Rebalancing & Position-Sizing Proposal (De-risk the Top-Three Concentration)

Prepared: 2026-08-25 | Source: concentration.json, scenarios.json, trim_lots.json, profit_targets.json, wash_sale_risk.json, harvest_trade_list.md

Read-only sizing proposal. This documents *what* and *why* only. No order is placed and nothing is auto-traded. Review every lot, confirm live prices, and consider commissions/fills before you act.

1. Current concentration & the problem

The AI cluster (HOOD + TSLA + NBIS + NVDA + AMZN + GOOGL, plus leveraged tails TSLL/TQQQ) is ~80% of the equity sleeve and ~68% of real base.

Why it matters (correlated single-name + cluster risk):

• Single-name blowup: HOOD alone is ~32% of real exposure. A -40% HOOD drawdown costs ~$43.3k while everything else is flat (scenarios.json).
• Correlated cluster drawdown: a -30% AI tape = ~$86.5k combined hit (core six -51.3k, TSLL 2x decay -35.2k). A -15% tape already = ~$43.3k. These are the $70–90k swings the dashboard flags — the leveraged tails exist purely to amplify them.
• There is almost no diversifying ballast in the equity sleeve: the "others" are themselves crypto/AI/noise, and the book is a single high-beta AI/retail-trading basket.

2. Proposed de-risking actions (keep the conviction, cut the tail)

Philosophy stays intact: trim the top from the top and cap the leveraged tail — do not exit the high-conviction core.

A. HOOD — trim to a target weight (~22–25% of real base)

• Target: reduce from $108.1k (32.2%) toward ~$75–82k (~22–25%). That is a trim of roughly $27–33k notional.
• Phase it. First step fits the existing tax-optimized trim_lots.json: sell 150 sh (LT lot, cost $41.06) ≈ $15.5k proceeds / ~$9.4k realized LT gain**. This single trim cuts HOOD to ~27.6% of the equity sleeve.
• If the target is ~22–25%, either extend the trim to the next LT lots later, or pair with the TSLL tail-cut below so the *combined* top-three weight drops under pressure without dumping the whole HOOD position.
• Why: this is the single largest, unhedged, non-diversifiable name. Trimming the profit is tax-efficient (long-term lots, only ~15% of the position) and does not abandon the thesis.

B. TSLA — the risk is option delta; keep it modest

• The +200 share net-long delta ($72.6k) is mostly LEAPs — you cannot cheaply size LEAPs down (wide bid/ask, high per-basis-point cost to strip delta).
• Do not rush to unwind the deep-ITM LEAP picks. Instead:

- Let the existing short legs carry more weight — with +4 long / -2 short, adding/re-opening a short leg against the long delta reduces net long without selling LEAPs at a loss. A handful of short contracts can shave several thousand in delta neutral-ly.
- **Or take profit on a *portion* of the LEAP delta** only if the LEAP is well in-the-money and profit-taking is part of an existing plan — keep it modest, do not exit the position.

• Net effect target: bring the ~21.8% option-delta weight down toward the low-teens % without converting a long thesis into a trade-out at a loss.

C. TSLL — cap the 2x leveraged tail (reduce to ~half)

• Currently $56.7k / 16.9% of real base — this is the single highest-volatility, highest-correlation line item (2x daily TSLA, so a -30% TSLA tape is -60% on TSLL).
• Trim to ~half (sell ~2,500–3,000 of the 6,000 sh) → cut $28–34k of notional leveraged exposure.
• Wash-sale aware — this is a loss position. Sell only the losing lots (the harvest_trade_list.md losers = $9.1k realized loss) and do not rebuy TSLL until 2026-09-23 to keep the loss recognized. The "reduction to half" is a *sales* action, not a re-entry, so it pairs cleanly with the harvest.
• Why: you keep a meaningful long TSLA view (the LEAP net-long stays), but you stop paying 2x decay for it. Capping the tail is the single biggest variance-reducer per dollar of notional sold.

Tax note — the harvest funds the trims

• The harvestable losses are currently ~$14.9k (CRWV $5,753 all-ST + TSLL $9,108 losing lots), in line with the $16.5–19.5k band when the losing TSLL/CRWV lots and any other losing lots are included.
• Realizing these offsets the ~$9.4k HOOD LT gain and the TSLL/other trims tax-neutrally (net recognized gains ≈ $0), letting you shrink the book without a large tax bill. This is the lever that makes the whole de-risk tax-efficient this calendar year.

3. Target allocation sketch (after-rebalance, % of real base ~$335.9k)

Rounded sketch; "of base" = % of the ~$335.9k real-exposure denominator. The goal is to keep conviction in AMZN / NVDA / GOOGL / NBIS while cutting the top-three to more manageable weights.

Top-three combined weight drops from ~71% to well under ~45%, and no single name carries >~25% of base. The AI-cluster % stays high (you are not exiting the thesis) but the *leveraged-amplified* portion is halved and the *single-name* HOOD concentration is cut roughly in half.

4. What NOT to do

• Don't sell the deep-ITM LEAP picks to raise cash at a loss. The LEAPs are the efficient, low-cost way to hold TSLA conviction; unwinding them early locks in realized losses and wide spreads. Size the LEAPs down only via short legs / modest profit-takes, never by dumping them for liquidity.
• Don't over-trim into a wash sale. TSLL/CRWV are loss positions with a 30-day (61-day centered) stay-out window to 2026-09-23. Trim them as *sales*, and don't rebuy the same/substantially-identical security in that window or the loss is disallowed. Don't rebuy HOOD back within the same trade that realizes its gain in a pattern the IRS treats as a wash on the trim lots.
• Don't chase diversification into unrelated low-conviction names. Rebalancing *out* of concentration by buying N new names you have no thesis on merely adds uncompensated noise and defeats the concentration you're deliberately running. If you want ballast, a broad low-cost index position is the honest tool — not scattered single names.
• Don't do it all in one frantic day. Trim the HOOD LT lot, harvest CRWV/TSLL, and cap the TSLL tail as a *sequence*; re-check live prices and fills between steps. This is a sizing decision, not a market-timing event.

*Generated by the vibe-dashboard sizing pipeline. Informational only — not tax or investment advice. No orders are placed by this document.*