Economic surprises and commodity future returns

Jupyter Notebook

Surprises in industry and construction activity are plausible predictors of short-term future returns for commodities that are heavily used in these sectors. This hypothesis can be evaluated using quantamental economic surprises, which are point-in-time differences between estimated market expectations and actual reported values of economic indicators, measured at a daily frequency.
We specifically assess the predictive power and economic value of global surprises in survey and production indicators across 32 economies. Country-level surprises are first standardised and then aggregated into global composites, weighted by each economy’s share in global industry. Since 2000, the resulting global surprise indicator has been a statistically significant predictor of returns for an industrial commodity futures basket. A simple daily trading strategy based solely on this signal would have generated material risk-adjusted returns with little correlation to global equity markets. Further hedging against non-growth-related factors, such as monetary risk proxies, can improve performance.

(more…)

Cross-country relative duration strategies with macro factors

Jupyter Notebook

Cross-country relative duration strategies take vol-adjusted fixed receiver positions in interest rate swaps of one currency area versus others. Similar to directional IRS strategies, cross-country trading factors can be based on point-in-time indicators of economic developments. The main difference is that cross-country factors are typically related to relative economic performance.
This post constructs 12 conceptual macro factors that theoretically should help predict cross-country vol-targeted duration returns. Empirically, all of them have displayed at least some modest predictive power, while their cross-correlations have been mostly modest and very diverse. Composite signals can be generated by conceptual parity or sequential machine learning methods. All of these have produced material, consistent, and uncorrelated risk-adjusted returns over the past two decades with few major drawdowns.

(more…)

Macro-aware risk parity

Jupyter Notebook

Risk parity is an investment strategy that allocates risk exposure equally across asset types through volatility-based calibration and leverage. A most profitable risk parity strategy in the past decades has been the equity-duration “long-long”, which harvests combined equity and long fixed-income risk premia, while containing return volatility through diversification. Alas, this position is vulnerable to “autonomous” monetary tightening shocks, i.e., surges in real interest rates without commensurate economic improvements.
The probability of such downside shocks depends on the macroeconomic environment. Hence, by adapting volatility-targeted risk-parity positions to economic overheating measures, strategies become macro-aware and more information-efficient. We show that simple point-in-time macro scores of excess inflation, growth, and overconfidence have predicted equity-duration returns across eight developed markets and for over three decades. Macro-aware signals have materially outperformed simple volatility-targeted positions, both in terms of absolute and risk-adjusted returns. The long-term effects of macro awareness on cumulative net asset value have been huge.

(more…)

Equity trend-following with market and macro data

Jupyter Notebook

The popularity of trend-following bears the risk of market excesses. Medium-term market price trends often fuel economic trends that eventually oppose them (”macro headwinds”). Fortunately, relevant point-in-time economic indicators can provide critical information on the sustainability of medium-term market movements and are a natural complement to standard trend signals.
This post illustrates the benefit of combining market and macro trends in equity markets. Since the 1990s, “robust” market price trend signals alone have created value for equity index future strategies in developed markets. However, risk-adjusted returns would have been enhanced materially if one adjusted market signals for natural macro headwinds, such as the state of the business cycle, inflation, and equity valuations.

(more…)

Boosting macro trading signals

Jupyter Notebook

Boosting is a machine learning ensemble method that combines the predictions of a chain of basic models, whereby each model seeks to address the shortcomings of the previous one. This post applies adaptive boosting (Adaboost) to trading signal optimisation. Signals are constructed with macro factors to guide positioning in a broad range of global FX forwards.
Boosting is beneficial for learning from a wide and heterogeneous set of markets over time, because it is well-suited for exploiting the diversity of experiences across countries and global economic states. Empirically, we generate machine learning-based signals that use regularized regression and random forest regression, and compare processes with and without adaptive boosting methods. For both regression types, machine learning prefers boosting as datasets get larger and, by doing so, creates more profitable signals.

(more…)

Quantamental economic surprise indicators: a primer

Jupyter Notebook

Quantamental economic surprises are point-in-time measures of deviations of economic indicators from expected values. There are two types of surprises: first-print events and pure revisions. First-print events feature new observation periods, and the surprise element depends on market expectations of the indicator. Market surveys can approximate such expectations, but only for a limited number of indicators. Quantamental surprises use econometric prediction models and can be calculated for all indicators and transformations, principally using the whole information state.
This post introduces economic surprises in global industry and construction and shows how they can be transformed into short-term macro trading signals for commodities. There is clear empirical evidence for the predictive power of such surprises for a basket of industrial commodity futures at a daily and weekly frequency. Related simulated PnL generation produces risk-adjusted alpha, albeit mainly in seasons of large swings in manufacturing and construction.

(more…)

Cross-country equity risk allocation with statistical learning

Jupyter Notebook

Macroeconomic factors plausibly cause divergences in equity market returns across countries. Factors related to monetary policy, financial conditions, and competitiveness should all systematically help detect such divergences. In the presence of rational inattention, point-in-time indicators should also have predictive power.
We apply statistical learning to investigate the signalling value of nine candidate macro factors for cross-country excess returns within major equity sectors for a set of 12 countries. Most factors turn out to be relevant predictors. Cross-country trading signals based on point-in-time macro factors and models would have added material uncorrelated PnL value to an equity portfolio.

(more…)

Conditional short-term trend signals

Jupyter Notebook

There are plausible relations between past and future short-term trends across and within financial markets. This is because market returns affect expected physical payoffs, risk premia, and the monetary policy outlook. However, the relations between past and future returns are unstable and often depend on the economic environment. As an example, this post shows that the impact of short-term commodity future trends on subsequent S&P500 future returns depends on the inflationary pressure in the U.S. economy. Empirical analysis suggests that macro-conditional trend signals outperform unconditional short-term trend signals regarding predictive power, accuracy and naïve PnL generations.

(more…)

FX trading signals: Common sense and machine learning

Jupyter Notebook

Two valid methods to combine macro trading factors into a single signal are “conceptual parity” and machine learning. Conceptual parity takes a set of conceptually separate normalized factors and gives them equal weights. Machine learning optimizes models and derives weights sequentially, potentially with theoretical restrictions. Both methods support realistic backtests. Conceptual parity works best in the presence of strong theoretical priors. Machine learning works best with large homogenous data sets.
We apply conceptual parity, and two machine learning methods to combine 11 macro-quantamental trading factors for developed and emerging market FX forwards in 16 currencies since 2000. The signals derived by all methods have been highly significant predictors and produced material and uncorrelated risk-adjusted trading returns. Machine learning methods have failed to outperform conceptual parity, probably reflecting that theoretical priors in the FX space are abundant while data are limited and heterogeneous.

(more…)

U.S. Treasuries and macro-enhanced trend following

Jupyter Notebook

Trend-following strategies rely on the persistence of market trends. Such persistence can arise from the gradual dissemination of information or behavioural biases. In light of these inefficiencies, trends that coincide with supporting economic information (macro tailwinds) are more likely to persist than those accompanied by opposing macro information (macro headwinds). As a result, a macro-enhancement of standard trend-following signals should produce better investment returns.
This post supports this proposition for the U.S. Treasury market over the past 32 years. It tests simple macro enhancement types for directional return trends and curve-flattening return trends. In all cases, macro enhancement would have materially improved predictive power and backtested trading profits. This echoes previous research for other asset classes that illustrated the complementarity of price and economic information in systematic trading.

(more…)