1/ Carry and Time-Series Momentum: A Match Made in Heaven
"Conditioning trading signals of time-series momentum on the sign of the basis (carry), improves the Sharpe ratio. The improvement in is greater during recessions (when it matters most)."
2/ The authors start with a basket of liquid futures (equities, commodities, fixed income, and foreign exchange).
Basis is the percentage difference between the prices of the front- and back-month futures contracts.
Time-series momentum is the sign of trailing 12-month return.
3/ "The slope coefficient is 0.89 for 1-month basis and 0.90 for 12-month basis with corresponding t-statistics of 13.35 and 14.85. The adjusted R² values are high, 0.67 and 0.68. This result indicates that basis is an important component of returns."
4/ Koijen, Moskowitz, Pedersen, Vrugt find that time-series momentum and carry are different strategies:
5/ "We account for transaction costs during portfolio rebalances and rolls of futures contracts... described in Exhibit 5.
"Sharpe ratio is calculated using monthly excess returns as the ratio of an annualized excess return with geometric compounding and an annualized SD."
6/ The authors modify time-series momentum by allowing it to go long or short only when the basis (carry) signal is in agreement.
The result is improved Sharpe ratios (for both inverse vol and equal weighting and for both 1-month and 12-month basis signals).
7/ "We find a consistent pattern of the strategy delivering a higher Sharpe ratio in economic terms, though the difference in performance is statistically significant only during early recessions.... a period associated with very negative performance of the stock market."
8/ Results are similar using Sortino ratios:
Adding the basis (carry) filter to time-series momentum generally improves performance. This is only statistically significant for the first half of recession periods, though this is arguably a time when the difference matters most.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
3/ "Value, momentum & defensive/quality applied to US individual stocks has a t-stat of 10.8. Data mining would take nearly a trillion random trials to find this.
"Applying those factors (+carry) across markets and asset classes gets a t-stat of >14."
2/ "The model's four terms describe different life stages for an individual who marries during the sample period. The intercept reflects the average life satisfaction of individuals in the baseline period [all noncohabiting years that are at least one year before marriage]."
3/ " 'How satisfied are you with your life, all things considered?' Responses are ranked on a scale from 0 (completely dissatisfied) to 10 (completely satisfied).
"We center life satisfaction scores around the annual mean of each population subsample in the original population."
1/ Short-sightedness, rates moves and a potential boost for value (Hanauer, Baltussen, Blitz, Schneider)
…
* Value spread remains wide
* Relationship between value and rates is not structural
* Extrapolative growth forecasts drive the value premium
… robeco.com/en-int/insight…
2/ "The valuation gap between cheap and expensive stocks remains extremely wide. This signals the potential for attractive returns going forward."
3/ "We observe a robust negative relationship between value returns and changes in the value spread.
"The intercept of ≈10% can be interpreted as a cleaner estimate of the value premium, given that it is purged of the time-varying effects of multiple expansions & compressions."
2/ Part 1: Basic directional strategies
Part 2: Adjusted trend, trend and carry in different risk regimes, spot trend, seasonally-adjusted carry, normalized trend, asset class trend
Part 3: Breakouts, value, acceleration, skew
Part 4: Fast mean reversion
Part 5: Relative value
3/ Related reading
Time-Series Momentum
Two Centuries of Trend Following
https://t.co/R6JQb6Cg96
Carry
https://t.co/poFk6OWQsO
Value and Momentum Everywhere
https://t.co/l0wVgAOrhL
2/ "The broadly similar pattern of adverse health and well-being reported as new-onset at 6- and 12 months among test-positives and test-negatives highlights the non-specific nature of these symptoms and suggests that multiple aetiologies may be responsible."
3/ Related reading:
Efficacy of Vaccination on Symptoms of Patients With Long COVID