Should a portfolio hold more stocks so there’s less money riding on each one? It sounds obvious — and it’s half right. The CFP curriculum splits risk into two kinds. Unsystematic risk — one company blowing up — shrinks every time you add a position. Systematic risk — the whole market falling — doesn’t care how many tickers you own. It is, in the curriculum’s own words, nondiversifiable. Position #38 through #54 will not save you in a bear market; your allocation to cash, gold, and low-beta sectors does that.
So the honest question isn’t “does more diversification help?” It’s “where does the benefit stop — and what does it cost to keep going?” For fifty years, the textbook answer was: it stops around thirty stocks. That answer turns out to be stale.
The rule traces to studies by Evans & Archer (1968) and Meir Statman (1987): plot portfolio volatility against the number of holdings and the curve flattens fast — most of the achievable risk reduction arrives by 20 holdings, nearly all of it by 30–40. Those studies were built when every trade carried a fixed commission. Diversification had a real price, so you stopped buying it early.
Three things changed. First, Statman himself revised the number. In 2004 he reran the math against modern, near-zero trading costs and concluded the optimal level of diversification exceeds 300 stocks — the marginal benefit of one more name never falls below its now-negligible marginal cost. Second, the market got noisier at the single-stock level. Campbell, Lettau, Malkiel and Xu, in the Journal of Finance (2001), showed that individual-stock volatility roughly doubled between the 1960s and 1990s while correlations between stocks fell — meaning it now takes roughly 50 stocks to deliver the diversification 20 used to provide. Third came the discovery that changed how we think about the whole question.
Read that again, because it inverts the usual fear. Stock returns are violently skewed: a tiny handful of enormous winners pay for everything, and most stocks individually disappoint. The biggest risk a concentrated portfolio runs is not holding a loser — a 2% position that goes to zero costs you two cents on the dollar. The biggest risk is missing the winners. A small portfolio is a repeated bet that the market’s next great wealth-creator is already on your list. Skipping the market’s biggest compounders isn’t caution — it’s a market-timing call wearing a seatbelt.
The counterargument has a famous name attached: Peter Lynch called over-diversification “diworsification.” But here is what gets left out. In One Up on Wall Street, diworsification describes companies squandering shareholder money on acquisitions they didn’t understand — and investors buying businesses they couldn’t explain. It was never a cap on portfolio size. Lynch ran Fidelity Magellan with roughly 1,400 positions and beat the market for thirteen straight years. The man whose name defends concentrated portfolios operated one of the most diversified stock portfolios ever managed.
What Lynch actually opposed — owning things for no articulable reason — we agree with completely. Every one of the 54 names in the book has a stated thesis, a dividend, and a weight cap. That is the difference between more positions and diworsification: intent.
Two CFP-textbook points complete the picture. Diversification quality is measured by correlation, not count — one genuinely uncorrelated asset (gold, T-bills) does more work than ten additional names from the same sector. And “correlations go to one” in a crisis: when panic hits, similar stocks fall together no matter how many you own. Forty consumer staples are not forty diversifiers. That is why the book’s crash protection lives in its allocation — beta 0.50, a T-bill sleeve, a gold hedge — not in its position count.
On July 4, 2026 we rebuilt the $500K tier of the Midterm Election Dividend book: 37 positions became 54, and the maximum single-name weight dropped from 4% to 2.5%. Two broken theses left the book. Nineteen dividend payers came in — and yes, that includes the big boys, through the only door a dividend book honestly has: the ones that actually pay. Microsoft, Apple, Broadcom, Alphabet, Meta and Texas Instruments all pay and grow dividends. Amazon and Tesla pay nothing; Nvidia’s yield is a rounding error — they stay in our growth books, where they belong.
The rebuild also closed the book’s biggest structural gap — it owned zero financials — with five dividend stalwarts led by JPMorgan and Chubb, added two energy aristocrats plus the natural-gas pipelines feeding data-center power demand, defense and rail names whose budgets don’t depend on election outcomes, and the income side of the AI buildout itself: the data-center landlord and the nuclear operator signing hyperscaler contracts.
| Through July 2, 2026 | 37-name book | 54-name book | S&P 500 |
|---|---|---|---|
| YTD return | +10.3% | +11.3% | +9.8% |
| 1-year return | +13.1% | +16.7% | +21.4% |
| Annualized (5-yr) | +8.8% | +12.3% | +13.0% |
| Sharpe ratio | 0.40 | 0.71 | 0.52 |
| Volatility | 12.0% | 11.7% | 17.2% |
| Max drawdown | −11.7% | −12.4% | −24.5% |
| Beta vs S&P 500 | 0.41 | 0.50 | 1.00 |
Hypothetical total return from daily closes of current holdings at target weights, before fees, five years through July 2, 2026. Past performance does not guarantee future results.
The backtest says what the research predicted. Adding the paying mega-caps and financials raised the five-year annualized return three and a half points while volatility went down — the Sharpe ratio jumped from 0.40 to 0.71, ahead of the S&P 500’s own 0.52. The cost was seven-tenths of a point of drawdown and nine points of beta. That is the Bessembinder effect in our own numbers: owning the winners cost less risk than excluding them.
The mandate. This is still an election-defensive income book: every holding pays a dividend, the T-bill and gold ballast is untouched, and at 0.50 beta it still takes half the market’s risk. More names made the book harder to hurt through any single company — they did not, and cannot, make a bear market cheaper. That job still belongs to the allocation. We didn’t abandon discipline to add positions; we followed the evidence to the number the modern research actually supports.