, 
                         edge_color='black', linewidths=1, font_size=10,
                         width=1, alpha=0.5)

    plt.savefig('dow_30_corr_coeff_spring.png')
    
    
    # plot the network
    plt.clf()
    networkx.draw_circular(G, with_labels=True, node_color='cornflowerblue', node_size=600, 
                           edge_color='black', linewidths=1, font_size=10,
                           width=1, alpha=0.5)

    plt.savefig('dow_30_corr_coeff_circular.png')

Visualizing Correlations Among Dow 30 Stocks Via NetworkX

Dow 30 Correlation Coefficient Matrix

Circular Graph

Spring Graph

Affiliate Disclaimer