Your portfolio already has a benchmark. You just haven't seen it.

Berkshire's top-15 13F book is 52% idiosyncratic — more than half of the variance can't be explained by market, sector, or subsector ETFs. That's an unusual profile for a portfolio of its size, and it lines up with how the book was actually built: a small number of high-conviction stock picks rather than factor bets.

The sector layer (26%) is heavier than you'd see in a typical large-cap book — that's the financials concentration (AXP + BAC + COF + CB + MCO ≈ 38% of capital) plus the energy cluster (CVX + OXY ≈ 12%) showing up as real sector co-movement, not just diversified beta.

Two patterns. The dominant position is also the most idiosyncratic, and a cluster of smaller positions sits at the extreme end of stock-specific risk:

The opposite pattern in the energy book — CVX (7.1%) is 81.8% sector ER, OXY (5.1%) is 64.1% sector ER. Those two positions are mostly an XLE bet wearing two tickers; their residual contribution is small.

The takeaway: there's no single thesis here. AAPL is a concentrated stock pick. CVX/OXY are a sector tilt. DVA is a microcap-style idio bet inside a mega-cap book. The 52% portfolio residual is the cumulative signal that these positions are not factor exposures in disguise.

AAPL · Apple Inc. — Information Technology / Hardware Subsector benchmark: XLK · 23-day vol 24.6% · as of 2026-05-14

Two things to flag here. First, the sector ER is essentially zero — at the L3 cascade level, AAPL is not co-moving with XLK. That's unusual for a megacap "tech" name; it tells you the typical "Apple = tech beta" framing is wrong once you orthogonalize against market. AAPL's tech-sector beta is largely already inside the SPY load (look at that −2.00 market hedge ratio — you'd short $2 of SPY per $1 of AAPL to neutralize the market piece alone).

Second, AAPL is 26.3% of the Berkshire book and 67.1% of its own variance is residual. That means:

26.3% × 67.1% ≈ 17.6% of total Berkshire variance lives inside AAPL-specific moves.

If AAPL underperforms its market+sector load by 10% in a quarter, that single residual hit shows up directly in the portfolio with no factor diversification to dampen it. The whole Berkshire residual budget (51.8%) is dominated by this one position — the next-largest residual contributor is AXP at roughly a third of AAPL's load. The 52% idio number isn't 15 stocks each chipping in a little; it's mostly AAPL.

This is the trade-off of a concentrated book: the conviction shows up as performance and as concentration risk in the same number.

Three levels, increasing completeness. The invariant across all three: total market-beta exposure removed is the same — what changes is which instruments carry it. Sector ETFs (XLF ≈ 1.1β, XLE ≈ 0.9β, XLP ≈ 0.6β) carry their own market beta, so shorting them removes market exposure on top of the sector tilt. The SPY balancing position shrinks as you add ETF layers; this is the institutional construction, not constant-SPY-across-tiers.

L1 hedge — market only (removes 19.6% of variance)

• Short ~$196,000 in SPY
• No sector or subsector hedges
• Right call if you want to keep the financials/energy/staples tilts as deliberate bets

L2 hedge — market + sector (removes 45.8% of variance)

• Short ~$30–50,000 in SPY (residual balancing — the sector shorts already remove most of the market beta)
• Net ~$262,000 across XLF / XLE / XLP (the three dominant sector loads)
• Total market exposure removed = same as L1; plus the 26.2pp sector layer
• Leaves the subsector layer and the 51.8% residual

L3 hedge — full systematic (removes 48.2% of variance)

• Short ~$4,000 in SPY (residual balancing — subsector ETFs add a small additional market-beta contribution)
• L2 sector trades unchanged
• Plus small subsector overlays
• Marginal improvement over L2 — the subsector layer is only 2.4% of variance here, so the operational complexity isn't usually worth it for a book like this

What's left after any of these: the 51.8% residual — Buffett's actual alpha bet. That's not hedgeable with ETFs. If you want to neutralize it, you'd need single-name shorts or puts on the biggest residual contributors (AAPL is the obvious one — 26% weight × 67% residual ≈ 18% of total variance lives in AAPL-specific moves).

Strategic question to ask yourself before you trade any of this: are you trying to own Buffett's stock-picking (in which case L2 is the right level — neutralize the macro, keep the alpha) or replicate the book without the directional exposure (in which case you also need to short the top residual names)? The hedge structure should follow the intent, not the other way around.