REVIEWING PORTFOLIO CONSTRUCTION-WITH A LITTLE HELP FROM MY FRIEND – GEMINI

 

REVIEWING PORTFOLIO CONSTRUCTION-WITH A LITTLE HELP FROM MY FRIEND – GEMINI

Portfolio Construction (PC) is a different skill from security selection. Security selection is the identification of a security and weighing up whether the risk/return makes sense. PC is more multi-faceted. There is the security selection question, then the sizing question and then the interplay of that security with the rest of the portfolio, like factor or theme exposure. Unconscious bets in the portfolio can be dangerous and usually surface after the damage is done. That is, where unidentified correlations exist, they can bite severely and unexpectedly. Then there is the completely separate question of appropriately sizing stocks, factors and themes to optimise returns; that is a complex task in itself.

Admittedly, I've always struggled with diversification for diversification's sake. If an asset is just there because it is different, that doesn’t make much sense to me. I want assets or stocks that I believe are attractive and will be worth much more in time. The challenge then becomes weighting those stocks and being aware of correlations and concentration risks. The flipside is that without taking risk, there is likely to be no excess returns, and it will just be reliant on luck. Therefore, there is a trade-off to be made, with the outcome being that the portfolio is concentrated enough to generate excess returns, but not overly concentrated to risk disaster, and to identify a selection of attractive stocks that are not dependent on the same inputs for their success and having the confidence and conviction to hold the strategy through inevitable tough times.

If anything needs an extra set of eyes, it is likely Port Con.

Background and Nuance

Since running my portfolio free from my employment compliance restraints, I have aimed for a 33X3% portfolio mix. That is 33 stocks equally weighted, as a guide. The stock numbers come from multiple studies indicating that 25-35 is the optimal stock number; of course, cross-correlation matters as well. My ultimate aim is to establish a diverse collection of quality franchises that are bought at reasonable prices and will grow earnings for a very long time.  In reality, that mix has waxed and waned due to stock performance, and I have also added progressively to a tail of interesting positions with less conviction. When a favoured stock has been cheap due to a SP decline, or better fundamentals, I have added; when expensive, reduced; and when a thesis is breached, sold. The result is no real portfolio structure, and one thing I do like is structure: why is something held and why at that weight. The other consideration is that as I get older, capital conservation slowly increases in importance compared to higher returns. Looking at the stock weights in my portfolio, I couldn’t see a logic behind the mix of position weights.

I have used Gemini and its sidekick NotebookLM for probably close to two years. The model has improved immeasurably over that time. The answers are now worth paying attention to. Besides that, I have uploaded an enormous amount of data into the system, the closed sandbox -  NotebookLM, and the broad LLM Gemini. Gemini knows my portfolio; it knows my investment philosophy, style and risk tolerances. That knowledge has been built up over many conversations. It also knows a multitude of questions I have asked about my portfolio holdings and other stocks.

The main aim of the prompt was to create a portfolio structure with a risk/return framework that I would find sensible to implement.

“Overall, I am after a low double-digit after-tax return on my total wealth.... I am more interested in stock weightings; how should the skew of stock weightings to generate the expected returns be comprised?....... You are a market analyst with exposure to the best portfolio construction practice and theory. How should the portfolio be comprised, keeping with quality growth, that is, relatively conservative stocks given the large equity exposure to my net wealth?”

The Reply and my responses are below.

To achieve a low-double-digit after-tax return on your total wealth when equities represent 67% of your net worth, your portfolio construction must resolve a critical mathematical constraint: your equity sleeve cannot afford dilution from equal-weighted mediocrity or sub-scale tail holdings. (suggests varying weights and eliminating the tail).

1. The Math of the Hurdle Rate: Total Wealth vs. Equity Sleeve

Your non-equity assets provide stability and steady income, but their expected long-term nominal returns are naturally bounded:

[Total Wealth: 100%]

   ├── Principal Residence (24%)  ──> ~4.5% p.a. (Tax-free capital growth) ──┐

   ├── DB Pension (6%)           ──> ~5.0% p.a. (Diversifying cash yield)   ├──> Generates ~1.49% to Total Wealth

   ├── Cash (3%)                 ──> ~3.5% p.a. (After-tax yield)          ──┘

   └── Equities Portfolio (67%)   ──> REQUIRED: 12.7% to 14.2% p.a. Net After-Tax & Fees

• The Return Equation: If your residence delivers ~4.5% long-term untaxed growth, your DB pension provides ~5.0% equivalent income, and cash earns ~3.5% after tax, these non-equity assets collectively contribute only ~1.49% to your total wealth return.
• The Equity Burden: To achieve a 10.0% net after-tax return on total wealth, your share portfolio must compound at 12.71% net after fees and taxes (~14%–15% pre-tax). If your target is 11.0%, the equity portfolio must deliver 14.20% net.
• Implication: Because 67% of your net worth is carrying the growth engine of your entire wealth, an overly conservative or overly dispersed equity portfolio will mathematically fall short of a double-digit total wealth outcome. (The maths is, of course, correct; the double-digit outcomes would be a great result given the lower risk I am willing to bear, IMO, so it is a stretch target that the LLM doesn’t know).

2. The Critique of the "30–40 Equal-Weighted (1/N)" Model

While equal weighting (e.g., 35 stocks @ ~2.8% each) provides psychological comfort, it introduces three structural drags for a quality growth investor:

1. Dilution of Exceptional Moats: When your highest-conviction monopolies (e.g., REA, CAR, Visa, TSMC, Constellation) compound at 20%+ ROIC, a flat 2.5%–3.0% position restricts their ability to drive the portfolio. A 100% gain on a 2.5% position adds just 2.5% to total portfolio return.
2. The "Zombie Tail" Cognitive Tax: In your current 36-line portfolio, 5 holdings are under 1.0% and 7 holdings are between 1.0% and 2.0%, totalling 10%. These 12 tail positions consume half your research and monitoring bandwidth, yet even if a 0.5% position triples, it moves your portfolio by only 1.0%.
3. Rebalancing & Tax Inefficiency: Strict equal weighting forces you to mechanically trim compounding winners (crystallising CGT) to top up lagging or lower-conviction names, directly undermining the asset value of your deferred tax liabilities. (Spreading the portfolio across numerous small holdings is a self-indulgence, and I have been aware that even if getting these positions correct, the impact on the portfolio would be muted. That makes sense to me, but it does raise other issues. Tax is a useful but relatively minor point.)

3. Recommended Portfolio Construction: The 3-Tier "Conviction Skew"

Rather than 30–40 equal weights, the optimal structure for a conservative retiree pursuing low double-digit after-tax compounding is 22 to 28 holdings organised into three distinct conviction tiers:

┌────────────────────────────────────────────────────────────────────────┐

│  TIER 1: Core Wide-Moat Monopolies / Tollbooths (6–8 Stocks | 35%–40%)   │

│  • Target Weight: 4.5% – 6.0% each  • Trim Ceiling: >7.5% – 8.0%      │

│  • Examples: REA, CAR, Visa, SPGI, Constellation (CSU), GOOGL/AMZN     │

├────────────────────────────────────────────────────────────────────────┤

│  TIER 2: Quality Growth Compounders / Franchise Leaders (10–14 Stocks) │

│  • Target Weight: 2.5% – 3.5% each  • Total Weight: 35%–40%           │

│  • Examples: ResMed, Aristocrat, Pinnacle, Macquarie, LVMH, TSMC, Meta │

├────────────────────────────────────────────────────────────────────────┤

│  TIER 3: Cyclical Troughs / Deep-Value Monopolies (4–6 Stocks | 10%–15%) │

│  • Target Weight: 1.5% – 2.5% each  • Re-rate / Promote to Tier 2     │

│  • Examples: Cochlear (at trough), Copart, Zoetis, Nick Scali          │

├────────────────────────────────────────────────────────────────────────┤

│  STRATEGIC CASH & LIQUIDITY BUFFER (3% – 5%)                           │

└────────────────────────────────────────────────────────────────────────┘

Key Rules of the Tiered Architecture

• The 2.0% Minimum Hurdle (Eliminate the Sub-Scale Tail): If a stock does not warrant at least a 2.0% allocation, it belongs on a research watchlist, not in your live portfolio. Exiting your sub-1.5% positions frees up ~10%–12% of capital to allocate into higher-conviction Tier 1 and Tier 2 anchors.
• Tier 1 (4.5%–6.0%): Reserved exclusively for businesses with pricing power, high ROIC (>20%), fortress balance sheets (net cash or very low debt), and minimal threat of technological disintermediation. Let these run up to 7.5%–8.0% before executing a trim.
• Tier 2 (2.5%–3.5%): High-quality businesses with slight cyclicality, capital expenditure requirements, or regulatory exposure that warrant standard position sizing.
• Tier 3 (1.5%–2.5%): High-asymmetry entry points—wide-moat businesses experiencing temporary operational or multiple-compression drawdowns (e.g., CSL or Cochlear at multi-year valuation troughs). As the thesis normalises, they either graduate to Tier 2 or get sold on recovery.

4. Tax-Aware Rebalancing Framework

To prevent (your) behavioural paralysis experienced in FY26 where overvalued stocks were held through large drawdowns due to tax fears:

Rebalancing Trigger	Action Protocol	Entity Routing
Position reaches >7.5%–8.0% of portfolio	Execute a mandatory 15%–25% partial trim back to a 5.0% baseline.	Prioritise selling highest-cost parcels or executing inside SMSF (15%/0% tax).
Valuation stretches >35% above 5-yr historical band	Execute a pre-committed 20% trim into cash or an undervalued Tier 3 peer.	Never exit 100% of a Tier 1 compounder purely on valuation; trim in tranches.
Thesis Breach / Structural Moat Impairment	Immediate 100% exit, regardless of tax consequence or capital loss.	Use realised capital losses to immediately offset gains from overvaluation trims.

PS the 35% valuation stretch equates to the SP showing about 0% valuation upside through my valuation process.

5. Conservatism Without Sacrificing Returns

Given that equities comprise 67% of your net wealth, true conservatism does not come from holding 40 diversified average businesses or low-yielding fixed interest.

True conservatism comes from balance sheet durability and earnings quality:

• Zero Financial Distress Risk: >80% of your equity capital should reside in companies with Net Cash balance sheets or Net Debt/EBITDA < 1.5x (e.g., REA, CAR, CSU, Visa, Copart, ResMed).
• High Cash Conversion: Favour companies where reported statutory profit converts at >90% into Free Cash Flow, avoiding capital-intensive or acquisition-dependent models.
• International Moat Arbitrage: Keep 40%–50% of your equity sleeve in global best-in-breed monopolies (Visa, S&P Global, Constellation, Big Tech) to eliminate the geographic concentration risk of Australia's narrow quality-growth universe.

CONCLUSION and SUMMARY—POSITIVES/NEGATIVES

Firstly, this framework is highly customised (to me), and I have directed the LLM with constraints that fit my temperament and style. That is, it is not transferable.

The suggested framework attempts to allocate risk and return with the goal of holding a conservative-quality growth stock portfolio without sacrificing returns.

The suggestions I like are the framework of holding a base of toll road/monopolies and trading a portion of these as valuations move around.

I like concentrating the fund to the three tiers with different risk/return and portfolio sizes. The portfolio holds together quite well, with some flexibility across the board and opportunism in the last tier.

The suggestions to spread stocks across geographies, fundamental risk constraints on debt limits and cash conversions, etc all speak to durability, conservatism, and quality.

The rebalancing triggers and reasonable tax realisations make sense as well. Balancing towards tax neutrality.

What is implied is that the stocks can be entered at reasonable prices and that the stock picking is of a good standard. Implicit in the framework is that I have conviction and patience to hold through any style reversals and let the businesses compound earnings, while the market may not be interested in these stocks for extended periods. Holding the portfolio as a private undertaking and not being under the influence of building a business and all the outside and inside influences that impact professional managers is a benefit of the retail investor.  Unlikely I could do this as a professional manager.

In summary, I like the framework and will move to implement it over the short to medium term. Certainly, it will reduce the research and monitoring efforts to only look at those stocks that are appropriate for the various tiers. Good stock picking does not disappear; it is always there. Conviction and patience, perhaps the most difficult aspect, do not disappear either.