Selection Process

• Take the top 100 ETFs by Assets. Compare each combination of pairs and find which pairs are cointegrated.
• Do a visual inspection to see which ones are suitable.
• Sort the pair spread time series by volatility (high to low volatility). Higher volatility (standard deviation) pairs are more likely to be profitable.
• Select the top N pairs from the sorted pair list

Remove the ones that are absolutely correlated or include? Looks like they move in absolutely together

cointegrated_pairs = [
('VOO', 'XLY'),
('VTI', 'XLY'), 
('BND', 'LQD'),
('BND', 'XLP'),
('IEFA', 'SPDW'),
('AGG', 'IEF'),
('AGG', 'SDY'),
('AGG', 'XLP'),
('VWO', 'VTEB'),

('IEMG', 'IWR'),
('IJH', 'VO'),
('GLD', 'VYM'),
('GLD', 'DVY'),
('GLD', 'VOE'),
 
('BNDX', 'VEU'),
('BNDX', 'SCHX'),
('BNDX', 'DGRO'),
('BNDX', 'SCHB'),
('BNDX', 'EFV'),
('BNDX', 'COWZ'),

('ITOT', 'XLY'),
('VCIT', 'VNQ'),
('VCIT', 'XLP'),
('MUB', 'LQD'),
('MUB', 'VNQ'),
('MUB', 'TIP'),

('VCSH', 's'),