Smart Trading M&A in the Semiconductor Industry

With the election now only two days away and most choices likely set, I’m sure that many of you are ready for at least a brief respite from the crescendo of related coverage. In that vein, we’re taking a break from our election series this week, but for those still inclined you may want to check out our previous two posts: “Mitigating Election-Driven Volatility in Your Portfolio” and “The Hidden Election Candidate – ESG”.

Since the start of 2019, the Semiconductor industry has been on a tear, outperforming the S&P500 by almost 70%.

More recently, during the pandemic, the industry has not only benefitted from the boom in work/play-from-home demand, but also from robust data center demand as cloud computing continues to expand rapidly. This rapid demand and need for scale has opened the door for the traditional tier-2 players to consolidate and challenge the historically dominant Intel:

1. On July 13, Analog Devices (ADI) announced it will acquire Maxim (MXIM) for $20B
2. On September 13, Nvidia (NVDA) announced it will acquire Arm Holdings (ARMH) for $40B
3. On Monday (Oct 26), AMD announced it will acquire Xilinx (XLNX) for $35B
4. On Thursday, (Oct 29), Marvell (MRVL) announced it will acquire Inphi (IPHI) for $10B

The Typical Merger Trade

This boom in M&A is a playground for a risk arbitrageur. Leading academic research suggests that richly valued companies typically signal a peak in their valuation, especially when they use their stock as currency to pay the acquisition price. Thus, a merger-arbitrage strategy for a stock-for-stock deal has the investor buying the target’s stock and short selling the acquirer’s to lock in the deal spread.

For example, using the Xilinx / AMD deal, the current % spread between Xilinx closing price as of Friday and the AMD deal price is computed as follows:

(AMD_Stock_Price * 1.7234 - XLNX_Stock_Price) / XLNX_Stock_Price = (75.29*1.7234 - 118.69)/118.69 = 9.3%

‍That’s a healthy return, especially if one believes that regulatory approval for the deal, while certainly time consuming, will likely be a go due to the highly competitive nature of the industry, the company’s complementary product sets, and the combined AMD’s market share (~25%) still far below Intel’s >60% share of CPU’s sold globally.

Conventional wisdom suggests to apply the traditional risk arb investment principals to this opportunity, but is there a better trade?

An Alternative for an AMD acquirer short

If you think the semiconductor industry has been on a tear, take a look at AMD’s stock price performance since the beginning of 2019:

Since 2019, AMD’s stock has outperformed the S&P500 by 272%. More recently, during the pandemic, AMD has not only benefitted from strength in its GPU businesses as laptop & gaming system sales soar due to work/play-from-home demand, but its cloud business has also benefitted from robust data center demand as cloud computing continues to expand rapidly. And lastly, market share leader Intel has struggled with manufacturing at the 7 nanometer node, which will result in a 1-year delay from its original plan, giving AMD a manufacturing advantage and chance to take more market share (AMD makes all of its chips at Taiwan Semiconductor, where 7nm chips have already been shipping for a year).

Consensus Wall Street estimates see roughly 18% upside for AMD over the next 12 months. With a continued tailwind, investors are potentially better off buying AMD instead of selling it, but then how will we capture the potential XLNX/AMD merger spread?

What risks should we hedge?

Interestingly, 72% of AMD’s stellar 272% S&P500 outperformance shown in the prior chart is attributable to factors such as Momentum, Growth, and the Semiconductor industry. Thus, our view on AMD’s total performance in the future should likely be decomposed into:

• Our view on AMD’s future idiosyncratic (alpha) performance
• Our view on future factor (growth, momentum, semiconductor industry) performance.

The chart below shows AMD’s historical and current predicted active risk (vs S&P500) of 45.54%, decomposed into factor and alpha components. This means that an investor shorting AMD is actually betting against:

• 73% of AMD’s future idiosyncratic (alpha) performance
• 27% on future factor (growth, momentum, semiconductor industry) performance, of which 8% is semiconductor industry related and the remaining 19% is style (growth, momentum, etc..)

With the election around the corner and high likelihood of significant market volatility, we certainly would want to hedge out the factor risk but likely minimize our hedge on the risks associated with the positive drivers of AMD and the semiconductor industry.

Building an AMD SmartHedge™

Leveraging Omega Point’s Smart Hedge builder, we were able to design a basket hedge that replicates AMD’s factor risk using the following rules:

Universe selection:

Our goal is to select highly liquid securities with low probability of tail risk. This yielded roughly 774 securities

Position sizing:

Our goal here is to size securities to minimize trade impact and further reduce tail risk by capping their size in the basket. This generated a basket of 103 securities.

‍Objective: build a basket that isolates AMD’s stock-specific (idiosyncratic) risk

‍As you can see, the basket is almost 100% factor-driven, and offers a far more diversified industry exposure than the pure semiconductor industry hedge provided by AMD.

Backtesting the hedge

While past performance is certainly not indicative of future results, backtesting allows us to determine whether our process appropriately captures the factor characteristics of AMD, while avoids the idiosyncratic elements.

The graph below compares AMD as a short hedge (LEFT), our AMD Smart Trade (MIDDLE), and an SMH (Semiconductor Industry ETF) hedge (RIGHT) from October 1st, 2019 through EOD 10/30/2020. Each hedge is rebalanced monthly.

We observe the following:

• Smart Hedge greatly reduces idiosyncratic performance (-31.01%) to (-3.92%) from taking a pure hedge on AMD.
• Relative to both AMD and the SMH, the Smart Hedge reduces drag from shorting the positive trending semiconductor industry.

The combination of both reducing idio + sector risk allows the Smart Hedge to outperform the AMD hedge by almost 40% and the SMH ETF hedge by almost 12% over the backtest period.

Conclusion

For an investor looking to gain a dual benefit of (a) positive semiconductor trend and (b) merger premium returns can apply the AMD SmartHedge™ approach across the three other semiconductor deals. If this is an area you’re interested in exploring further, please contact us and we’d be happy to setup a time discuss these types of trades.

That’s a wrap, we hope that you and your families had an enjoyable Halloween with many more treats than tricks, and your week goes superbly. On our end we’ll be absorbing the headlines and crunching the numbers during what is sure to be a momentous week, stay tuned for what we uncover in our next issue.

US & Global Market Summary

US Market: 10/26/20 -10/30/20

• The market suffered its worst week since March, driven by soaring COVID cases and lack of information around vaccines and government stimulus, as well as uncertainty around the US election and the aftermath of its outcome.
• Earnings season has largely indicated recovery in the more cyclical sectors, with many companies still facing challenges stemming from the pandemic. Thus far, 70% of the S&P 500 constituents have beaten consensus by one standard deviation or more (according to GS).
• After the worst quarter in history, US 3Q GDP bounced back to the tune of +33.1% on an annualized basis, besting consensus estimates of 32%. The big print was driven by growth in business and residential investment along with stronger consumer activity.
• This week, investors worldwide will be focused on the upcoming US elections, up and down the ballot.

Normalized Factor Returns: Axioma US Equity Risk Model (AXUS4-MH)

• Profitability continued its reversion and is now firmly in Neutral territory at -0.21 SD below the mean (recent bottom was -1.09 SD below the mean on 10/7).
• Both Growth and Value remained flat on the week, both trending slightly below their respective means.
• Earnings Yield exited Overbought space as it continued to descend from a peak of 1.53 SD above the mean on 10/8.
• Volatility fell into negative normalized territory and Market Sensitivity went deeper into negative space as investors eschewed riskier positions.
• Size was once again the biggest loser as it fell nearly 1/2 a standard deviation, now sitting at -0.71 SD below the mean.
• US Total Risk (using the Russell 3000 as proxy) finished the week up 16bps, after popping to as high as 23.7% on 10/28.

‍Normalized Factor Returns: Axioma Worldwide Equity Risk Model (AXWW4-MH)

• Value was the biggest riser in a quiet week for global factors, now sitting at +0.4 SD above the mean.
• Profitability was the other winner as it climbed back towards the mean (now at -0.31 SD below the mean).
• Unlike the US, global Volatility saw a slight decline and now sits perfectly at the mean.
• Earnings Yield saw continued weakness, as it saw a slight downward move that landed it in negative space. This factor was +1.53 SD above the mean on 10/7.
• Global Risk (using the ACWI as proxy) increased by 11bps this week.

Regards,
Omer