Option Strategies and Market Signals: Do They Add Value to Equity Portfolios?

Abstract

This study explores an innovative approach to incorporating option strategies into equity portfolios. It presents an alternative direction that institutional investors could take to overcome their current challenges, in a context where traditionally diversified portfolios of only equity and fixed-income assets have shown weaknesses that make it difficult for these investors to achieve their performance goals within their risk limits. We test whether a set of well-known backward-looking signals from equities markets and less-researched forward-looking ones from options markets can be used to improve the efficiency of two option strategies, namely covered call and protective put. The trend signal appears to be the one that adds the most value to both strategies. This study also shows that increasing complexity through additional trading rules does not improve the results of the more basic option strategies that make use of the signals.

1. Introduction

Institutional investors, such as pension funds, have faced challenging times in achieving their investment mandates in recent years. The past decade, marked by a very low or even negative interest rate environment, prompted an increase in equity risk exposure in these investors’ portfolios, in order to meet return objectives that allow them to match their ever-increasing liabilities. This low-rate policy ended in 2022, followed by a rapid rise in interest rates across much of the developed world, causing a large drop in valuations from both equities and fixed-income assets, in unison. Traditional institutional portfolios, which seek diversification by allocating mainly to equities and bonds, suffered the most, as the low or even negative correlation between these two asset classes vanished. In this context, institutions should reconsider their diversification strategies and explore expanding their universe towards alternative sources of income or portfolio protection. When used correctly and efficiently, options can be powerful tools to adjust the risk and return profiles of equity portfolios, offering investors more possibilities to express their views or objectives. For instance, a covered call strategy can enhance yield without adding downside risk, while a protective or long put strategy can protect the equity portfolio from large drawdowns. However, efficiently integrating option strategies into equity portfolios, from the simplest to the most complex ones, is not straightforward. Empirical studies show that plain vanilla, “passive” put protection is not a favorable strategy to protect an equity portfolio, as the long-term cost in terms of option premia required is simply too expensive, despite the protection attained being more effective than other competing strategies, such as a long–short equity, market-neutral trend-following program [1].

In this study, we test whether a set of backward-looking and forward-looking trading signals can be used to build option strategies that are incorporated into two base equity portfolios. The backward-looking signals, derived from the equity markets, have been tested and exploited by equity asset managers to build active portfolios in long-only format with the aim of outperforming their benchmark indices, or even in long–short, market-neutral formats. Here, we test price momentum, historical volatility, and trend signals. The forward-looking signals, derived from the options markets, have been less researched by practitioners and academics. These signals are the implied volatility and skew. Their interest lies in that they represent investors’ perceptions of future volatility or large movements of a certain underlying asset; hence, the name used here is “forward-looking”.

These signals, forward- and backward-looking, should help in selecting the underlyings to which option positions are taken, which is akin to the selection of underlyings in an equity-only portfolio. This is the aim of our research. After testing the relevance of these signals in the Swiss equity market, we examine whether they are also useful for building a set of option strategies that can be incorporated into Swiss equity portfolios.

The remainder of the article is organized as follows. After a brief literature review in Section 2, in Section 3 we first introduce the dataset and then describe the methodologies used, first, to backtest the existence and relevance of the signals in Swiss equity markets, and second, to build and test the added-value of two strategies (a covered call and a protective put) within a Swiss equity portfolio that apply these same signals to select underlyings on which to take option positions. Section 4 presents the results and Section 5 discusses the main findings and their implications.

2. Literature Review

The literature on portfolio management and risk management offers a rich variety of approaches and methodologies aimed at improving the risk-adjusted performance of institutional portfolios. To some degree, this study explores three of these approaches: the reliance on quantitative trading signals, the incorporation of option strategies, and the study of allocation methods in equity portfolios.

Trading signals have long been researched and applied in practice by quantitative investors. However, most signals are nothing else but the product of a long data mining process made by individuals or institutions with commercial interests, which offer promising results in the backtests but fail to deliver them in real life. The authors of [2] expose this concerning reality by showing the high rate at which “risk factors” are being discovered, with more than 400 “existing” ones already. The authors of [3] compressed this zoo of factors to only 15. While many signals fade once they become widely known, others persist over time, as they reflect a market inefficiency or alternative sources of risk premia.

In this study, we examine three well-known backward-looking trading signals: price momentum (relative, rank-based), low historical volatility (also relative), and trend (time series, absolute signal). These signals have been widely tested across various markets and sample periods. The authors of [4] were pioneers in formally testing the “momentum” signal in the US stock market. The authors of [5] found the momentum signal to be relevant across geographies and asset classes. Furthermore, the authors of [6] demonstrated the existence of the momentum effect even among risk factors. The “trend” signal has also attracted considerable academic and practitioner interest. Studies such as [7,8,9,10,11] not only document the performance of trend-following strategies but more importantly emphasize their desirable characteristics when included in equity portfolios. Due to their convex profile, showing strong performance on highly rising markets and also during large market drawdowns, these strategies are considered to have hedging properties, without the usual cost of other hedge strategies, such as the protective put strategy. The low-risk anomaly gained popularity through the authors of [12], who showed empirically that lower-risk stocks outperformed their higher-risk counterparts. In their study, they use the CAPM beta as a risk measure. More recently, the authors of [13] found this mispricing to be even more pronounced when volatility, and not beta, is used as the risk measure.

The forward-looking signals found in options data, implied volatility and skew, have been less studied, despite their strong potential as they represent unique information on investors’ market sentiments. The authors of [14] used these same signals to build trading strategies on US large-capitalization equities that outperformed their benchmarks.

Regarding option strategies, the author of [15] stresses how these can be used to enhance a portfolio’s performance and outperform the market. The authors of [16,17,18] explore the effectiveness of portfolio insurance strategies using options. Moreover, the authors of [1] compare the hedging effectiveness of a long-put strategy vs. a trend-following long–short strategy.

The last approach to portfolio management examined in this study relates to allocation methods for equity portfolios. The authors of [19] showed that a naïve equally weighted (EW) allocation outperformed optimized portfolios, with the latter being heavily impacted by small estimation errors in return or risk parameters. The authors of [20] introduced the equal risk contribution portfolio (ERC), which, instead of assigning equal weights to each asset, allocates weights such that each asset contributes equally to overall portfolio risk, which is measured by volatility. Their study analytically demonstrates that ERC weights always lie between those of the EW and the global minimum variance (GMV) portfolios. The authors of [21] proposed a version of ERC that uses expected shortfall (also known as Conditional Value-at-Risk or CVaR) as the risk measure. The authors of [22] also proposed the use of CVaR as the risk measure but employed it as the objective function to be minimized (similar to volatility in the GMV method). Finally, the authors of [23] suggested placing constraints on the optimized weights derived from the mean-variance allocation, in order to reduce the impact from estimation errors.

All these cited studies have influenced the authors of this article. Yet to the best of our knowledge, no previous research has explored the application of trading signals to build option strategies and integrate them into equity portfolios (this article is an extension of a previous one from the same authors [24], which was presented at the 6th International Conference on Finance, Economics, Management and IT Business in April 2024 in Angers, France (FEMIB 2024). The article focused on the long-put strategy only).

3. Materials and Methods

3.1. Data

The data used in this research, which focuses on the Swiss market, comes from two datasets: an equity dataset and an options dataset. The equity dataset includes daily stock prices, dividends, and volume data from all stocks that are part of the Swiss Performance Index (SPI) and, therefore, comprises more than 200 stocks from both large-, mid-, and small-capitalization companies. After excluding stocks whose missing data represents more than 3% of their own sample period available from beginning to end (due to lack of trading volume), we have 163 underlyings. The excluded stocks are small-capitalization and very small-capitalization («micro-caps») companies, with no sufficient trading volume and daily observations to backtest the strategies reliably. Despite the large number of stocks being excluded, all of these small- and micro-caps represent less than 1% of the Swiss equities market. The sample period spans from 1 January 2003 to 31 January 2022. However, note that many underlyings have a shorter sample period available. The options dataset includes only those companies that are part or have been part of the Swiss Market Index (SMI), and therefore it includes only large-capitalization companies. Small-capitalization companies lack the necessary trading volume and liquidity in the options markets. Restricting the investment universe to underlyings with sufficient liquidity in the options market significantly limits the pool of eligible stocks, effectively excluding small-capitalization firms. At first glance, this constraint may introduce selection bias and, potentially, survivorship bias. Nevertheless, all strategies—whether incorporating options or not—are evaluated within the same constrained universe, thereby mitigating the influence of selection bias to the extent feasible. Furthermore, implementing the proposed strategies using less liquid underlyings would be impractical, as options on such assets, if available at all, tend to trade infrequently and are characterized by wide bid–ask spreads.

There are 27 underlyings in the time series and data for each underlying consists of the historical parameters to construct volatility surfaces: the set of implied volatilities and skew values for all combinations of maturities and strike prices that were available at each date, from March 2006 to January 2022, corresponding to millions of data points. This allows us to accurately value any option, call, or put, with any combination of strike and maturity, at any point in time and for any underlying asset in the sample. The options dataset also includes the same type of options data for the SMI benchmark. For a proxy of the Swiss risk-free rate, we use the overnight rate SARON.

3.2. Backtesting the Relevance of Trading Signals

The first step is to investigate the existence and pertinence of the trading signals in the Swiss equities market. This section introduces the signals in more detail, describes the testing methodology, and presents the main results.

3.2.1. Trading Signals

The first group of signals uses information from the stocks’ past behaviors, being backward-looking. It is made of momentum, low historical volatility, and trend signals.

Momentum (MOM): Past performance is used to define whether an underlying shows a positive momentum signal, a negative one, or no signal. MOM is a relative, ranking-based signal. At each rebalancing period, all underlyings are ranked according to their past return on a predefined window (6, 12, or 24 months). The top quantiles (25% or 33%) are classified as past “winners” and therefore have a positive signal. The bottom quantile, the past “losers”, is a negative one. Being a relative signal implies that, provided that the total number of underlyings remains constant, there will always be the same number of “winners” and “losers”, even though the names might change over time.

Historical volatility (HVOL): Past volatility is the measure to rank the underlyings. Then, at each rebalancing period, the underlyings with the highest historical volatility show a negative signal, while the ones with the lowest volatility show a positive one. The estimation windows and quantiles tested are the same as the momentum signal.

Trend (TREND): Similarly to momentum, trend also uses past performance to define the signal. However, trend is an absolute signal (i.e., not ranking-based). At each rebalancing, all underylings with a positive performance over predefined windows have a positive trend signal and all those with a negative return have a negative trend signal. Therefore, at any rebalancing period, all stocks in the portfolio can show a positive trend, a negative one, or no trend signal. In order to have a positive trend, an underlying must have a positive return over the past 12, 6, and 3 months. We also test a variation that excludes any underlying that shows a positive trend over the last 1-month period.

The second group of signals is those that are derived from the options markets, and express investors’ sentiment or expectations on future market risk, hence the name “forward-looking”. These signals are the implied volatility and skew. To calculate these signals, a time series of daily implied volatility and skew values for 12-month maturity and at-the-money (ATM) options in each underlying is generated.

Implied volatility (IVOL): At each rebalancing, underlyings are ranked by their implied volatility observed over the previous 3 or 7 days. Then, similar to the historical volatility signal, the stocks with the largest implied volatility show a negative signal, while the ones with the lowest value show a positive signal. The quantiles tested are the same ones with momentum and historical volatility (25% and 33%).

Skew (SKEW): Skew measures the difference between the implied volatility of an ATM option vs. an in-the-money (ITM) or an out-of-the-money (OTM) option, with the same maturities. The strategy ranks the underlyings by the most recent skew measure (average values of the previous 3 or 7 days), imposing a negative signal to those underlyings with the highest skew value and a positive one to stocks with the lowest skew value.

Despite the forward-looking signals having been less researched than backward-looking ones, there is a priori no rationale to expect some signals to outperform the others over the entire sample period. Yet the convex return profile of trend-following strategies could suggest that the TREND signal is the most suitable to protect portfolios from the worst equity market drawdowns, especially when implemented in the long-put strategy, which underperforms in less volatile market regimes.

3.2.2. Testing Methodology

In order to test for the relevance and added value of these signals in the Swiss market, for each signal we backtest two long-only portfolios: one that invests in the underlyings with a positive signal (EW) and another one that allocates to stocks with a negative signal (also EW). The signals are calculated, and the portfolios are rebalanced monthly. To ensure that our results are robust, we use multiple combinations of percentiles (or quantiles) and lookback windows to calculate the signals and choose the underlyings. Then, we compare the Sharpe ratios of these two long-only portfolios.

Additionally, we backtest a self-financed long–short, market-neutral (in nominal, CHF terms) portfolio. At each rebalancing date, it invests long into underlyings with a positive signal, EW, and sells short the same amount, also EW, those underlyings with a negative signal. The monthly rebalancement enforces the gross leverage ratio to 2 and the net leverage (long–short) of zero each month.

3.3. Applying the Trading Signals to Option Strategies to Complement Equity Portfolios

Next, we test whether the same trading signals can be used to build active option strategies that, when incorporated into a base equity portfolio, can improve the portfolio’s results. We implement two base option strategies: a covered call strategy that aims at providing a yield enhancement instrument to the portfolio, at the expense of some upside gain, and a long put strategy whose aim is to partially preserve the portfolio during market drawdowns, at the cost of put premia, which sacrifice some return each period (the aim of this article is to evaluate the effectiveness of trading signals within two base equity option strategies. While more complex strategies such as collars, straddles, or strangles could be used, focusing on these two is sufficient to assess the added value of the signals, which can later be applied to more advanced structures).

The signals should help in selecting the underlyings to which the option positions are taken, with the purpose of minimizing the loss on the upside gain (covered call) or the amount spent on option premia (long put), but without compromising the objectives of long-term strategies which must offer attractive risk–return profiles. In both option strategies, additional trading rules (TRs) are tested, with the intention of improving the efficiency of the basic strategies. These TRs are explained in more detail below.

In order to test their added value, these sets of option strategies are incorporated into two base equity allocations, the EW and the GMV. We use two testing methodologies: classic historical backtests, in which the strategies are implemented from the beginning to the end of the sample period, and bootstrap tests, where we apply the strategies to 500 randomly selected one-year subsamples. In both cases, the portfolios, including the options’ positions, are rebalanced every 6 months, coinciding with the options’ maturities. Likewise, the portfolios’ value and return results are calculated every 6 months (calculating every 6-month period, coinciding with the rebalancing dates and options’ maturities, allows us to compute the portfolio’s P&L using the options’ payoffs. Estimating all the options’ values at a higher frequency, e.g., monthly, to calculate the portfolio’s value would require computational resources that exceed our capabilities).

3.3.1. Covered Call for Yield Enhancement

The covered call strategy aims at providing a fixed income-like return stream by collecting option premia from selling, at each rebalancing date, 110 OTM (i.e., the strike price is set at a level 10% above the spot price) call options with 6-month maturities, on underlyings that are part of the portfolio and show a negative signal (negative trend or momentum, high historical or implied volatility, or high skew). The use of “negative” signals here must minimize the opportunity cost since it should minimize the likelihood that the underlying will increase. The signal is applied using the 25th percentile and the prior 12-month returns for MOM and HVOL and the most recent 3-day returns for IVOL and SKEW. As TREND is an absolute signal, no percentile is applied and its windows are the 12, 6, and 3 months (the three of them must be satisfied). The signals are re-calculated at each rebalancing date for all the underlyings that compose the base equity portfolio.

The equity-only portfolios, with no options, are the BASE portfolios (BASE, SMI, OPT, LEV, TR1, and TR2 are the names used in the result tables in Section 4 to classify the different portfolios tested). As an additional benchmark, we include a portfolio that takes option positions on the SMI benchmark (“SMI” portfolio). The number of options on the index is beta-adjusted. The OPT portfolios are the ones in which call options are sold on the negative signal underlyings. Moreover, as a benchmark, we include an “ALL signal” that sells calls on all underlyings. Then, we build a LEV portfolio that applies a 20% leverage to equities positions of the OPT portfolio. The premise here is that, as covered calls slightly reduce downside risk (by collecting the premia), a certain degree of leverage can be added, in order to take on more equity risk and enhance the return coming from the underlyings. TR1 is a strategy that adds a trading rule to the OPT strategy: between two rebalancing dates (i.e., during the “investment period”), if an option’s delta decreases to 0.1 or less (i.e., the call is deeply OTM), the call options are bought back, in order to close the positions at a very low cost, thus realizing a profit. Finally, TR2 (for covered call, TR2 is only implemented in the backtests.) is a strategy that adds a different trading rule to the OPT portfolios: after selecting the underlyings with a negative signal to which call options should be sold, it applies a limit of implied volatility. If the implied volatility of an option is below 10%, no options are sold; if it is above 30%, options are sold, regardless of the underlyings showing a signal or not. Stated otherwise, this rule sells options when they are expensive, collecting a high premium, and avoids selling them when they are cheap.

3.3.2. Long Put for Downside Risk Mitigation

The long put strategy is constructed as follows: 90 OTM (i.e., the strike price is 10% below the spot) put options with 6-month maturities are purchased on underlyings that show a negative signal, regardless of being present in the base equity portfolios or not (this is relevant for GMV only, as for EW some positive amount will always be allocated to all underlyings). Instead of applying a purely protective put strategy (one that would buy the exact number of put options that offsets the number of underlyings in the portfolio), a global target of 2.5% of the portfolio value is set to be spent on put premia at each rebalancing, which acts as a “risk budget”. This amount is used to purchase put options on all underlyings with signals, on an EW basis (in practice, due to the varying number of underlyings available in the investment universe, this “risk budget” spent on put options premia is not exactly 2.5% but varying within an interval between 2.3% and 2.8%. For TREND, as it is an absolute signal, it can theoretically range between 0% and 10% of the portfolio value, but the average is equal to 2.44%, which is very close to the budget). This method means the put strategy in the equity portfolio can be partially protective, fully protective, or even in some cases speculative (e.g., when protection is over 100% of the underlying’s value in the portfolio, the combined put and equity positions will benefit from a large drawdown, rather than simply offsetting its impact. The extreme case is when a put is bought on an underlying with no weight in the GMV portfolio). The signals are calculated using the same parameters (percentile and historical window) as with the covered call strategy (we also tested the classic protective put strategy that takes the exact put positions to offset the payoffs (negative) of the underlyings. However, managing positions with a defined “risk budget” proved to be more pertinent and efficient).

The “SMI” portfolio takes long put positions on the SMI, which are beta-adjusted. OPT buys put options on underlyings that show a negative signal. LEV uses a 20% leverage to enhance performance at the expense of some increase in risk, which is expected to be mitigated through the put protection. TR1 applies the following trading rule to OPT: if, during the investment period, a put option’s delta reaches −0.9 or below (i.e., deeply ITM), the put is sold, locking a large profit from this position, which could eventually be lost if there is a quick reversal in the underlying’s price. TR2 implements a similar rule as in TR2 of the covered call strategy previously described: regardless of the signal, put options are bought if the implied volatility of any underlying is below 10% and no position is taken if the implied volatility is above 30%, avoiding the purchase of expensive options and benefiting from supposedly cheap (very low implied volatility) options.

4. Results

4.1. Backtesting the Relevance of Trading Signals

Table 1. Annualized return of the long–short portfolios and Sharpe ratio differential of the two long-only portfolios.

Percentile	Lookback Window	Ann. Ret.	SR Diff.
MOM
	6 M	8.75%	0.71 ***
33%	12 M	5.37%	0.52 **
	24 M	2.08%	0.29
	6 M	11.02%	0.81 ***
25%	12 M	5.71%	0.55 **
	24 M	1.65%	0.29
HVOL
	6 M	−5.43%	0.30
33%	12 M	−3.46%	0.37 *
	24 M	−3.06%	0.37 *
	6 M	−5.62%	0.40 **
25%	12 M	−4.19%	0.44 **
	24 M	−3.77%	0.41 **
TREND
	12–6–3 M	11.36%	1.00 ***
	12–6–3 (ex–1) M	11.14%	0.93 ***
IVOL
33%	3 D	−4.93%	0.03
	7 D	−5.12%	0.04
25%	3 D	−5.95%	0.10
	7 D	−6.37%	0.08
SKEW
33%	3 D	6.98%	0.34
	7 D	6.11%	0.29
25%	3 D	8.13%	0.38
	7 D	8.32%	0.38

Ann. Ret. indicates the long–short portfolios’ annualized returns. SR diff. indicates the difference in the Sharpe ratio between the long-only “positive signal” portfolio and the “negative signal” portfolio. For SR diff., the asterisks indicate the following: *** significant at 99% level; ** significant at 95% level; and * significant at 90% level.

summarizes the results of the signal tests in the Swiss equities market. It shows the long–short portfolio’s annualized return and the Sharpe ratio differential between the long-only portfolio with positive signal underlyings and the portfolio with negative signal underlyings. The results are displayed for all combinations of percentiles and lookback windows. The results in Table 1 provide evidence of strong MOM and especially TREND signals: all portfolio combinations show a positive return, considerably and consistently high for the TREND signal. Moreover, all Sharpe ratio differentials are positive and significant at a minimum 95% level, except the 25th percentile and 24M window case for the MOM signal. The HVOL signal also shows positive and significant Sharpe ratio differentials for almost all combinations. However, the L/S portfolios’ returns are always negative. This is due to the construction process of the L/S portfolio: at each rebalancing date, it enforces nominal neutrality (CHF terms) but not risk neutrality. Therefore, the short leg’s volatility is substantially larger than the long leg’s volatility.

Table 2. Descriptive statistics of the two long-only portfolios and the long–short portfolio.

Port.	Start	End	Ann. Ret.	Vol.	SR	CVaR95	MaxDD	Ret. DD1	Ret. DD2	OS Beta
MOM
High	15.01.04	24.05.22	12.49%	14.97%	0.86	−2.39%	−48.39%	−46.72%	−26.85%	0.72
Low	15.01.04	24.05.22	4.06%	18.17%	0.31	−2.81%	−63.51%	−62.54%	−33.52%	0.81
L/S	15.01.04	24.05.22	5.71%	14.82%	0.45	−2.26%	−57.56%	28.66%	6.65%	−0.09
HVOL
Low	15.01.04	24.05.22	6.84%	8.22%	0.85	−1.30%	−33.90%	−32.86%	−23.36%	0.43
High	15.01.04	24.05.22	6.44%	20.85%	0.40	−3.19%	−70.19%	−69.81%	−33.96%	0.95
L/S	15.01.04	24.05.22	−4.19%	16.55%	−0.18	−2.40%	−70.88%	89.11%	10.86%	−0.49
TREND
Pos	15.01.04	24.05.22	12.11%	13.14%	0.94	−2.12%	−30.96%	−26.64%	−18.86%	0.53
Neg	15.01.04	24.05.22	−1.32%	17.22%	0.01	−2.61%	−58.63%	−53.99%	−34.73%	0.72
L/S	15.01.04	24.05.22	11.14%	15.66%	0.75	−2.30%	−43.34%	47.43%	17.94%	−0.17
IVOL
Low	01.02.06	29.06.20	4.99%	14.52%	0.41	−2.19%	−35.66%	−31.89%	−24.55%	0.70
High	01.02.06	29.06.20	4.72%	28.50%	0.30	−4.32%	−72.44%	−71.05%	−36.72%	1.28
L/S	01.02.06	29.06.20	−5.95%	22.87%	−0.15	−3.51%	−80.61%	87.62%	14.76%	−0.56
SKEW
Low	01.02.06	29.06.20	8.12%	21.89%	0.47	−3.29%	−54.02%	−51.53%	−33.77%	1.00
High	01.02.06	29.06.20	−0.63%	23.34%	0.09	−3.69%	−74.95%	−73.17%	−36.13%	1.09
L/S	01.02.06	29.06.20	8.13%	15.19%	0.59	−2.12%	−34.37%	70.19%	1.65%	−0.06

For each signal, the table shows the statistics of two competing equity long-only portfolios: one that invests in stocks with a positive signal and another invested in stocks with a negative signal, and a long–short portfolio that is long (short) the underlyings with a positive (negative) signal. Ann. Ret. is the annualized return; Vol. is the annualized volatility; SR is the Sharpe ratio; CVaR₉₅ is the Conditional Value-at-Risk at 95% level and expressed in daily returns; MaxDD is the portfolio’s maximum drawdown; Ret. DD1 and Ret. DD2 are the portfolios’ returns during the first and second SPI index drawdown periods from Table A1 in the Appendix A; and OS Beta is the out-of-sample beta coefficient from the regression of each portfolio against the SPI benchmark. The parameters to construct the portfolios are the following: percentile = 25%, lookback window = 12 M (MOM, HVOL), 12–6–3 (ex–1) M (TREND), and 3 D (IVOL, SKEW).

illustrates in more detail the performance and risk of the long–short and the two long-only portfolios. The examples of Table 2 are constructed taking the 25th percentiles and the following lookback windows: 12 months for MOM and HVOL, 12–6–3 (ex–1) months for TREND, and 3 days for IVOL and SKEW. The columns Ret. DD1 and Ret. DD2 represent the portfolios’ return during two equity market drawdowns, detailed in

Table A1. Swiss Performance Index (SPI) drawdowns.

Drawdown	Cause	Return	Start	End
DD1	GFC	−54.99%	01.06.07	09.03.09
DD2	COVID-19	−27.11%	19.02.20	23.03.20

SPI drawdowns are calculated from peak to trough. GFC refers to the 2007–2010 Global Financial Crisis. COVID-19 refers to the equity market drawdown due to the beginning of the coronavirus pandemic.

in the Appendix A. It evidences how the short leg’s volatility is larger than the long leg’s volatility, as the realized volatility of the Low HVOL portfolio is 8.22% and the one of the High HVOL is 20.85%. In addition, the out-of-sample beta (OS Beta, measured against the SPI benchmark) of the L/S portfolio is −0.49. The same effect occurs with the IVOL signal. Nevertheless, in the IVOL case, the Sharpe ratio differentials are not statistically significant. With regard to the SKEW signal, all L/S portfolios show a positive performance but none of the Sharpe ratio differentials are statistically significant, despite being positive. The columns Ret. DD1 and Ret. DD2 in Table 2 reveal that all these signals can be helpful to mitigate the impact of the worst equity market drawdowns: the loss that the “positive signal” portfolios suffer during the two periods of market stress, the GFC and the COVID-19 drawdowns, is remarkably lower than that of the “negative signal” portfolios. Consequently, the L/S portfolios performed very well during these periods, especially during the GFC drawdown. Incorporating these signals into an equity portfolio can be useful not only to enhance its performance but especially to reduce equity market risk.

4.2. Applying the Trading Signals to Option Strategies to Complement Equity Portfolios

Next, we turn to the results of implementing the options strategies within the two base equity allocation frameworks, EW and GMV. In this part of the analysis, the portfolios are constructed and valued every six months, coinciding with the maturity dates of the options used in the strategies. We begin by presenting the results of the covered call strategy, focusing first on the historical backtests and continuing with the findings from the bootstrapped samples, followed by the results of the long-put strategy.

4.2.1. Covered Call for Yield Enhancement

The results from the historical backtests for the covered call portfolios and the variants tested (SMI, LEV, TR1 and TR2) are summarized in

Table 3. Historical backtested results: covered call.

Port.	Measure	SIGNAL
		ALL		TREND		MOM		HVOL		IVOL		SKEW
		EW	GMV	EW	GMV	EW	GMV	EW	GMV	EW	GMV	EW	GMV
	Ann. Ret.	3.39	5.3
BASE	Worst 6 M	−29.5	−18.8
	Ratio	0.12	0.28
	Ann. Ret.	2.97	4.97
SMI	Worst 6 M	−27.5	−17.6
	Ratio	0.11	0.28
	Ann. Ret.	2.89	5.01	3.76	5.64	3.18	5.16	3.29	5.31	3.3	5.18	4.04	5.32
OPT	Worst 6 M	−24.1	−15.6	−25.4	−17	−27.4	−18.8	−27.3	−18.8	−27.3	−18.8	−27.7	−18.8
	Ratio	0.12	0.32	0.15	0.33	0.12	0.27	0.12	0.28	0.12	0.28	0.15	0.28
	Ann. Ret.	3.27	5.63	4.25	6.32	3.59	5.8	3.72	5.96	3.73	5.82	4.55	5.97
LEV	Worst 6 M	−30.4	−19.3	−32.3	−21.1	−34.6	−23.3	−34.6	−23.4	−34.6	−23.4	−35.1	−23.3
	Ratio	0.11	0.29	0.13	0.3	0.1	0.25	0.11	0.26	0.11	0.25	0.13	0.26
	Ann. Ret.	2.93	5.07	3.85	5.74	3.22	5.11	3.37	5.33	3.33	5.19	4	5.25
TR1	Worst 6 M	−24.5	−16.0	−25.8	−17.2	−27.5	−18.8	−27.4	−18.8	−27.4	−18.8	−27.9	−18.8
	Ratio	0.12	0.32	0.15	0.33	0.12	0.27	0.12	0.28	0.12	0.28	0.14	0.28
	Ann. Ret.	3.1	5.62	3.81	5.76	3.28	5.48	3.29	5.31	3.3	5.18	4.12	5.61
TR2	Worst 6 M	−24.1	−15.6	−25.4	−17	−27.4	−18.8	−27.3	−18.8	−27.3	−18.8	−27.7	−18.8
	Ratio	0.13	0.36	0.15	0.34	0.12	0.29	0.12	0.28	0.12	0.28	0.15	0.3

For each strategy, the results show the average historical annualized return (%), Ann. Ret., worst 6-month period return (%), Worst 6 M, and the ratio of avg. return to worst 6 M return, Ratio.

. For each variant, Table 3 exhibits the historical annualized return, the worst 6-month period return, and the ratio of annualized return to the worst 6-month return (in absolute terms). The historical backtests suggest that only TREND and to a lesser extent SKEW signals add value to the BASE portfolios. The TREND portfolios are especially superior, showing a higher return and at the same time lower risk, except for the levered portfolio whose risk is above the BASE, as expected, but it still outperforms on risk-adjusted terms. The other signals do not appear to add value. Finally, additional trading rules, TR1 and TR2, do not outperform simpler option strategies, OPT. This is true for all signals.

The results from the bootstraps tests are shown in

Table 4. Bootstrapped results: covered call.

Port.		SIGNAL
	Measure	ALL		TREND		MOM		HVOL		IVOL		SKEW
		EW	GMV	EW	GMV	EW	GMV	EW	GMV	EW	GMV	EW	GMV
	Avg. Ret.	6.23	5.97
BASE	Worst 5%	−27	−16.8
	Ratio	0.23	0.36
	Avg. Ret.	6.1	5.77
SMI	Worst 5%	−27.1	−16.9
	Ratio	0.23	0.34
	Avg. Ret.	5.68	6.74	5.73	6.38	5.28	5.95	5.87	5.99	5.47	5.88	6.69	6.04
OPT	Worst 5%	−20.7	−12.7	−23.3	−14.1	−25.6	−16.5	−24.7	−16.7	−25.1	−16.7	−23.9	−16.3
	Ratio	0.27	0.53	0.25	0.45	0.21	0.36	0.24	0.36	0.22	0.35	0.28	0.37
	Avg. Ret.	6.76	8.04	6.82	7.6	6.29	7.08	6.99	7.14	6.51	7	7.98	7.2
LEV	Worst 5%	−25.2	−15.5	−28.2	−17.2	−31.1	−20.2	−30	−20.4	−30.4	−20.4	−29	−20
	Ratio	0.27	0.52	0.24	0.44	0.2	0.35	0.23	0.35	0.21	0.34	0.28	0.36
	Avg. Ret.	3.19	5.2	4.75	5.84	4.75	5.88	5.06	5.95	4.6	5.85	6.35	5.93
TR1	Worst 5%	−23.4	−16	−25.5	−15.9	−26.1	−16.6	−25.9	−16.8	−25.8	−16.8	−25.5	−16.4
	Ratio	0.14	0.32	0.19	0.37	0.18	0.35	0.2	0.36	0.18	0.35	0.25	0.36

For each strategy, the results show the average annual return over the 500 bootstraps (%), Avg. Ret., the worst 5% period return (%), Worst 5%, and the ratio of avg. return to worst 5% return, Ratio.

. For each portfolio, Table 4 displays the average annual return over the 500 randomly selected periods, the worst 5% annual return (5th percentile of the 500 samples, equivalent to a Value-at-Risk measure), and the ratio of average return to the worst 5% return. The bootstrap results confirm the findings from the backtests, even though the TREND and SKEW signals only outperform in terms of return/risk ratio (and SKEW shows better results than BASE for EW but similar for GMV). It appears that the OPT strategy that sells calls on all underlyings at each rebalancing (“ALL” signal) outperformed the portfolios that make use of signals. Finally, the difference in performance and risk, and especially the outperformance in risk-adjusted terms of all GMV portfolios with respect to their EW counterparts is noticeable, highlighting the importance of choosing the correct equity allocation method.

4.2.2. Long Put for Downside Risk Mitigation

The results from the backtests of the long-put strategy, presented in

Table 5. Historical backtested results: long put.

Port.		SIGNAL
	Measure	ALL		TREND		MOM		HVOL		IVOL		SKEW
		EW	GMV	EW	GMV	EW	GMV	EW	GMV	EW	GMV	EW	GMV
	Ann. Ret.	3.39	5.3
BASE	Worst 6 M	−29.5	−18.8
	Ratio	0.12	0.28
	Ann. Ret.	0.00	2.84
SMI	Worst 6 M	−16.9	−10.9
	Ratio	0.00	0.26
	Ann. Ret.	−3.51	−2.63	3.77	5.15	2.55	4.09	3.01	4.55	3.14	4.68	2.26	3.81
OPT	Worst 6 M	−10.6	−13.3	−13.3	−8.2	−25.6	−15.2	−25.3	−14.9	−25.3	−14.9	−17.6	−8.6
	Ratio	−0.33	−0.2	0.28	0.63	0.1	0.27	0.12	0.31	0.12	0.31	0.13	0.44
	Ann. Ret.	−4.53	−3.34	4.27	5.8	2.89	4.63	3.41	5.14	3.55	5.28	2.56	4.32
LEV	Worst 6 M	−13.2	−16.1	−16.0	−9.8	−31.3	−18.3	−31.4	−18.2	−31.5	−18.3	−22.0	−10.3
	Ratio	−0.34	−0.21	0.27	0.59	0.09	0.25	0.11	0.28	0.11	0.29	0.12	0.42
	Ann. Ret.	−0.97	0.02	3.58	5.03	3.44	5.01	3.53	5.12	3.54	5.13	2.39	4.04
TR1	Worst 6 M	−18.3	−13	−20.8	−10.8	−27.9	−17.5	−28.1	−17.7	−28.1	−17.7	−25.5	−15.1
	Ratio	−0.05	0.00	0.17	0.46	0.12	0.29	0.13	0.29	0.13	0.29	0.09	0.27
	Ann. Ret.	−0.67	1.07	2.80	4.66	−0.67	1.07	−0.67	1.07	3.48	5.36	2.29	4.14
TR2	Worst 6 M	−29.5	−18.8	−29.5	−18.8	−29.5	−18.8	−29.5	−18.8	−29.5	−18.8	−29.5	−18.8
	Ratio	−0.02	0.06	0.09	0.25	−0.02	0.06	−0.02	0.06	0.12	0.28	0.08	0.22

For each strategy, the results show the average historical annualized return (%), Ann. Ret., worst 6-month period return (%), Worst 6 M, and the ratio of avg. return to worst 6 M return, Ratio.

, also indicate that the TREND signal is the most effective among those tested for selecting the underlyings on which to take option positions. Specifically, the TREND signal substantially reduces the portfolio’s risk without sacrificing annual return when compared with the BASE signal, the equity-only portfolio, with no option positions. This suggests that the TREND signal offers a compelling balance between downside protection and performance preservation. The SKEW signal contributes positively to the GMV allocation, although it fails to show the same effectiveness under the EW allocation scheme. Consistent with the findings from the covered call strategy, the introduction of additional trading rules, TR1 and TR2, does not consistently improve the performance of the portfolios. Their impact appears to be in the best-case marginal and in most cases negative. Finally, as observed previously, portfolios constructed under the GMV allocation scheme outperform their EW counterparts in all settings: whether options are included or not and regardless of the signal and trading rule applied.

The bootstrapped results from

Table 6. Bootstrapped results: long put.

Port.		SIGNAL
	Measure	ALL		TREND		MOM		HVOL		IVOL		SKEW
		EW	GMV	EW	GMV	EW	GMV	EW	GMV	EW	GMV	EW	GMV
	Avg. Ret.	6.44	6.67
BASE	Worst 5%	−24.2	−14.6
	Ratio	0.27	0.46
	Avg. Ret.	2.25	3.72
SMI	Worst 5%	−23.6	−15.6
	Ratio	0.1	0.24
	Avg. Ret.	−8.62	−8.96	4.59	4.57	3.15	3.22	4.69	4.75	4.83	4.91	2.12	2.15
OPT	Worst 5%	−28.7	−29.1	−18.5	−11.7	−17.8	−10.2	−13.6	−9.1	−14.9	−9.2	−20.6	−12.4
	Ratio	−0.3	−0.31	0.25	0.39	0.18	0.31	0.34	0.52	0.32	0.54	0.1	0.17
	Avg. Ret.	−10.39	−10.8	5.47	5.45	3.74	3.82	5.59	5.67	5.76	5.85	2.5	2.54
LEV	Worst 5%	−34.4	−34.8	−22.6	−14	−21.2	−12.3	−16.6	−10.9	−17.8	−11.1	−24.6	−15.1
	Ratio	−0.3	−0.31	0.24	0.39	0.18	0.31	0.34	0.52	0.32	0.53	0.1	0.17
	Avg. Ret.	−4.1	−4.42	5.53	5.57	4.96	5.02	5.55	5.66	5.74	5.86	3.33	3.36
TR1	Worst 5%	−26	−27.7	−16.5	−11.4	−15.4	−8.7	−15.1	−8.4	−15.2	−8.8	−20.1	−11.6
	Ratio	−0.16	−0.16	0.34	0.49	0.32	0.58	0.37	0.67	0.38	0.66	0.17	0.29
	Avg. Ret.	−2.7	−3.25	5.49	5.6	4.78	4.86	6.72	6.78	7.22	7.27	3.69	3.65
TR2	Worst 5%	−31.2	−26.4	−24.9	−14.2	−24.9	−14	−22.2	−12.4	−23.7	−12.6	−22.4	−13.6
	Ratio	−0.09	−0.12	0.22	0.39	0.19	0.35	0.3	0.55	0.3	0.58	0.16	0.27

For each strategy, the results show the average annual return over the 500 bootstraps (%), Avg. Ret., the worst 5% period return (%), Worst 5%, and the ratio of avg. return to worst 5% return, Ratio.

diverge from the ones of the backtests. The TREND signal no longer outperforms the BASE portfolio. Instead, the HVOL and IVOL signals emerge as the most effective, outperforming the BASE portfolio in risk-adjusted terms, with and without trading rules. The different findings from bootstraps vs. historical backtests indicate that the added value of these signals is dependent on the market regime. In fact, a closer look at the backtested results shows that the outperformance from TREND is mainly due to its superior performance during the 2008 GFC drawdown and the years immediately following (these detailed results can be provided upon request). No signal works for every time period and market regime and a more active management, for instance by conditioning the signals on the different market regimes, could potentially improve the results.

5. Discussion

The use of quantitative trading signals has been extensively researched by academics and widely implemented by equity asset managers, with the aim of outperforming a passive benchmark or building profitable long–short, market-neutral strategies. In this study, we examine whether a set of well-known backward-looking signals from equity markets and less explored forward-looking signals originating from options markets can add value to the construction of two classic option strategies that, when incorporated into two base equity portfolios, should improve their risk-adjusted performance. The first strategy, the covered call, aims to provide a yield-enhancing mechanism to the portfolio, using the signals to minimize the lost upside potential from rising underlyings that traditional covered call strategies suffer from. The second is a long-put strategy whose objective is to reduce capital loss in the event of adverse market movements but without compromising long-term performance. Therefore, the aim is to reduce as little as possible the impact on portfolio performance caused by the purchase of options through the use of signals, which should help identify the correct underlying assets on which to place option positions.

As a first step, the existence and pertinence of these signals in the Swiss market are analyzed by simulating two competing long-only portfolios and a long–short portfolio. The TREND signal and, to a lesser extent, the MOM signal are found to show statistically significant results.

When applying these same signals to the options’ strategies and incorporating them into the base equity portfolios, the TREND signal tends to be the most effective both in the covered call and long put strategies. SKEW also adds value to the covered call strategy. The results from the long-put strategy outline that no signal works in all periods and market regimes, even if the TREND signal here again helps to reduce the drawdown in the event of a severe and lasting market decline. Hence, testing a more active strategy that conditions the use of signals to market regimes could be worthwhile. Yet our results show that more complexity, in this case adding trading rules, is not necessarily synonymous with better performance. Finally, the findings highlight the importance of choosing the correct base equity allocation strategy, as GMV portfolios outperform all their EW counterparts.

These empirical findings must be viewed with certain limitations. The options dataset is only composed of Swiss large-capitalization underlyings. When testing the option strategies, the resulting portfolios tend to be very concentrated on a few assets. In practice, portfolios with small-capitalization companies might be more diversified. However, the lack of liquidity in the options markets for small-capitalization underlyings would render these strategies very difficult to be implemented. Even with large-capitalization companies, there could at times be large bid–ask spreads in certain options, which could incur high costs for the strategies. Indeed, bid–ask spreads or transaction costs have not been considered in this study. The impact of transaction costs should be limited, given that portfolios are rebalanced only twice a year.

Moreover, other parameters could be tested (setting different strikes, different maturities, or rebalancing frequencies). In our option strategies, the “positive side” of signals (e.g., positive trend) is not used. One could examine option strategies that consider all signals, negative as well as positive. In the same vein, strategies that combine options (e.g., collars, straddles, …) could be tested.

Finally, recent studies [25] suggest that advanced machine learning (ML) techniques can be used to identify patterns or tendencies in the behavior of individual stock prices. These techniques could be implemented as complements or even substitutes for our more basic trading signals, with the aim of building more effective option strategies. As with almost all economic sectors, the use of ML and artificial intelligence (AI) in finance is increasing rapidly, not only in the context of asset management but also in other fields, such as risk management [26] and credit risk analysis in banking [27].

In summary, this article contributes to the recent literature on quantitative asset management by exploring an alternative direction that institutional investors could take in order to surpass their current challenges, in a context where traditional diversification through equities and fixed-income assets has been shown to suffer from many weaknesses. For investors who need to adopt more innovative investment approaches, the results presented here offer some guidance for their decision-making process.