Demographic Changes Predict Stock Market Returns
Log in to collectAcademic paper
Strategy in a nutshell
Invest in 15 developed countries based on senior population share: go long in countries with below-average elderly populations, short where it exceeds one standard deviation above the mean. Portfolios rebalance annually.
Economic rationale
Market trends are influenced by generational asset supply and demand. Middle-aged investors buy more, driving prices up, while older populations sell to fund retirement, exerting downward pressure. Demographic shifts create predictable effects on equity returns.
Backtest performance
Annualised return8.93%
Beta0.022
Win rate64%
Full Python code
from AlgorithmImports import *
import numpy as np
#endregion
class DemographicChanges(QCAlgorithm):
def Initialize(self):
self.SetStartDate(2000, 1, 1)
self.SetCash(100000)
self.symbols = {'AU':'EWA', 'AT':'EWO', 'BE':'EWK', 'CA':'EWC', 'FR':'EWQ', 'DE':'EWG', 'IT':'EWI', 'JP':'EWJ', 'NL':'EWN', 'ES':'EWP', 'SE':'EWD', 'CH':'EWL', 'GB':'EWU', 'US':'SPY'}
for symbol in self.symbols:
data = self.AddEquity(self.symbols[symbol], Resolution.Daily)
data.SetLeverage(10)
self.demographic_data:Symbol = self.AddData(DemographicData, 'DemographicData').Symbol
self.recent_month:int = -1
def OnData(self, data:Slice) -> None:
if self.recent_month == self.Time.month:
return
self.recent_month = self.Time.month
# rebalance once a year
if self.Time.month != 1: return
age_data = {}
if self.demographic_data in data and data[self.demographic_data]:
for symbol in self.symbols:
age_data[symbol] = data[self.demographic_data].GetProperty(symbol)
if len(age_data) == 0:
self.Liquidate()
return
age_data_values = [x[1] for x in age_data.items()]
age_data_mean = np.mean(age_data_values)
age_data_std = np.std(age_data_values)
long = []
short = []
for symbol in self.symbols:
if age_data[symbol] < age_data_mean - age_data_std:
long.append(self.symbols[symbol])
elif age_data[symbol] > age_data_mean + age_data_std:
short.append(self.symbols[symbol])
# liquidate
invested = [x.Key.Value for x in self.Portfolio if x.Value.Invested]
for symbol in invested:
if symbol not in long + short:
self.Liquidate(symbol)
# trade exxecution
long_count = len(long)
short_count = len(short)
for symbol in long:
if symbol in data and data[symbol]:
self.SetHoldings(symbol, 1 / long_count)
for symbol in short:
if symbol in data and data[symbol]:
self.SetHoldings(symbol, -1 / short_count)
# Demographic data - population 65+ yo
# NOTE: IMPORTANT: Data order must be ascending (date-wise)
from dateutil.relativedelta import relativedelta
class DemographicData(PythonData):
def GetSource(self, config, date, isLiveMode):
return SubscriptionDataSource("data.quantpedia.com/backtesting_data/economic/population_65_over.csv", SubscriptionTransportMedium.RemoteFile, FileFormat.Csv)
def Reader(self, config, line, date, isLiveMode):
data = DemographicData()
data.Symbol = config.Symbol
if not line[0].isdigit(): return None
split = line.split(';')
# Data example:
# year;AU;AT;BE;CA;FR;DE;IT;JP;NL;ES;SE;CH;GB;US
# 2000;12.4;15.4;16.7;12.5;16.2;16.2;18.1;17.4;13.5;16.5;17.3;15.2;15.8;12.4
#
# YEARLY DATA
data.Time = datetime.strptime(split[0], "%Y") + relativedelta(months=12) # NOTE: Preventing of look ahaead bias. Add 12 months so this year we see last year's data.
data['AU'] = float(split[1])
data['AT'] = float(split[2])
data['BE'] = float(split[3])
data['CA'] = float(split[4])
data['FR'] = float(split[5])
data['DE'] = float(split[6])
data['IT'] = float(split[7])
data['JP'] = float(split[8])
data['NL'] = float(split[9])
data['ES'] = float(split[10])
data['SE'] = float(split[11])
data['CH'] = float(split[12])
data['GB'] = float(split[13])
data['US'] = float(split[14])
data.Value = float(split[1])
return data