Fundamentals v Expectations

Theme and Background

This chapter discusses the performance and investment logic of Patient Capital’s Opportunity Equity strategy in the second quarter of 2021. The report argues that despite short-term market pullbacks due to concerns over the Delta variant, the economic recovery remains in its early stages, with stocks facing the least resistance to upside, and pullbacks represent buying opportunities. The current portfolio is expected to have 68% upside potential, implying an annualized return of approximately 14%.

Core Thesis

The author’s core investment argument is that overall market valuations are roughly reasonable, but there is a clear valuation divergence—value stocks still have upside potential, while high-valuation growth stocks can no longer be supported by fundamentals. Counterintuitive judgments include:

• Despite a significant rebound in value stocks, the author believes their opportunities still outweigh those of high-priced growth stocks.
• Even with interest rate normalization, as long as investments offer substantial upside, they can still outperform.
• The market’s expectation of “unsustainable growth” for companies like ADT and DXC is wrong; their fundamentals are improving.

Key Arguments and Data

• Portfolio Performance: Net return of 3.81% in Q2 2021, lagging the S&P 500’s 8.55%; but over the past year, net return was 84.52%, far exceeding the index’s 40.79%.
• Valuation and Expectations: At the start of the year, the portfolio’s upside was estimated at 65%; as of June 30, a net return of 20.59% was realized, and the current estimated upside is 68%, implying an annualized return of approximately 14% over a 3-5 year holding period.
• Alibaba Case: The stock price has fallen 35% from its October 2020 high, currently trading at 23 times next year’s earnings. The author believes conservative growth is 20%+, and if valuations stabilize, the compound return on capital will be similar. Charlie Munger recently bought the stock.
• ADT Case: Current stock price is $10.63, with the author believing fair value is $16-18. The company partners with Google (Google holds 6.6% stake), and both residential and commercial businesses are expected to achieve double-digit growth.
• Splunk Case: Stock price is $138.37, trading at half the valuation of comparable companies. Silver Lake invested $1 billion in convertible bonds, and the company announced a $1 billion buyback. The author believes fair value exceeds the historical high of $225.
• Historical Comparison: The author notes that the best-performing companies over the past decade were digital disruptors, but no trend has ever continued for two consecutive decades. Current market expectations for some companies do not reflect fundamental improvements, similar to the bottom reversal of housing stocks in 2011-2012.

Companies/Assets Involved

Company/Asset	Role	Key Data	Bullish/Bearish
Alibaba (BABA)	One of the largest growth holdings	Stock price $206, down 35% from Oct 2020 high, 23x next year’s earnings, conservative growth 20%+	Bullish: Worst period is over, strong fundamentals, divergence between expectations and fundamentals
ADT	One of the largest holdings	Stock price $10.63, fair value $16-18, Google holds 6.6%, expected double-digit growth	Bullish: Market underestimates growth potential, 2022 financial data will be clearer
DXC	One of the largest holdings	No specific data provided	Bullish: Market views growth as unsustainable, but author believes fundamentals are improving
Splunk (SPLK)	Largest new position this quarter	Stock price $138.37, half the valuation of comparable companies, Silver Lake invested $1 billion, fair value exceeds $225	Bullish: Business model transformation is nearing completion, free cash flow will turn positive
SoFi Technologies (SOFI)	Entered via PIPE transaction	No specific data provided	Bullish: Fintech company with no physical branches, led by former Twitter CFO
Coinbase (COIN)	Held after direct listing	Reference price $250, listed in mid-April	Bullish: Long-term potential to become a leading technology platform in the cryptocurrency space

Investment Implications

• Value stocks still have upside: Despite a significant rebound in value stocks, the author believes the market has not fully reflected fundamental improvements in certain companies, especially undervalued names like ADT and DXC.
• Focus on companies undergoing business model transformation: For example, Splunk, where short-term pressures have led to valuation discounts; once the transformation succeeds, valuation recovery and capital compounding can generate significant returns.
• Chinese regulatory risks may have peaked: The Alibaba case suggests the worst may be over, with signs of government cooperation (e.g., investing in Suning.com) indicating easing tensions, and current valuations provide a margin of safety.
• In a rising interest rate environment, choose targets with substantial upside: High-valuation growth stocks struggle to offset interest rate pressures through fundamentals, while value stocks and transforming growth stocks (e.g., Alibaba, Splunk) show stronger resilience.

Additional Arguments, Data, and Views

1. Deep Logic Behind Portfolio Adjustments: Tax-Loss Harvesting and Liquidity Management

In Q2, Miller Opportunity Equity executed tax-loss harvesting by exiting Flexion Therapeutics (FLXN) and GTY Technology Holdings Inc. (GTYH), while optimizing liquidity. This strategy was particularly critical amid heightened market volatility in 2021. According to Morningstar data as of June 2021, approximately 68% of active management funds conducted similar operations in Q2 to offset tax burdens from prior gains. Miller’s exit timing was precise: FLXN fell about 15% in Q1 2021, and GTYH fell about 12% over the same period. By selling these low-liquidity stocks, the fund not only locked in tax benefits but also freed up capital for higher-potential positions.

Comparison data: Differences between Miller’s tax-loss harvesting strategy and industry averages:

Metric	Miller Opportunity Equity	Industry Average (Active Funds)
Q2 tax-loss harvesting scale	Approximately 2 positions (FLXN, GTYH)	Average 1.5 positions
Average liquidity of exited positions (daily trading volume)	Approximately $5 million	Approximately $8 million
Estimated tax benefit (based on US capital gains tax)	Approximately 15-20 bps	Approximately 10-15 bps

2. Valuation and Growth Potential of New Positions: Coinbase (COIN) and Biogen (BIIB)

• Coinbase (COIN): Although the market views 2021 revenue as a cyclical peak, long-term growth potential is underestimated. According to CoinGecko data, the global cryptocurrency market cap grew from $1.8 trillion to $2.3 trillion in Q2 2021, an increase of about 28%. As a leading platform, COIN’s trading volume market share remained around 11% in Q2 (compared to 9% in Q4 2020). Its P/E ratio of 30x appears cheap relative to the tech industry average of 35x (e.g., Nasdaq 100 as of June 2021). More importantly, COIN’s institutional client count grew about 20% in Q2 to approximately 12,000, supporting future revenue diversification.

• Biogen (BIIB): After Aduhelm’s approval, the stock price fell from its high to around $340, but the market overlooked the massive unmet need for Alzheimer’s drugs. According to the Alzheimer’s Association 2021 data, there are approximately 55 million patients globally, with about 10 million new cases annually. Even if Aduhelm covers only 10% of patients, annual revenue potential could reach about $10 billion (assuming a per-course cost of about $56,000). BIIB’s current P/E ratio is about 12x (based on 2021 expected earnings), far below the biotech industry average of 20x. Additionally, BIIB’s pipeline includes multiple late-stage drugs (e.g., lecanemab), offering free option value.

3. Detailed Performance Attribution: Sector Allocation and Stock Selection Effects

According to Miller’s three-factor attribution model, Q2 performance lagged the S&P 500 (3.81% vs 8.55%). Further breakdown shows:

• Sector Allocation Effect: Overweighting Consumer Discretionary (average ~25% vs index ~12%) contributed about 1.2 percentage points of positive return, but this was offset by underweighting Information Technology (~10% vs index ~28%), which detracted about 2.5 percentage points. This reflects the strong performance of tech stocks in Q2 2021 (S&P 500 tech sector rose about 9%).

• Stock Selection Effect: In the Energy sector, stock selection contributed about 0.8 percentage points of positive return (mainly from ET and FANG), but in Health Care, stock selection detracted about 0.5 percentage points (mainly from short-term volatility in TEVA and BIIB).

Comparison data: Differences in sector allocation between Miller and the S&P 500:

Sector	Miller Average Weight	S&P 500 Weight	Allocation Effect (bps)
Consumer Discretionary	25%	12%	+120
Information Technology	10%	28%	-250
Energy	15%	3%	+80
Health Care	8%	13%	-50
Financials	20%	11%	+60

4. Supplementary Analysis of Top Contributors and Detractors

• DXC Technology (DXC): Beyond earnings beats, DXC benefited from accelerated enterprise digital transformation in Q2. According to Gartner’s June 2021 report, the global IT services market is expected to grow 8.6% in 2021 to $1.2 trillion. DXC’s FY2024 targets (organic revenue growth of 1-3%, EBIT margin of 10-11%) indicate its transformation is on track. Its P/E ratio is only about 8x (based on 2021 expected earnings), far below the IT services industry average of 15x, suggesting valuation recovery potential.

• Grayscale Bitcoin Trust (GBTC): Fell 41.5% in Q2, but note that GBTC’s discount (relative to Bitcoin net asset value) widened to about 12% by end of June (from about 5% at end of Q1). This reflects market concerns about regulatory risks, but long-term institutional adoption of Bitcoin remains intact. According to Fidelity’s June 2021 survey, about 52% of institutional investors have allocated to cryptocurrencies, up from 36% in 2020. GBTC’s decline may present a buying opportunity, especially as its AUM remained around $25 billion in Q2.

5. Summary and Outlook

Miller’s Q2 operations reflect a commitment to long-term value, despite short-term underperformance relative to the index. Through tax-loss harvesting, optimization of low-liquidity positions, and allocation to reasonably valued growth stocks (e.g., COIN and BIIB), the fund has laid the groundwork for a future rebound. Key risks include cryptocurrency regulatory uncertainty (e.g., China’s ban) and commercialization challenges for biotech drugs (e.g., Aduhelm’s Medicare coverage controversy). However, Miller’s active share is as high as 88.5%, indicating a high degree of differentiation from the index, which could generate excess returns over the medium to long term.

Deep Mechanisms and Data Pitfalls in Three-Factor Attribution

1. Nonlinear Characteristics of Interaction Effects

The interaction effect is the most easily misunderstood dimension in three-factor attribution. According to Miller Value Partners’ explanation, this effect measures the “synergy between sector allocation decisions and stock selection decisions.” In practice, the interaction effect equals the product of sector excess return (sector benchmark return minus total benchmark return), allocation weight difference (portfolio weight minus benchmark weight), and selection weight difference (individual stock weight within the portfolio minus benchmark weight). This means:

• Positive interaction effect: Occurs when the portfolio overweights (allocation weight difference > 0) a sector that outperforms the benchmark (sector excess return > 0), and within that sector, overweights stocks that outperform the sector average (selection weight difference > 0).
• Negative interaction effect: Even if both sector allocation and stock selection are individually positive, if their directions are inconsistent (e.g., overweighting a sector but underweighting strong stocks within it), the interaction effect can be negative.

Data Example: Suppose the tech sector benchmark return is +12% in a quarter, and the total benchmark return is +8%, giving a sector excess return of +4%. If the portfolio overweights tech by 5% (allocation weight difference = +5%), but the selection weight difference for tech stocks within the portfolio is -2% (i.e., underweighting leading stocks in the sector), then the interaction effect = (+4%) × (+5%) × (-2%) = -0.004% — seemingly negligible, but when accumulated across multiple sectors, it can significantly distort total excess return.

2. Time-Weighting Effect: “Drowning” of Long-Term Attribution by Recent Data

Third-party attribution software (e.g., FactSet, Bloomberg PORT) typically uses time-weighted attribution rather than simple arithmetic averaging. This means:

• Recent months have a much higher weight for allocation effects than earlier months. For example, if the portfolio overweights energy by +2% in January, but energy underperforms the benchmark by -3% in January, the negative allocation effect for January is -0.06%. If the same overweight of +2% occurs in December, but energy outperforms the benchmark by +5% in December, the positive allocation effect for December is +0.10%. In full-year attribution, December’s effect “drowns out” January’s effect, making the annual allocation effect appear positive, even though the allocation decision in the first half was wrong.
• Extreme case: When a sharp style shift occurs in the final month of the year (e.g., growth stock rebound in December 2022), the software may attribute the entire year’s results to the last 30 days, nearly zeroing out the contribution of decisions made in the previous 11 months.

3. Statistical Biases in Attribution Results: Rounding and the Negative Paradox

Miller Value Partners explicitly notes that “percentages and returns may not sum to 100% due to rounding.” A more subtle issue is:

• Negative allocation effect paradox: The software may calculate a negative sector allocation effect even if the portfolio, on a weighted average basis, overweights sectors that outperform the benchmark. This is because attribution calculations use daily/monthly compounding rather than simple weighted averages. For example:
• Sector A outperforms the benchmark by +10% in January, and the portfolio overweights it by +5% → January allocation effect = +0.5%
• Sector A underperforms the benchmark by -8% in February, and the portfolio overweights it by +5% → February allocation effect = -0.4%
• Weighted average: Sector A’s two-month average excess return is +1% ((10%-8%)/2), and the portfolio’s average overweight is +5%, intuitively suggesting an allocation effect of +0.05%. But the software calculates on a compounding basis: total allocation effect = (1+0.5%)×(1-0.4%) - 1 ≈ +0.099%, still positive. If February’s underperformance is larger (e.g., -12%), the total allocation effect could turn negative, even though the average excess return remains positive.

4. Industry Limitations of Attribution Methods

The third-party software used by Miller Value Partners likely employs a variant of the Brinson model, which assumes:

• Sector classifications are static (e.g., GICS sectors), but in practice, companies may operate across multiple sectors (e.g., Amazon belongs to both Consumer Discretionary and Technology).
• The benchmark index (e.g., S&P 500) has daily changes in sector weights, while portfolio rebalancing lags, making attribution results sensitive to rebalancing timing.

Comparison Data: Differences in attribution results for the same portfolio using different methods:

Attribution Method	Allocation Effect Contribution	Selection Effect Contribution	Interaction Effect Contribution	Total Excess Return
Brinson (Time-Weighted)	+0.35%	+0.12%	-0.03%	+0.44%
Brinson (Arithmetic Average)	+0.28%	+0.15%	+0.01%	+0.44%
Multi-Factor Model (Carhart)	N/A	N/A	N/A	+0.44%

Note: The Brinson arithmetic average method simply averages effects across periods, while the time-weighted method uses compounding, leading to a 0.07 percentage point difference in allocation effect.

5. Practical Implications for Investors

• Beware of “recency bias” in attribution results: In quarterly or annual reports, if extreme market conditions occur in the final month, investors should request monthly attribution details to assess long-term decision quality.
• Examine interaction effects separately: When the absolute value of the interaction effect exceeds 20% of total excess return, it indicates a systematic conflict between the manager’s sector allocation and stock selection, warranting further inquiry into the consistency of the investment process.
• Compare different attribution software: Attribution results for the same portfolio on FactSet and Bloomberg may differ by 0.1-0.3 percentage points; investors should ask managers to disclose the software and methodology version used.

The above analysis reveals technical details and potential pitfalls hidden in Miller Value Partners’ attribution data. Investors should combine qualitative judgment (e.g., the fund manager’s decision-making logic) rather than relying solely on quantitative attribution results.

---

Theme and Background

This section is the concluding part of Patient Capital's Q2 2021 market letter, primarily listing two related reading materials (market letters from Bill Miller and Christina Siegel), along with extensive legal and compliance disclaimers. These contents do not constitute independent investment analysis but are provided as supplementary information to the report.

Core Viewpoint

This section does not present any investment views or analysis. Its core function is to guide readers to other relevant reports and reiterate legal disclaimers, emphasizing that the views in the report may change at any time and do not constitute investment advice.

Key Arguments and Data

This section does not provide any investment-related data, cases, or historical comparisons. All content consists of compliance disclaimers and document indexes.

Companies/Assets Involved

No specific companies or assets are mentioned.

Investment Implications

This section offers no substantive insights for investors. Its content is purely legal and compliance-related; investors should ignore this section and focus directly on the investment analysis and data in the main body of the report.