Intelligence

Tax Strategy

8 min read

What it is

Tax Strategy turns your real fills into three decision tools: a Tax-Loss Harvest view that finds your biggest unrealized losses + suggests conservative non-substantially-identical replacements; a What-If Scenarios tool that compares Sharpe / Vol / Max DD / VaR / Annualized Return across two portfolios using real Binance OHLCV; and a My Lots ledger reconstructed FIFO from your bot fills. Stable-quote-only USD anchoring — volatile pairs (e.g. ETH/BTC) are skipped because USD basis cannot be assigned without an oracle.

How to think about it

This is one of the platform's signature features. Most retail tax tools either miss DEX activity entirely or fabricate USD basis on volatile-quote pairs. We do neither — fills with no honest USD anchor are skipped and disclosed, not invented. Treat all output as estimates; consult a CPA before acting.

Step-by-step

  1. Tab 1 — Tax-Loss Harvest

    Surfaces every open lot currently sitting at a loss. Each card shows the lot details (open date, cost basis, current value, short vs long term), the loss amount, estimated tax savings at YOUR tax rate, and 1-3 conservative replacement options (e.g. ETH → WSTETH, BTC → WBTC, SOL → MSOL). Replacements maintain similar economic exposure without being substantially identical.

  2. Harvest: Effective tax rate slider (10-50%, default 30%)

    Your combined federal + state marginal rate for this gain category. US federal long-term capital gains range from 0-20%; short-term is your ordinary income rate. Add state on top (CA ~13%, NY ~10%, TX/FL 0%). 30% is a reasonable starting estimate for most middle-bracket US filers. Non-US users: estimate your local equivalent.

  3. Harvest: Min harvest amount slider ($5-500, default $25)

    Hides opportunities smaller than this loss. Default $25 filters out trivial harvests where transaction fees would eat the savings. Raise to $100+ for cleaner output if you have many tiny lots. Lower to $5 if your tax rate is high and even small losses matter.

  4. Tab 2 — What-If Scenarios

    Side-by-side risk metrics for TWO portfolios — your current allocation vs a proposed alternative. Computes annualized return, annualized volatility, Sharpe, max drawdown, and daily VaR 95% over real Binance OHLCV. Each weight must sum to 1.0. Assets without Binance history are excluded and disclosed.

  5. What-If: Lookback window (30 / 90 / 180 / 365 days)

    30d = noisy, useful for very recent regime questions. 90d (default) = the standard portfolio analytics window. 180d = covers most market mini-cycles. 365d = covers a full crypto cycle phase. Longer windows give more stable metrics but may dilute recent regime signal.

  6. What-If: Reading the deltas

    Green Δ = proposed is better on that metric (higher return / lower vol / higher Sharpe / lower DD / higher VaR). Red Δ = worse. The trade-off picture matters more than any single number — a +0.2 Sharpe lift at the cost of +5% max drawdown might or might not be worth it depending on your tolerance.

  7. Tab 3 — My Lots

    FIFO lot ledger built from your real bot fills. Top 4 cards: Open Cost Basis, Open Market Value, Unrealized P&L, Realized P&L. Table below: per-lot detail (asset, qty, basis, current value, unrealized P&L, open date). Skipped pairs (non-stable quote like ETH/BTC) shown in an amber footer.

  8. Conservative replacement rationale

    The replacement map (visible at /api/tax-strategy/replacement-map) deliberately avoids substantially-identical tokens. ETH → WSTETH because WSTETH is a staked-ETH wrapper with distinct issuer (Lido) and tradeable economics. BTC → WBTC because WBTC is BridgeDAO-issued, not native BTC. Never blind-swaps a token for an exact-mirror IOU.

  9. Manual execution only

    haythix does NOT auto-execute harvest trades. Every suggested swap requires you to open Swap or CEX Trading and place the trade manually. Engine-enforced, not just a UI convention — there is no auto-harvest mode and there will not be one. Tax decisions deserve human review.

Tips & pitfalls

  • Not tax advice. Run output past a CPA before acting, especially for large lots or first-time harvest in a new jurisdiction.
  • IRS has not formally extended the 30-day wash-sale rule to crypto, but conservative practitioners treat it as if it does. The replacement map is built around this assumption — replacements are NOT substantially identical.
  • Save or print the Harvest output before executing. The disclosures (your tax-rate assumption, wash-sale caveat, replacement rationale) are part of an audit trail your CPA will appreciate.
  • Volatile-quote pairs (ETH/BTC, SOL/ETH, etc.) are deliberately skipped from the lot ledger. We refuse to invent a historical USD price for them. Coming post-launch: optional oracle-based pricing for these pairs with explicit disclosure.
  • Short-term loss is more valuable to harvest than long-term loss — short-term gains taxed higher = bigger savings per dollar harvested. The page labels each lot as short vs long term.
  • Run What-If BEFORE harvesting. If your proposed post-harvest portfolio has worse Sharpe + higher DD, the tax savings may not justify the risk shift.
  • Wash-sale rules differ by jurisdiction. UK has 30 days; Canada has 30 days; Australia has 12 months; most EU states have no formal rule. Adjust your replacement timing accordingly.