top of page
Leapday Quantitative Research Team

Market-neutral portfolio construction with an event-based signal PART 3


Summary


In this report we construct a market-neutral portfolio using the third and final entry point for Leapday’s REACTION signals which is generated before the open on the second day of trading subsequent to a company’s earnings announcement. We then combine this portfolio with the portfolios built in our previous two reports, which use the REACTION signals generated 60 minutes after the open and just before the close on the first day of trading subsequent to a company’s earnings announcement. The combined portfolio that uses all 3 entry points (t0 open + 60 min, t0 close, t1 open) produces superior performance with metrics that are stronger than any individual entry point on its own or any pair of entry points. This combined portfolio has an annualized return of 9.9% between 2016-2020, a Sharpe ratio of 2.22, a maximum drawdown of -3.2%, and a correlation to SPY of 0.07. All of the portfolio results are inclusive of conservative trading cost assumptions. These results are robust to stock liquidity.



Read as PDF







Introduction


In previous reports, we constructed event-driven portfolios using REACTION signals released 60 minutes after the open on the first day the market is open after a company’s earnings announcement (t0 open + 60min) and signals released just before the close on the first day the market is open after an announcement (t0 close). In this report, we will focus on the third and final set of REACTION signals, namely those released prior to the open on the second day the market is open after an announcement (t1 open). The motivation is to show the performance of the final entry point of signals in a realistic portfolio setting and to study the relationship between all three entry points to demonstrate how a portfolio that combines all signals performs.


We will proceed by providing an overview of the signals, establishing the portfolio assumptions, presenting the results of a portfolio built using the t1 open signal, presenting the results of a portfolio that trades all the signals, and discussing the findings.



Signals


Leapday’s REACTION signals are available for 3 entry points after an event day, which is the first day the market is open subsequent to a company’s earnings announcement. These 3 entry points are: (1) 60 minutes after the open on the event day (t0 open + 60min); (2) just before the market close on the event day (t0 close); (3) before the market open on the next trading day after the event day (t1 open). Our previous reports focused on the first two entry points. This study will focus on the t1 open signals. This signal is available for 4 holding periods: 3d, 5d, 10d and 15d. Like previous reports, we use the 3d hold signal in the portfolio construction below.


The REACTION signals range from -100 to +100 indicating the direction and magnitude of the signal. Like our previous two reports, we will use approximately the top one-third of signals ordered by signal magnitude. This approach provides a balance between using the strongest signals and having enough trading activity to utilize available capital. Table 1 below shows the number of signals these settings correspond to between 2016 and 2020. With this constraint we observe over 1,000 trades a quarter.




Portfolio Construction


As outlined in the previous section, we will use the strongest one-third of t1 open REACTION signals that are released before the open on the second day the market is open subsequent to a company’s earnings announcement to build a simulated portfolio. We will focus on the 3d hold signal to keep this report consistent with the prior studies. Below are all the assumptions used in our portfolio construction.


  • Enter a trade at the open on the first day the market is open subsequent to a company’s earnings announcement (t1 open) and exit the trade on the close 3 days later (t3 close).

  • Create a market-neutral portfolio by hedging trades using an opposite side trade in SPY with a beta-adjusted notional value that offsets the beta risk of the event trade.

  • Starting capital of $50M and assuming constant capital throughout the portfolio (i.e. - we do not reinvest profits).

  • Portfolio leverage used during periods of high activity.

  • Trade no more than 6 percent of the starting capital per position.

  • Trade no more than 2 percent of a company’s historical average trade value per position. (The historical average trade value is calculated using the trailing 20 day average traded value.)

  • Beyond the two constraints on trade size stated above, trade equal weight.

  • Round-trip trading costs are based on breakpoints for liquidity group as follows:

    • Low (<$3M historical average trade value): 20 bps

    • Mid ($3M-$25M historical average trade value): 10 bps

    • High (>$25M historical average trade value): 5 bps

    • Costs are allocated half on entry and half on exit

    • In the metrics presented below, we assume $0 in hedge costs (for the SPY hedge) and the hedge does not use any equity or margin



Results


Below we present metrics from a portfolio that trades the top one-third of the t1 open 3d signal and holds those trades for 3 days using the portfolio assumptions stated above.



The metrics associated with the chart above are as follows. We include metrics without costs (Table 2) and with costs (Table 3).



Below in Table 4 are additional metrics for the portfolio between 2016-2020 including maximum drawdown and correlation to the broader market.



It is worthwhile to note that the above metrics for annual return are based on constant capital in which trade size does not grow with the portfolio and instead is based on the starting capital. There is no reinvestment of profits. When we present annual portfolio growth metrics above, these results inherently understate the annual return because the return is calculated using cumulative portfolio value, while trade sizing is held constant (e.g. - the 2018 portfolio annual return is based on growth of portfolio since the end of 2017, not the starting capital in January 2016 that the trade size is based on). Below in Table 5 we present annual metrics based on the constant capital used.



One additional caveat to note is that the majority of events take place during a few weeks toward the middle of each quarter. Therefore a portfolio trading these signals will be in mostly cash during the beginning and end of each quarter. Table 6 below shows that the median day uses only 21% of the initial capital. Thus, annualized returns presented above are conservative for a portfolio setting since an investor would have significant capital to deploy into additional strategies during much of the year and return on capital used is significantly higher.



Finally, all the metrics presented above do not filter out any companies based on market capitalization or trade value. Given the starting portfolio size and the restrictions on the portfolio construction that cap how much of a company’s trade value can be traded, lower liquidity companies do not have a significant impact on the total return. However, to give additional reassurance of the value of the REACTION signals across high liquidity companies, we present metrics below in Tables 7 - 9 for a portfolio that only trades companies with an average daily trade value over $25M using all the same assumptions as before.



As shown above, constraining trades to only highly liquid companies results in a small deterioration in the overall performance of the portfolio, but metrics remain strong.



Combined Portfolios


This report built a portfolio with the t1 open signal. Intuitively, one may expect similarities between this signal and the t0 close signal from our previous report as they are only separated by the overnight session. And indeed, the performance looks similar at first glance. Let’s examine the correlation between the return streams of the two portfolios. Table 10 below shows this correlation.



A correlation of 0.77 suggests that the two return streams are fairly similar but leaves some room for incremental improvement by combining the portfolios. Below we present metrics from a portfolio that trades the top one-third of the t0 close 3d signal and holds those trades for 3 days and also trades the top one-third of the t1 open 3d signal and holds those trades for 3 days. Portfolio assumptions are the same for both sets of signals.



Metrics associated with the chart above are presented below in Tables 11 - 14.



As the tables above indicate, performance of the combined t0 close and t1 open portfolios is better than either individual portfolio. Between 2016-2020 and inclusive of trading costs, the t0 close portfolio returned 11.0% annualized, had a Sharpe ratio of 1.90, and max drawdown of -4.3%. The t1 open portfolio returned 9.8% annualized, had a Sharpe ratio of 1.85, and max drawdown of -3.1%. The combined portfolio exhibits better risk-adjusted metrics with an annualized return of 10.4%, a Sharpe ratio of 1.99, and a max drawdown of -3.5%. The result indicates that despite a fairly high correlation of returns between the t0 close and t1 open signals, there is incremental improvement from utilizing both entry points.


The natural next and final step is to combine signals from all 3 entry points (t0 + 60 min, t0 close, t1 open) into one portfolio and examine performance. Below we present metrics from a portfolio that trades the top one-third of the t0 open + 60 min 3d signal and holds those trades for 3 days, trades the top one-third of the t0 close 3d signal and holds those trades for 3 days, and trades the top one-third of the t1 open 3d signal and holds those trades for 3 days. The latter two use the portfolio assumptions stated in this report. The t0 open + 60 min uses almost identical assumptions and the primary distinction is that trading costs are doubled for the t0 open + 60 minute signal to account for using the midpoint of the spread at the time the signal is generated for the entry price.



Metrics associated with the chart above are presented below in Tables 15 - 18.



As the tables above indicate, performance of a portfolio that combines all 3 entry points is stronger than any individual portfolio and stronger than a portfolio that combines only 2 entry points. Between 2016-2020 and inclusive of trading costs, the combined portfolio using all 3 entry points has an annualized return of 9.9%, a Sharpe ratio of 2.22, and a max drawdown of -3.2%. These results highlight the value of using all three entry points from the Leapday REACTION signals.



Discussion


We have demonstrated that building a market-neutral portfolio that combines all three entry points of Leapday’s REACTION signals produces strong annual returns, high Sharpe ratios, and small drawdowns with results that are robust to trading costs and various market regimes. We have also shown that this combined portfolio results in better risk-adjusted metrics than a portfolio that only trades a single or two entry points, highlighting the power of using multiple entry points.


The metrics presented above are based on a simple standalone strategy and sophisticated investors can use the signals in a variety of ways to boost performance even further. The signals can be used as an input in a model or any decision making framework alongside other inputs to outperform the standalone strategy. Furthermore, investors can use more advanced portfolio construction methods. In this study, we treat all signals that fall in the top one-third of signals equally. In our whitepaper, however, we demonstrate that the top 5% of signals outperform the top 10% and so forth. This leaves room for further boosts in performance by weighting trades according to the magnitude of the signal. Finally, the entry and exit rules used in this portfolio are simple. The portfolio performance can be further improved by optimizing the hold to incorporate additional information gauged from the onset of the trade until the optimal exit.



Conclusion


In this report we build a market-neutral portfolio using signals generated after earnings. We focus on the REACTION t1 open 3d signal, which is generated before the open on the second day of trading subsequent to a company’s earnings announcement, and create a portfolio that uses the top one-third of these signals based on signal magnitude to enter trades at the t1 open and hold that trade for 3 days. Inclusive of costs, this portfolio has an annualized return of 9.8% between 2016-2020, a Sharpe ratio of 1.85, a maximum drawdown of -3.1%, and a correlation to SPY of 0.07.


In this report we also create a portfolio that combines the t1 open 3d signals introduced in this report and the t0 open + 60min signals and t0 close signals from our previous two reports. The result is a combined portfolio that inclusive of costs has an annualized return of 9.9% between 2016-2020, a Sharpe ratio of 2.22, a maximum drawdown of -3.2%, and a correlation to SPY of 0.07. These metrics are stronger than any individual entry point on its own or any pair of entry points and highlight the value of using all of the entry points from the Leapday REACTION signals.


The metrics presented in this report are conservative given the trade cost assumptions, portfolio position sizing constraints that we use, and the cyclical nature of earnings events which allows investors to deploy capital elsewhere during periods of low earnings activity. These results indicate that investors and traders can use the REACTION signal to drive an independent trading strategy or as an input for a model or decision making framework.



 


Disclaimer


This material is solely for informational purposes and is not an offer or solicitation for the purchase or sale of any security, nor is it to be construed as legal or tax advice. References to securities and strategies are for illustrative purposes only and do not constitute buy or sell recommendations. The information in this report should not be used as the basis for any investment decisions. We make no representation or warranty as to the accuracy or completeness of the information contained in this report, including third-party data sources. The views expressed are as of the publication date and subject to change at any time. Hypothetical performance has many significant limitations and no representation is being made that such performance is achievable in the future. Past performance is no guarantee of future performance.

Comments


bottom of page