——
Level — · Streak —
— XP
Mastery—weighted
Quizzes 30d—last 30 days
Cert progress—active level
Rank—global
Loading...
The page behind this dialog is live. Create a free account or sign in and you'll land right back on it.
How much to allocate — the science of bet sizing
Two investors generate the same 60% win rate, identifying winners and losers with identical analytical skill. Investor A sizes every position at 5%; Investor B sizes winners at 2% and losers at 12% by 'averaging down.' Same skill, same names, same time horizon. Investor A compounds at roughly 12% per year; Investor B compounds at zero — or worse — because the asymmetric sizing converts every wrong call into an oversized portfolio drag. Position sizing decides more of your long-term return than stock selection does. The math is unforgiving: a 75% loss requires a 300% gain to recover, and a 90% loss requires 900%. This lesson is about how much to allocate to each idea — and why the question is far more important than which idea to pick.
answers a question most retail investors don't even realize they're answering: how much of my portfolio do I bet on this idea? The answer matters more than the analysis behind the idea, because investing isn't a single decision — it's a series of decisions, and the long-run return is the geometric mean of those outcomes. The geometric mean punishes large losses far more than the arithmetic mean does, which is why the math of ruin (a 75% loss requires 300% gain to recover) drives almost every serious sizing framework. Two structural facts about sizing. First: the right size depends on your edge and your portfolio's tolerance for variance — not on how excited you are about the idea. Second: most retail investors size by emotion (sell winners early, hold and double-down on losers) and produce reverse-Pareto portfolios where their largest positions are their worst ideas.
Two competing sizing philosophies coexist in the literature. The MODERATE sizing tradition (Markowitz/Sharpe/most institutional advisors): cap any single position at 3-5% of the portfolio, with the bulk in broadly diversified ETFs. The argument is that you cannot reliably distinguish your true edge from luck, so you should size assuming you might be wrong on any given name. Bogle's 'Common Sense on Mutual Funds' (1999) takes this to the logical conclusion: own broad-index ETFs, accept market returns, and put the savings on costs into compounding rather than into stock-picking. The CONCENTRATED sizing tradition (Buffett/Munger/Klarman/Pabrai): when you have GENUINE analytical edge in a well-understood business, size up to 10-25% per position because diversification dilutes your edge across positions you understand less well. Buffett at Berkshire has historically held roughly 75% of equity assets in his top-5 names. The reconciliation, as in m8_l1: both frameworks are correct given their assumptions; the discipline is honestly assessing whether you have the edge required for concentration.
For active stock pickers who want concentration but not portfolio-destruction risk, the consensus framework (Pabrai's Dhandho Investor, Klarman's Margin of Safety, and the institutional value-investing tradition) is conviction-based sizing. High-conviction positions: wide moat, materially undervalued, identifiable catalyst, within circle of competence. Size: 5-8%, with absolute maximum 10% for the highest-conviction names. Moderate conviction: good business at fair price, no clear catalyst. Size: 2-4%. Speculative: interesting thesis with material uncertainty (turnaround, early-stage growth, special situation). Size: 0.5-2%. Index core: foundational ETFs that don't depend on stock-specific calls. Size: 20-40%. The discipline: NEVER size up because of a recent gain or down because of a recent loss; size based on the BUSINESS quality and valuation, not the stock price action. The platform's Insights and Valuation tabs answer the conviction question; your own honest assessment of circle of competence answers whether you should be concentrating at all.
Two extreme cases illustrate why position sizing matters. Berkshire Hathaway's equity portfolio has historically been concentrated: the top-5 holdings have typically represented approximately 75% of equity assets. As of recent 13F filings, the top-5 is approximately Apple, Bank of America, American Express, Coca-Cola, and Chevron — though the exact rank order shifts quarterly and the Apple position alone has been a 35-50% share of the portfolio at times. Buffett's argument: you only get a small number of genuinely-great ideas in a lifetime; when you find one, size it appropriately rather than diluting your edge across mediocre alternatives. The strategy works for Buffett because he has demonstrable circle-of-competence depth in a small number of businesses. The opposite extreme: Long-Term Capital Management, founded 1994 by John Meriwether with Nobel laureates Robert Merton and Myron Scholes on the team. LTCM ran a portfolio of statistical-arbitrage trades — small expected-value edges, sized up via leverage to 25x equity. The strategy worked beautifully through 1995-1997 (returns of ~20-40% per year). In August 1998, Russia defaulted on its debt and credit spreads widened far more than the LTCM models had assumed possible. The 25x leverage converted a small adverse move into a near-total wipeout: LTCM lost approximately \$4.6 billion of capital in roughly five weeks and required a Federal Reserve-organized rescue to avoid a systemic banking crisis. The lesson isn't 'concentration is wrong' — Buffett's concentrated portfolio has compounded at ~20% for decades. The lesson is that sizing depends on whether the edge is genuine and the leverage is sustainable. Buffett sizes up genuine business edge with no leverage; LTCM sized up small statistical edges with extreme leverage. The first survives crises; the second doesn't. Sources: Berkshire Hathaway 13F filings (most recent quarterly); Roger Lowenstein, 'When Genius Failed' (2000), the canonical LTCM history.
The Portfolio Hub displays each position as a percentage of total portfolio value, with target-vs-actual comparisons against your IPS limits. Concentration warnings fire when any single position exceeds the IPS-defined maximum. The Valuation tab on each holding answers the conviction question (multi-method valuation, margin of safety) — high conviction requires multi-method agreement, not just enthusiasm. The Insights tab summarizes the analyst case for and against, so you can see whether your conviction is supported by the broader investor base or you're effectively betting against consensus (sometimes correct, but always worth knowing). For the math-of-ruin asymmetry, the /screener page lets you compare current holdings against alternatives — a position drawdown that has reached -40% requires a 67% gain to recover, and the alternative may have a much better expected forward return. The Portfolio Lab page (Phase 2) will surface a Kelly-style position-sizing calculator for users who want to formalize the framework; until then, the half-Kelly conservative sizing rule above is sufficient.
The single most damaging position-sizing behavior in retail investing is averaging down into a position whose thesis is deteriorating. The mechanism: you buy at 4%, the stock falls 25%, fundamentals remain stable, you 'add to your conviction' at 6% — fine so far. Then fundamentals start to deteriorate (margin compression, KPI weakness, competitive pressure). Instead of selling on the thesis change, you average down again because the price is 'even cheaper' — except the price is cheaper because the business has gotten worse. Three more cycles of averaging down, and your largest position by dollar weight is your worst-performing thesis — the exact opposite of what conviction-based sizing intends. Klarman calls this 'the catalyst for permanent capital loss' in 'Margin of Safety' (1991): cheapness on a stock's price-vs-history chart is meaningless when the underlying earnings power has structurally declined. The discipline: separate sizing decisions from price decisions. If the THESIS is intact, the lower price IS an averaging-down opportunity (within position-size limits). If the thesis has deteriorated — as measured by KPIs, margin trends, competitive position — the right action is to sell, not buy more. The Ratios and KPIs tabs answer this question; your portfolio's 'cost basis' column does not. The hardest disciplinary moment is averaging down on the position that has dragged your portfolio for two quarters and might become your largest single loser. Buffett's 1989 letter discusses this exact pattern: 'There is no point in buying any more Apricot just because Apricot has fallen — you have to ask whether the underlying business has changed.' (Paraphrased; the full passage is in the 1989 chairman's letter.)
Few bets, big bets, infrequent bets. The Dhandho framework is built on John Kelly's optimal bet-sizing formula and the recognition that you only get a small number of truly attractive opportunities in a lifetime. When you do find one — when the heads-I-win-a-lot, tails-I-don't-lose-much asymmetry is unmistakable — size up. Index investing and small-position diversification are the right framework when you don't have edge; concentrated sizing is the right framework when you do. The mistake is mixing the two — running a diversified portfolio that you act on with concentration's confidence, or running a concentrated portfolio when you don't actually have the edge concentration requires.