Decision Support

Four reasoning frameworks Claude applies whenever your question calls for one: pre-mortem before a buy, rebalance when concentration drifts, tax-harvest at year-end, and scenario analysis when something in the news looks scary. Slash commands are there for explicit invocation. Otherwise Claude triggers them on its own.

/pre-mortem · "why might this go wrong?"

Five concrete failure modes for a position over the next 12 months - each tied to actual data (macro signal, news, position weight), with dollar impact on your net worth and one mitigation. Designed to surface what you're missing, not to confirm the trade.
Auto-fires when you're considering a buy

You

Thinking about adding 50 more TSLA shares. Is this safe?

Claude

You hold 392 shares already (82% of portfolio). Adding 50 more pushes that to 87%. Five ways this hurts you in the next 12 months:

1. Q2 deliveries miss. Consensus is 480k units. Last quarter missed by 8%. A 10% earnings miss → ~$22K hit on your portfolio (5.4% of net worth).
2. Margin compression. Auto gross margin trending down 4 quarters running...
3. Robotaxi narrative breaks. Optimus/FSD timelines have slipped twice...
4. Sector rotation out of mega-cap tech. XLK -8% in last regime shift...
5. Single-name regulatory. NHTSA investigations active...

Most underprepared for: #1. Earnings is in 11 days. Mitigation: hold off on the add until after the print, or hedge with a deep-OTM put for ~$400 premium covering 5% downside.

/rebalance · drift vs target

Compares current allocation to your stated target, flags every drift >5pp, and suggests specific trades, with tax cost on sells, original transaction reason check, and wash-sale flags. Auto-fires when concentration in a single ticker or sector exceeds 30%.
Auto-fires on concentration drift

You

How's my portfolio balance these days?

Claude

TSLA is 82%. Your target was 40%. That's 42pp over. To rebalance, sell 191 shares ($54K). 73 of those have a $245 cost basis (long-term gain ~$10K, ~$2.4K tax @ 24%). The other 118 are recent buys with an average cost of $310, so sell those last to defer tax. Your last entry says "adding to the dip." TSLA is now +60% from that buy, so that thesis has already played out. Recommendation: sell 73 long-term-gain shares now and defer the rest to next tax year.

/tax-harvest · year-end loss scan

Scans every position for unrealized losses, applies the 30-day wash-sale rule by checking recent buys, suggests similar-but-not-substantially-identical replacements, and estimates tax savings at common brackets. Auto-fires when you mention taxes or year-end planning.
Auto-fires near year-end / tax mentions

You

It's November. Anything I should be doing for taxes?

Claude

Three positions with unrealized losses totaling $14,200:

RIVN -$8,400 (-32%) · 200 shares · ✓ harvest-eligible
SOFI -$3,100 (-18%) · ⚠ blocked (you bought 50 more on Nov 3)
PLTR -$2,700 (-9%) · ✓ harvest-eligible

Eligible loss: $11,100. Tax savings: $2,664 @ 24%, $3,552 @ 32%, $4,107 @ 37%.
Replacements: RIVN → consider TSLA (or wait 31 days to rebuy). PLTR → no clean equivalent, accept the gap.

※ general guidance, not tax advice. Confirm with your accountant.

/scenario · "what if X happens?"

Models a market shock or macro shift on your actual portfolio. States the beta assumptions upfront so you can override and re-run. Auto-fires when you ask "what if" or seem anxious about a macro event.
Auto-fires on hypotheticals

You

What if the S&P drops 20% from here?

Claude

Assumptions: TSLA β=1.4, NVDA β=1.6, AAPL β=1.2, KO β=0.6 (state these upfront so you can challenge them).

S&P -20% → portfolio -27% (≈$118K loss, 27% of net worth). Worst hit: NVDA -32%, TSLA -28%. Most resilient: KO -12%. Net worth would drop from $441K → $323K, putting you 73% from FIRE target instead of 44%.

One defensive move with the lowest expected-return cost is trimming 50 TSLA shares. That cuts portfolio beta from 1.35 → 1.18 and recovers about $15K of the projected loss. This is an option, not a panic call. Your runway is still 60 months at current expenses.

When you want to invoke explicitly

Type the slash command in Claude Desktop. Each enforces a strict tool-call sequence so the analysis is anchored to your real holdings.
/pre-mortemFive failure modes + dollar impact + one mitigation
/rebalanceCurrent vs target, specific trades, tax + wash-sale checks
/tax-harvestLoss scan, wash-sale rule, replacements, tax savings by bracket
/scenarioShock modeling with explicit beta assumptions you can override
MIT License