Equity · Global · stocks · Monthly · Global Free

ESG Factor Momentum Strategy

Academic paper

Paper linkhttps://www.pm-research.com/content/iijinvest/25/2/113

Strategy in a nutshell

The investment universe consists of stocks in the MSCI World Index. Paper uses MSCI ESG Ratings as the ESG database. The ESG Momentum strategy is built by overweighting, relative to the MSCI World Index, companies that increased their ESG ratings most during the recent past and underweight those with decreased ESG ratings, where the increases and decreases are based on a 12-month ESG momentum. The paper uses the Barra Global Equity Model (GEM3) for portfolio construction with constraints that can be found in Appendix 2. Therefore, this strategy is very specific, but we aim to present the idea, not the portfolio construction. The strategy is rebalanced monthly.

Economic rationale

Socially responsible investing is getting more and more popular among investors. The increased attention leads to an increasing worldwide amount of money invested by responsible investors either for profit or non-profit motives. The main idea of the paper is simple. The research hypothesizes that firms which have improved the ESG the most are expected to outperform other firms. One possibility is to look at the level of ESG scores. Still, the advantages of a better-rated ESG portfolio are expected to be apparent only in the long term, for example, because of increased cash flows.

On the other hand, the momentum compares the rise of the ESG scores. According to the paper, it is expected that the market could react to a change in rating in a relatively short time period. Therefore, the high ESG momentum stocks should be outperforming lower ESG stocks. Moreover, the lowest ESG stocks should underperform.

Backtest performance

Annualised return2.23%

Volatility2.50%

Beta0.822

Sharpe ratio0.15

Sortino ratio0.13

Maximum drawdown26.11%

Win rate64%

Full Python code

1from AlgoLib import *
2from numpy import floor
3#endregion
4
5class ESGFactorMomentumStrategy(XXX):
6
7def Initialize(self):
8self.SetStartDate(2009, 6, 1)
9self.SetCash(100000)
10
11# Decile weighting.
12# True - Value weighted
13# False - Equally weighted
14self.value_weighting = True
15
16self.symbol = self.AddEquity('SPY', Resolution.Daily).Symbol
17
18self.esg_data = self.AddData(ESGData, 'ESG', Resolution.Daily)
19self.tickers = []
20
21self.holding_period = 3
22self.managed_queue = []
23self.quantile = 10
24
25# Monthly ESG decile data.
26self.esg = {}
27self.period = 14
28
29self.latest_price = {}
30
31self.selection_flag = False
32self.UniverseSettings.Resolution = Resolution.Daily
33self.AddUniverse(self.CoarseSelectionFunction, self.FineSelectionFunction)
34
35def OnSecuritiesChanged(self, changes):
36for security in changes.AddedSecurities:
37    security.SetFeeModel(CustomFeeModel())
38    security.SetLeverage(10)
39
40def CoarseSelectionFunction(self, coarse):
41if not self.selection_flag:
42    return Universe.Unchanged
43
44self.latest_price.clear()
45
46selected = [x for x in coarse if (x.Symbol.Value).lower() in self.tickers]
47
48for stock in selected:
49    symbol = stock.Symbol
50    self.latest_price[symbol] = stock.AdjustedPrice
51
52return [x.Symbol for x in selected]
53
54def FineSelectionFunction(self, fine):
55fine = [x for x in fine if x.MarketCap != 0]
56
57momentum = {}
58
59# Momentum calc.
60for stock in fine:
61    symbol = stock.Symbol
62    ticker = symbol.Value
63    # ESG data for 14 months is ready.
64    if ticker in self.esg and self.esg[ticker].IsReady:
65        esg_data = [x for x in self.esg[ticker]]
66        
67        esg_decile_2_months_ago = esg_data[1]
68        esg_decile_14_months_ago = esg_data[13]
69        
70        if esg_decile_14_months_ago != 0 and esg_decile_2_months_ago != 0:
71            # Momentum as difference.
72            # momentum_ = esg_decile_2_months_ago - esg_decile_14_months_ago
73            
74            # Momentum as ratio.
75            momentum_ = (esg_decile_2_months_ago / esg_decile_14_months_ago) - 1
76            
77            # Store momentum/market cap pair.
78            momentum[stock] = momentum_
79
80if len(momentum) <= self.quantile:
81    return Universe.Unchanged
82
83# Momentum sorting.
84sorted_by_momentum = sorted(momentum.items(), key = lambda x: x[1], reverse = True)
85quantile = int(len(sorted_by_momentum) / self.quantile)
86long = [x[0] for x in sorted_by_momentum[:quantile]]
87short = [x[0] for x in sorted_by_momentum[-quantile:]]
88
89long_symbol_q = []
90short_symbol_q = []
91
92# ew
93if not self.value_weighting:
94    if len(long) != 0:
95        long_w = self.Portfolio.TotalPortfolioValue / self.holding_period / len(long)
96        long_symbol_q = [(x.Symbol, floor(long_w / self.latest_price[x.Symbol])) for x in long]
97    
98    if len(short) != 0:
99        short_w = self.Portfolio.TotalPortfolioValue / self.holding_period / len(short)
100        short_symbol_q = [(x.Symbol, -floor(short_w / self.latest_price[x.Symbol])) for x in short]
101# vw
102else:
103    if len(long) != 0:
104        total_market_cap_long = sum([x.MarketCap for x in long])
105        long_w = self.Portfolio.TotalPortfolioValue / self.holding_period
106        long_symbol_q = [(x.Symbol, floor((long_w * (x.MarketCap / total_market_cap_long))) / self.latest_price[x.Symbol]) for x in long]
107    
108    short_symbol_q = []
109    if len(short) != 0:
110        total_market_cap_short = sum([x.MarketCap for x in short])
111        short_w = self.Portfolio.TotalPortfolioValue / self.holding_period
112        short_symbol_q = [(x.Symbol, -floor((short_w * (x.MarketCap / total_market_cap_short))) / self.latest_price[x.Symbol]) for x in short]
113
114self.managed_queue.append(RebalanceQueueItem(long_symbol_q + short_symbol_q))
115
116return [x.Symbol for x in long + short]
117
118def OnData(self, data):
119new_data_arrived = False
120
121if 'ESG' in data and data['ESG']:
122    # Store universe tickers.
123    if len(self.tickers) == 0:
124        # TODO '_typename' in storage dictionary?
125        self.tickers = [x.Key for x in self.esg_data.GetLastData().GetStorageDictionary()][:-1]
126
127    # Store history for every ticker.
128    for ticker in self.tickers:
129        ticker_u = ticker.upper()
130        if ticker_u not in self.esg:
131            self.esg[ticker_u] = RollingWindow[float](self.period)
132        
133        decile = self.esg_data.GetLastData()[ticker]
134        self.esg[ticker_u].Add(decile)
135        
136        # trigger selection after new esg data arrived.
137        if not self.selection_flag:
138            new_data_arrived = True
139
140if new_data_arrived:
141    self.selection_flag = True
142    return
143
144if not self.selection_flag:
145    return
146self.selection_flag = False
147
148# Trade execution
149remove_item = None
150
151# Rebalance portfolio
152for item in self.managed_queue:
153    if item.holding_period == self.holding_period:
154        for symbol, quantity in item.symbol_q:
155            self.MarketOrder(symbol, -quantity)
156                    
157        remove_item = item
158        
159    elif item.holding_period == 0:
160        open_symbol_q = []
161        
162        for symbol, quantity in item.symbol_q:
163            if quantity >= 1:
164                if symbol in data and data[symbol]:
165                    self.MarketOrder(symbol, quantity)
166                    open_symbol_q.append((symbol, quantity))
167                    
168        # Only opened orders will be closed        
169        item.symbol_q = open_symbol_q
170        
171    item.holding_period += 1
172    
173if remove_item:
174    self.managed_queue.remove(remove_item)
175
176class RebalanceQueueItem():
177def __init__(self, symbol_q):
178# symbol/quantity collections
179self.symbol_q = symbol_q  
180self.holding_period = 0
181
182# ESG data.
183class ESGData(PythonData):
184def __init__(self):
185self.tickers = []
186
187def GetSource(self, config, date, isLiveMode):
188return SubscriptionDataSource("data.quantpedia.com/backtesting_data/economic/esg_deciles_data.csv", SubscriptionTransportMedium.RemoteFile, FileFormat.Csv)
189
190def Reader(self, config, line, date, isLiveMode):
191data = ESGData()
192data.Symbol = config.Symbol
193
194if not line[0].isdigit():
195    self.tickers = [x for x in line.split(';')][1:]
196    return None
197    
198split = line.split(';')
199
200data.Time = datetime.strptime(split[0], "%Y-%m-%d") + timedelta(days=1)
201
202index = 1
203for ticker in self.tickers:
204    data[ticker] = float(split[index])
205    index += 1
206    
207data.Value = float(split[1])
208return data
209
210# Custom fee model.
211class CustomFeeModel(FeeModel):
212def GetOrderFee(self, parameters):
213fee = parameters.Security.Price * parameters.Order.AbsoluteQuantity * 0.00005
214return OrderFee(CashAmount(fee, "USD"))