TMT Model Portfolio Update: Tempering Volatility
SEE LAST PAGE OF THIS REPORT Paul Sagawa / Tejas Raut Dessai
FOR IMPORTANT DISCLOSURES 203.901.1633 /.553.9827
psagawa@ / firstname.lastname@example.org
December 3, 2017
TMT Model Portfolio Update: Tempering Volatility
The sharp tech downturn on Nov 29 wiped away 4 months of outperformance for our model portfolio, which dropped 392 bp vs. the S&P500 and 106 bp vs. the tech components of that index in a single trading session. Since July 20, the portfolio is up 350bp vs. the broader market, but is now down 70bp against the tech benchmark. While the long-term performance of our 15-stock portfolio has been excellent, the high correlation between many constituents leaves it vulnerable to occasional painful adjustments like this. Enterprise SaaS software currently makes up a third of our portfolio, and, while we are enthusiastic for all the names, the correlation in their performance to each other and to the portfolio suggests that their impact is redundant. With MSFT and CRM as core elements, and with WDAY and NOW less correlated, lower in market cap, and thus, more likely to be acquired, we have decided to remove ADBE, which trades at P/S premium to WDAY despite materially lower growth. In its place, we are adding CIEN, which has launched an intriguing chip platform for optical SDN functionality and should benefit from aggressive 5G spending from global carriers and an upgrade cycle to 100/400Gbps systems.
- Our model portfolio was badly hit by last week’s tech sell off. At market close on Nov. 28, our 15-stock model portfolio had outperformed the S&P500 by 743 bp and the tech components of that index by 34 bp since the last update on July 20. Subsequently, the sharp downturn in tech stocks erased much of our upside vs. the overall market, and brought returns below the tech benchmark. Leading the drop were the SaaS software and AI-lined chip names that had extraordinary performers prior to last week.
- Historical performance has been excellent but volatile. The purpose of our portfolio is to identify stocks likely to be beneficiaries of paradigmatic change in the TMT space, with a distant investment horizon in mind. Our overarching research thesis points to a “Cloud/AI Era”, driven by platform leaders – GOOGL, AMZN and MSFT – and benefiting forward-thinking companies in digital, SaaS, new semis, and 5G. This approach has led to generally strong performance – the portfolio is up 3910 bp vs. the S&P500 since inception 6 years ago – but has been prone to periods of extreme volatility. We tie this to overconcentration in specific themes and to high correlation amongst our stock selections.
- The overweight on SaaS software is likely excessive. With considerable overlap from companies with efforts in multiple themes, we have 4 cloud/AI platforms, 4 digital media, 5 SaaS software, 3 AI/cloud components, 2 5G tech, 2 e-commerce, 1 autonomous transportation and 1 disruptive carrier. While we remain enthusiastic for the potential of our SaaS names (MSFT, CRM, ADBE, WDAY and NOW), having a third of the portfolio with the same primary growth driver gives the group too much emphasis, contributing to the wrenching moves. With core holdings (GOOGL, AMZN and MSFT) participating across multiple themes, we are comfortable with our other weightings, although we seek new, high-quality names in 5G, e-commerce and self-driving.
- Our portfolio is high Beta but also highly correlated. The overall beta of our portfolio is 1.30, reflecting a substantial volatility relative to market movements. This is exacerbated by a high degree of correlation between the normalized daily performance of the stocks in the portfolio. Of 15 stocks, 10 have had better than 0.40 correlation between their normalized performance and the performance of the entire portfolio. This is not surprising, given the skew toward highly valued growth stocks, but greater diversity would likely temper some of the volatility. We evaluated several possible new constituents that participate in our growth themes, assessing them on both upside promise but also their potential to counterbalance other elements of the portfolio
- ADBE’s correlation makes it redundant. Amongst our SaaS stocks, we see MSFT as a strategic core holding. Of the others, there is a temptation to replace WDAY after the tech sell-off amplified the reaction to WDAY’s poorly timed 3Q17 earnings release, taking the stock down 13% in 3 days. However, we believe this reaction is overdone, with strong growth and guidance overriding nits about billing details and the downturn leaving it at lower multiples than its peers. Instead, we are removing ADBE, despite it having appreciated 124% since we added it 2 years ago. While we still believe that it can post upside surprises against expectations, it sports lofty valuation metrics and it is highly correlated with the portfolio and MSFT, making it the best candidate for replacement.
- Self-driving, 5G, AI hardware. Our portfolio addresses near-term opportunities created by the longer-term paradigmatic shifts of the Cloud/AI era. SaaS is one of these, but there are several others that are worth stronger representation in the portfolio. For self-driving, we considered automotive component suppliers – IFNNY, RNECY, TXN, etc. – with upside from a global transition to a shift to electric vehicles and from automated driver assistance technologies. While we are well represented in hyperscale/AI components with NVDA, XLNX and QCOM, we considered adding a semiconductor design tool name – e.g. SNPS, CDNS, etc. At the end, we chose to increase our exposure to 5G, believing that global carrier spending in 2018 will well above current expectations.
- Replacing ADBE with CIEN. CIEN missed the tech rally over the past few years, repeatedly cycling between $16 and $24, and currently sitting toward the upper end of that range. This relatively weak performance tracked to weak results as the optical equipment market troughed. However, we see three positive factors that should lift the stock well above this level. First, there are clear signs that carrier capex for 5G will be more aggressive than earlier assumptions (http://www.ssrllc.com/publication/5g-rising-global-carrier-competition-to-drive-capex/) and that optical systems for backhaul will be a key element. Second, the introduction of 400Gbps speeds appears likely to spur upgrades for long-haul optical systems. Finally, CIEN has introduced an ASIC DSP that enables much more flexible provisioning of bandwidth, mirroring the move to SDN architectures for data networking. This solution could become a standard for hyperscale operators, particularly as CIEN has licensed the chip to systems level rivals.
Rebalancing the Portfolio
The big sell-off of high-value, high-beta tech stocks hurt our model portfolio far more than it did the broader market and our benchmark, the tech components of the S&P500. We had ended the day Tuesday having outperformed that benchmark by 34 bp, and the whole index by 743 bp, since the last update in July. In 3 trading days, we dropped 492 bp vs. the S&P and 182 bp vs. the tech components. Over the 6 years that we have published the portfolio, its performance has been excellent but marked by volatile responses to periodic shifts in sentiment.
This volatility is, in part, an inevitable condition of a concentrated portfolio of 15 large cap TMT stocks chosen for their exposure to the major growth themes arising from our research. Still, given that few of our clients would be comfortable with such risk, we examined our selections to see if we could reduce the volatility a bit without sacrificing our self-imposed mandate to choose names that illustrate our thesis of generational paradigm change to a Cloud/AI era of technology.
To this end, we considered two related issues. First, does the portfolio, as constructed, give reasonable weight to the major investment themes that we have championed? With considerable overlap from companies with efforts in multiple themes, we have 4 cloud/AI platforms, 4 digital media, 5 SaaS software, 3 AI/cloud components, 2 5G tech, 2 e-commerce, 1 autonomous transportation and 1 disruptive carrier. In this context, we are a bit too heavily weighted toward SaaS software and short on suggestions for 5G, ecommerce, self-driving and e-commerce.
Second, are the stocks of our portfolio overly correlated with one another? Normalized for performance against the tech benchmark, our stocks average 24% correlation – perhaps we can do better. 11 of our names had positive correlations to the overall model performance over the past two years. Given the generally strong performance of the portfolio over that time, that is not necessarily a bad thing. Still, the correlations show how much the volatility of highly correlated stocks like NVDA and ADBE can swing the overall portfolio.
Given this, we have decided to remove ADBE, reducing our weighting on SaaS software and addressing correlation concerns. We stress that we still believe ADBE is likely to outperform relative to expectations, but its valuation is relatively rich and its $81B market cap makes acquisition at a significant premium unlikely vs. other SaaS names. In its place, we are adding CIEN. Optical stocks have been mired in a global downcycle, but 5G and an upgrade cycle to 100/400Gbps systems is beginning to yield fruit and some welcome investor optimism. CIEN has also introduced an ASIC DSP that extends SDN-like flexibility to the optical domain. Radically, CIEN will supply this chip to its systems rivals in a bid to establish it as a standard. We believe this architecture will be very well received by hyperscale platform operators, giving it a third impetus for growth.
Expectations for CIEN are modest, as is its valuation – at a forward P/E of less than 11x it is the cheapest stock in our portfolio. Moreover, the correlation of its relative performance to the rest of the portfolio is negative 0.55, perhaps a partial antidote to our overall volatility.
Another November Tech Swoon
In the 4 months since we last updated our 15-stock large cap model portfolio, the equally weighted shares are up 350 bp vs. the S&P500, but down 70 bp against the tech components of the broader index. This performance is sharply skewed by the last three trading sessions, which saw our portfolio down 431 bp. Had we published the update at the close on Tuesday, November 28, we would have been up 743 bp against the S&P and 34 bp against the benchmark. Given the risky nature of our portfolio selections, which are chosen as emblematic of our thesis of a generational sea change to a Cloud/AI era of computing, sharp changes – both up and down – have been common across a long-term record of excellent relative performance (Exhibit 1, 2 & 3).
Exh 1: SSR TMT Large Cap Model Portfolio
Within the portfolio, QCOM (23.7%), ADBE (19.7%), CRM (14.7%), MSFT (14.7%) and AMZN (13.0%) all delivered double digit gains, while KEYS (4.0%), GOOGL (3.3%), NFLX (1.8%), TMUS (-0.4%) and WDAY (-2.5%) were the relative laggards. QCOM’s upside performance was driven by the hostile interest taken by AVGO, while TMUS’s weak showing included a negative reaction to the failure of discussions with S to yield a merger. NFLX suffered a bit from DIS’s decision to withdraw its programing from the streaming platform to jumpstart its own similar service in 2019. We are not concerned, as we believe NFLX strength stems from its strong strategy with original programming building on a considerable first mover advantage. Similarly, we remain confident that KEYS will appreciate once 5G spending gains steam in 2018, and that GOOGL will make strides in both IaaS cloud hosting and autonomous transportation even as its advertising franchises continue to take global share.
Exh 2: Relative Performance of SSR’s TMT Model Portfolio versus Benchmarks
The biggest hit from the late November drop was on WDAY – which compounded things by releasing good, but perhaps not quite good enough, results and guidance on Wednesday and fell 13% through Friday. We expect that the relative fall for WDAY would not have been nearly so precipitous without the backdrop of the overall market action. High flyers NVDA (-7%), NFLX (-7%) and XLNX (-6%) suffered the next most serious wounds from the drop.
Exh 3: Relative Performance of Model Portfolio versus Benchmarks since Inception
Can We Fix Our Volatility?
Given the purpose for our portfolio – a concentrated list of high conviction names representative of the investment themes generated by our thematic research – volatility is an occupational hazard. Our selections are skewed to high-beta, high-multiple stocks positioned to benefit from the emerging “Cloud/AI Era” that we believe is replacing the device-focused “PC Era” that preceded it (Exhibit 4). The portfolio comprises 15 names, each with equal weight (rebalanced at each update). We think of Alphabet, Amazon and Microsoft as our “core” holdings, although our construction does not give them greater emphasis than the other stocks. While we are enthusiastic for all the other stocks in the portfolio, perhaps a change or two could provide better stability without sacrificing long-term potential returns.
We see two primary causes for our volatility problem. First, the selections are not well balanced across our investment themes. We have 5 stocks, a third of the portfolio, whose primary business is enterprise SaaS software applications. This is too much. In contrast, we have three chip vendors focused on AI/Cloud/5G opportunities, three digital media/advertising leaders, then one aggressive wireless carrier
Exh 4: SSR TMT Model Portfolio Themes
(TMUS), one traditional IT company looking to evolve into a Cloud/AI butterfly (IBM), one 5G testing company (KEYS) and an e-commerce juggernaut (you know who). Ideally, we would replace one of our SaaS names with a company well positioned for opportunities in an area where we are a bit light – e.g. e-commerce, self-driving, AI assistants, or 5G.
The second factor in our volatility is the correlation between the relative performance to the market of the various names in the portfolio (Exhibit 5, 6, 7 & 8). Ideally, all would appreciate vs. the benchmark, but if one or more stocks were to slip, other stocks might pick up the slack. If all the stocks are correlated, any rise or fall is amplified. Our portfolio constituents average a 0.24 correlation to the group. Looking more deeply, 10 of the stocks are positively correlated, while five are negatively correlated. NVDA’s extraordinary performance gives it outsized influence over the total returns despite equal weighting, thus the near perfect correlation to portfolio returns. Still, if a highly correlated name were swapped for a negatively correlated one, without sacrificing opportunity, performance might be less volatile.
Exh 5: TMT Model Portfolio Constituents Correlation Matrix – 2 Years
Exh 6: TMT Model Portfolio Returns Relative to S&P 500 Tech – 2 Years
Exh 7: TMT Model Portfolio Constituents Correlation Matrix – YTD