AI-Assisted Hedging Strategies

For long-term investors and long-only asset managers, finding effective hedging strategies is one of the most elusive goals in portfolio management. In recent years, several new ETFs and mutual funds were launched with the aim to provide some downside protection at the cost of limiting upside gains. Many of these funds use ideas similar to the widely adopted collar option strategies and its variations. However, these approaches have proven to be expensive and ineffective during the recent bear market.
 

At Big Data Federation, we believe that harnessing big data and examining markets from multiple angles empowers us to develop cost-effective hedging strategies capable of enhancing the risk-return profile of the overall hedged portfolio. Key ingredients here include the ability to systematically monitor several signals and to execute the hedging trades with discipline.
 

In our hedging mini-series publications, we build hedging strategies and evaluate how these strategies and the core portfolio behave under different market conditions using OdinUltra’s Portfolio Builder functionality. Users can build hedging strategies by applying complex logic on fundamental, technical, economic, events, or macro indicators. Users can also import their own time series of returns of their preferred strategies.
 

For our first example, we consider a typical large-cap portfolio represented by SPDR S&P 500 ETF Trust (SPY). Assume a user wants to develop a strategy to hedge this core holding, SPY, when the markets have run up significantly. We will use the technical indicator known as the Bollinger Band. When SPY is above the 2-standard deviations (std) upper 50-day Bollinger Band, we can write:

1. 0.5 std out-of-the-money (OTM) monthly calls, or

2. At-the-money (ATM) monthly calls, or

3. 0.5 std in-the-money (ITM) monthly calls.

The notional amount of these calls is assumed to be 100% of AuM. Users can quickly backtest these hedging strategies through the Single Strategy functionality and compare through the Multi-Strategy functionality.

The NAV curve of these strategies are as follows:
As observed on the chart, the third strategy performs the best. Additional performance metrics are available from OdinUltra.

In the following chart, we display the performance of SPY versus SPY hedged with the third strategy. Note that when the hedging strategy is triggered, we have a covered call striked at 0.5-std in-the-money and about 1 month to expiration. The hedged portfolio outperforms the unhedged SPY benchmark in total return and risk-return measures, such as Sharpe ratios, in the testing period.

Backtesting these strategies requires our calculation engine to accurately identify the correct options. Data cleansing, modeling, and validation are essential parts of the process. For a typical quantitative researcher, generating the above results can take anywhere from a few days to several weeks, depending on data availability and technology infrastructure. With OdinUltra, these tasks can be accomplished in just a few minutes. Users can also download position and NAV details for further study from OdinUltra if so desired.

In conclusion, systematic and disciplined hedging strategies can be built on OdinUltra using complex logics to improve risk-return characteristics. Users can quickly generate detailed backtest results and may further improve on these with multiple hedging strategies using OdinUltra, which empowers asset managers to leverage their domain expertise throughout the entire investment process to explore, construct, verify, and deploy investment ideas and theses quickly. 

To learn more details on these strategies or construct your own hedging strategies, you may request a trial at OdinUltra.ai.