Backtest Period
1965-2012
Markets Traded
Equities
Maximum Drawdown
Period of Rebalancing
Monthly
Return (Annual)
8.5%
Sharpe Ratio
0.25
Standard Deviation (Annual)
18%
Original paper
SSRN-id3313212.pdf607.1KB
Abstract
We use a novel dataset of company-level innovation measures to identify the most innovative industries based on counts of their applications for foreign worker visas and their patent applications and grants. We are able to build portfolios which overweight these innovative industries and which generate economically significant excess returns, especially in the 2013-2018 period, with low turnover. The results do not appear to be fully explained by risk factors, and the same innovation measures do not predict returns at the single stock level.
Keywords:Â innovative efficiency, asset allocation
Trading rules
- Focus on stocks of companies that have non-missing market equity and SIC classification code from CRSP, and non-negative book equity data at the end of previous fiscal year from COMPUSTAT
- The company must hold a minimum of one patent issued within the rolling scope of the past five years.
- Excluded are financial firms with one-digit SIC codes of six and stocks that are priced below one dollar a share at the beginning of the holding period to reduce the impact of micro-cap stocks
- Technology-linked return (TECHRET) is measured as the average monthly return of technology-linked firms in the technology space, weighted by pairwise technology closeness
- Technology closeness (TECH) between firms is defined as the uncentered correlation between all pairs, counted for every firm i, and month t, based on the number of patents in USPTO technology class over the rolling past five years
- TECH acts as a weight in calculating the average stock return of technology-linked firms and is biased toward firms which are more technologically close to the focal firm.
- TECHRET for firm i and month t is defined as a sum for every i different from j of TECHij multiplied by RETj (past month return of technology-linked firm j) divided by sum of TECHij for every i different from j
- At the beginning of each month, sort all firms into deciles based on the return of their technology-linked portfolios in the previous month
- Invest in the top 10% and divest from the lowest 10%.
- The portfolio is value-weighted and is restructured every month.
Python code
Backtrader
import backtrader as bt
class MyStrategy(bt.Strategy):
def __init__(self):
# Filter stocks by market equity, SIC classification code, and book equity data
compustat = self.datas[0].compustat # assuming COMPUSTAT data is added as the first data feed
crsp = self.datas[1].crsp # assuming CRSP data is added as the second data feed
self.universe = list(set(compustat & crsp)) # intersection of COMPUSTAT and CRSP
self.universe = [s for s in self.universe if s.book_equity_data > 0] # non-negative book equity data
self.universe = [s for s in self.universe if s.price_at_beginning_of_holding_period >= 1] # exclude stocks priced below $1 a share
# Set up rolling-window patent filter
self.patent_window = bt.indicators.PatentWindow(patent_class="TECH", years=5) # assuming patent data is added as an indicator
# Set up technology-linked return
self.techret = bt.indicators.TechnologyReturn(universe=self.universe, patent_class="TECH")
# Set up decile filter
self.decile = bt.indicators.Decile(universe=self.universe, period=1, techret=self.techret)
# Set up value-weighted portfolio and rebalancing
self.weights = {} # dict to hold weight for each stock
self.rebalance_monthly = bt.timer.MonthBegin(0) # rebalance at the beginning of each month
def next(self):
if not self.rebalance_monthly:
return
# Calculate stock weights
for i, s in enumerate(self.universe):
if self.decile[i] == 10:
self.weights[s] = s.market_cap / sum([x.market_cap for x in self.universe if self.decile[self.universe.index(x)] == 10])
elif self.decile[i] == 1:
self.weights[s] = -s.market_cap / sum([x.market_cap for x in self.universe if self.decile[self.universe.index(x)] == 1])
else:
self.weights[s] = 0
# Place orders based on weights
for s in self.universe:
size = self.weights[s] * self.broker.get_cash() / s.close[0]
self.order_target_size(s, size)
# Reset weight dict
self.weights = {}