Cross-account wash sale detection for stocks, options, and ETFs — local-first, IRS Pub 550 rules, with a tax-harvest planner and pre-trade simulator.
Project description
Cross-account wash sale detection for stocks, options, and ETFs — local-first, IRS Pub 550 rules, with a tax-harvest planner and pre-trade simulator.
Why · Quickstart · Features · Usage · Rules · Service · Architecture · Development
Why net-alpha
When you trade across multiple brokerages, each platform tracks wash sales only within its own ecosystem. A loss sale on Schwab can be silently neutralized by a repurchase on Fidelity — you won't find out until tax season, long after the window to act has closed. The problem compounds when you trade options alongside the underlying, or ETFs that track the same index.
net-alpha is a local-first Python tool that gives you a single, unified view of your wash sale exposure across every account, asset class, and tax year — before it's too late to act.
[!NOTE] The package is published to PyPI as
wash-alphabut the CLI command isnet-alpha. Both names refer to the same project.
See it in action
Screenshots below use the built-in demo dataset — TSLA, NVDA, AAPL, SPY puts, an SPX §1256 call, plus a handful of open holdings (MSFT, AMZN, GOOGL, META, AMD, …). Pick "Try the demo" on the welcome screen to play with the same scenarios live.
Wash-sale ledger — every match across accounts, with confidence label and rule citation. The NVDA row is the case no single broker would ever flag: a loss closed in schwab/taxable is silently neutralized by a buy in schwab/ira 14 days later. TSLA and AAPL are confirmed same-account round-trips. SPY is an exact-strike-and-expiry options match — the lag is -8d because the replacement leg was opened 8 days before the loss close.
Per-symbol drilldown — NVDA's timeline pulls lots from both accounts into one view, with realized / disallowed KPIs and the cross-account violation pinned underneath. Every row links back to the source trade so the chain is always auditable.
Pre-trade simulator — propose Sell 20 AAPL @ $170 and see realized P&L, FIFO lot consumption, and the wash-sale verdict before you place the order. The lot-strategy table compares FIFO / LIFO / HIFO / MIN_TAX / MAX_LOSS side-by-side, with each strategy's wash-sale verdict computed independently. Suggestion chips at the top surface the largest unrealized loss (NVDA −$3,098) and gain (AMD +$4,503) for one-click sims.
Portfolio dashboard — KPIs, equity curve, cash & contributions overlay, monthly realized P&L bars, allocation rollup, and top movers — every account combined in one view, with toggles to drill in per-account.
Quickstart
net-alpha requires Python 3.11+ and uv.
# 1. Install with the local web UI
uv tool install 'wash-alpha[ui]'
# 2. (Optional) Install the always-on background service
net-alpha service install
# 3. Open the dashboard (browser opens automatically)
net-alpha ui
The UI runs locally at http://127.0.0.1:8765. With the service installed, prices refresh every 4 hours and forward-looking wash-sale checks run daily; without it, net-alpha ui runs ephemerally (Ctrl-C to stop).
[!TIP] First run? Pick "Try the demo" on the welcome screen for a guided tour with sample data — no CSV needed. Replay any time from
/tour.
Prefer the terminal? The CLI works without UI extras:
uv tool install wash-alpha
net-alpha schwab.csv --account personal --detail
Features
Detection
- Cross-account intelligence — match a loss sale on one broker against a repurchase on another in a single pass. The whole point of the tool.
- §1091 wash sales — equities, options (exact contract or same underlying), and ETFs (same ticker or same-index pair). Every match has a Confirmed / Probable / Unclear confidence label with rule citation.
- §1256 awareness — index options (SPX, NDX, RUT, VIX, …) are recognized as §1256 contracts: wash-sale-exempt with statutory 60/40 LT/ST classification.
- §1092 straddles —
net-alpha straddlessurfaces literal straddles, married puts, non-qualified covered calls (with QCC test), and vertical spreads. Holding-period suspension warnings attach to affected lots. - Bundled "substantially identical" pairs — major index-tracking ETFs (S&P 500, Nasdaq-100, Russell 2000, …). Extend with your own at
~/.net_alpha/etf_pairs.yaml(additive, never replaces defaults).
Planning
- Pre-trade simulator —
/simshows FIFO lot consumption, realized P&L, and a per-account cross-account wash-sale verdict before you execute. Suggestion chips surface the largest unrealized loss, wash-sale risk, and largest unrealized gain. - Lot-selection strategies — compare FIFO / LIFO / HIFO / MIN_TAX / MAX_LOSS side-by-side on a Sell sim, with each strategy's wash-sale verdict computed independently.
- Tax-harvest planner —
/tax/harvest/planturns the harvest queue into a ranked, editable plan (greedy by tax saved, capped by §1211's $3,000 ordinary-loss limit). Honors user-declared PositionTargets so it never closes something you want to keep. - Forward-looking watchlist — the always-on service forward-simulates every PositionTarget daily and surfaces wash-sale / §1091 risk inline as colored pills on the Plan view.
- Action Inbox — a single panel rolls urgent items (imminent wash-sale tripwires, §1092 holding-period suspensions, broken reconciliation, missing basis) into one place.
Reporting
- After-tax performance —
/tax?view=performanceshows realized P&L after estimated taxes, with a tax-drag breakdown and an ST/LT/§1256 mix bar. - Capital-loss carryforward — auto-derived from prior years honoring §1212(b) cross-category netting and §1211 $3K cap, with hand-editable overrides at
/settings/carryforward. - Auditable explanations — every wash-sale flag includes rule citation, source trades, match reason, math, and confidence reasoning — inline in the UI or via
--detailon the CLI. - Per-symbol reconciliation — cross-checks computed P&L against your broker's Realized G/L file (Schwab supported); discrepancies surface inline on the ticker page.
- Data hygiene rollup — Imports page groups missing-basis / no-quote / missing-date rows so you can fix them in one pass.
Local & private
- 100% local, zero-knowledge — your trade data, accounts, and P&L never leave the box. Symbols are sent to Yahoo Finance for live quotes only; disable with
prices.enable_remote: false. - No CDN at runtime, no Node — htmx, Alpine, ApexCharts, Lucide icons, and fonts are vendored under
web/static/. - Manual trade CRUD — add, edit, transfer, or delete trades from the web UI; wash sales recompute over the affected window automatically.
Usage
Local web UI (primary surface)
net-alpha ui [--port 8765] [--no-browser] [--reload]
| Page | Highlights |
|---|---|
/ Portfolio |
KPIs, allocation, equity & cash curves, top movers, options & short-options panels, wash-sale watch |
/positions |
Holdings (all / stocks / options / at-loss / closed), drag-to-reorder Plan view, multi-lot basis editor |
/sim |
Pre-trade simulator with suggestion chips and lot-strategy comparison table |
/tax |
After-tax performance, harvest queue, plan-builder, projection setup, wash-sale + exempt-match listings |
/imports |
Drop-zone upload, preview/commit, per-import detail, data-hygiene buckets |
/ticker/{symbol} |
Per-symbol timeline, lots, reconciliation, lot edit + add-trade forms |
/settings |
Profile, density, ETF pairs, carryforward, service controls, about |
Server-side HTMX + Alpine, dies on Ctrl-C.
CLI
# Import a CSV → recompute wash sales → print watchlist + YTD impact
net-alpha schwab.csv --account personal [--detail]
# Pre-trade simulation across every account holding the ticker
net-alpha sim TSLA 10 --price 180
# §1092 straddles currently open
net-alpha straddles [--detail]
# Manage past imports (recomputes wash sales on remove)
net-alpha imports
net-alpha imports rm 3 --yes
# v1 → v2 schema migration helper (v2.0.x line only)
net-alpha migrate-from-v1 --yes
Adding a broker
Bundled at launch: Schwab (transactions + Realized G/L for audit reconciliation). To add another:
- Implement the
BrokerParserProtocol atsrc/net_alpha/brokers/<name>.py. - Register it in
brokers/registry.py. - (Optional) Add a Realized G/L parser at
src/net_alpha/audit/brokers/.
How the rules work
net-alpha strictly follows IRS Publication 550. A wash sale is triggered when you sell a security at a loss and buy a substantially identical security within 30 days before or after the sale.
| Asset type | Scenario | Confidence |
|---|---|---|
| Equities | Sold ticker X at a loss, bought ticker X within 30 days |
🟢 Confirmed |
| Options | Sold option at a loss, bought same option (exact strike + expiry) | 🟢 Confirmed |
| Options | Sold option at a loss, bought option on the same underlying | 🟡 Probable |
| ETFs | Sold ETF at a loss, bought the exact same ETF ticker | 🟢 Confirmed |
| ETFs | Sold ETF at a loss, bought ETF tracking the same index (e.g., SPY → VOO) |
🔵 Unclear |
§1256 contracts
Broad-based equity index options (SPX, NDX, RUT, VIX, OEX, XSP, …) are exempt from §1091. Closed §1256 P&L is split 60/40 LT/ST per §1256(a)(3), regardless of holding period. Open positions at year-end are not marked-to-market in v1; consult your 1099-B / Form 6781.
§1092 straddles
Same-underlying offsetting positions caught by IRC §1092:
- Literal straddles — long call + long put on the same underlying
- Married puts — long stock + long put on the same underlying
- Non-qualified covered calls — long stock + short call that fails the §1092(c)(4) Qualified Covered Call test
- Vertical spreads — long + short option of the same series, opposite legs
[!NOTE] The QCC test in v1 is a conservative approximation of IRS Notice 2003-31's lowest-qualified-benchmark step table — it errs on the side of flagging deeper-ITM calls. Loss-deferral computation, the §1092(b) modified wash-sale rule, identified-straddle elections, and correlation-based stock/ETF offsets are deferred to v2.
Always-on service
net-alpha service is an opt-in launchd-supervised process that hosts the FastAPI app and runs two background jobs:
price_refresh— every 4 hours, refreshes quotes for held + targeted tickers.washsale_watch— daily 04:00 + on every manual import. Forward-simulates each PositionTarget and surfaces wash-sale / §1091 risk in the Plan view.
Lifecycle
| Command | Effect |
|---|---|
net-alpha service install |
Provision runtime venv, write launchd plist + sandbox profile, (re)load the agent. Idempotent. |
net-alpha service start / stop / restart |
Lifecycle. stop survives reboots until start. |
net-alpha service pause / resume |
Freeze background jobs but keep the dashboard reachable. |
net-alpha service status [--json] |
Health report. |
net-alpha service logs [-f] |
View / tail the service log. |
net-alpha service uninstall |
Remove plist, sandbox, and runtime venv. Data at ~/.net_alpha/net_alpha.db is preserved. |
The dashboard at /settings/service exposes the same controls in a UI surface, plus recent-runs history and a status pill in the site header.
What install does
Everything the running service needs lives inside ~/.net_alpha/, the only path the sandbox profile allows the service to write to:
~/.net_alpha/venv/— dedicated runtime venv withwash-alphainstalled from your local project source.~/.net_alpha/bin/net-alpha-wrap— the launchd wrapper script.~/.net_alpha/run/sandbox.sb— the sandbox-exec profile applied to the running service.~/Library/LaunchAgents/com.netalpha.service.plist— the LaunchAgent registration.
[!IMPORTANT] The runtime venv intentionally lives outside
~/Documents/. Under launchd, the service runs with a TCC identity that cannot read user-data folders, so the project's own.venv/is unreadable;~/.net_alpha/is TCC-clear.
The runtime venv is a non-editable snapshot of your project source at install time. To make the running service reflect new code or pyproject.toml changes, re-run net-alpha service install. For active development, run net-alpha … directly from your editable .venv — the LaunchAgent is for the always-on service, not the dev edit-run loop.
Architecture
CSV → BrokerParser → Trade (Pydantic) ──► SQLite (~/.net_alpha/net_alpha.db)
│
├─► Wash-sale engine (incremental ±30-day window)
├─► §1256 classifier (60/40 LT/ST split)
├─► §1092 straddle detector
├─► Reconciliation (against broker Realized G/L)
└─► Portfolio / planner / pricing (pure functions)
│
┌─────────────┴─────────────┐
▼ ▼
Typer CLI FastAPI + HTMX UI
│
launchd-supervised service
(price_refresh + washsale_watch)
- Stack — Python 3.11+,
pydanticv2,sqlmodelover SQLite,typerCLI, optionalfastapi+ Jinja + HTMX UI, APScheduler for the service,ruff,pytest. Managed withuv. - Storage — single SQLite DB at
~/.net_alpha/net_alpha.db; cross-year window detection works because all trades live in one place. Schema versioned via ametatable; hand-writtenALTER TABLEmigrations. - Engine — wash-sale recompute is incremental: only the ±30-day window around affected trade dates is recalculated on import or removal. Every emitted violation is reproducible from source trades.
- Web UI — optional subpackage at
src/net_alpha/web/. Calls only existing public seams (Repository, engine functions, audit/portfolio/planner pure functions); no business logic inweb/. Static assets vendored — no CDN at runtime, no Node toolchain.
Development
uv sync --extra ui --extra dev # install everything
uv run pytest # tests (or: make test)
uv run ruff check . # lint
uv run ruff format . # format
make build-css # rebuild the Tailwind bundle
Disclaimer
[!IMPORTANT] This tool is for informational purposes only and does not constitute tax or legal advice. Wash sale rules — especially around options and ETFs — involve unsettled areas of tax law. Anything labeled
ProbableorUnclearshould be reviewed with a qualified CPA before making filing decisions.
If net-alpha saves you a CPA call this tax season, consider starring the repo — it's the best signal for other traders looking for a tool they can trust with their tax data.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file wash_alpha-0.64.0.tar.gz.
File metadata
- Download URL: wash_alpha-0.64.0.tar.gz
- Upload date:
- Size: 5.3 MB
- Tags: Source
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/6.1.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
65a676e14d8042431436772f3322d283adc438d590324c2ea905ad20ffd0bef5
|
|
| MD5 |
a4f0accbc28c99c7a965209a0a22ba7f
|
|
| BLAKE2b-256 |
b3e6941ce990c7cba2a1e2b8dcbde8fc5c729a9e60111ce025176a279e957218
|
Provenance
The following attestation bundles were made for wash_alpha-0.64.0.tar.gz:
Publisher:
release.yml on chen-star/net_alpha
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
wash_alpha-0.64.0.tar.gz -
Subject digest:
65a676e14d8042431436772f3322d283adc438d590324c2ea905ad20ffd0bef5 - Sigstore transparency entry: 1515278567
- Sigstore integration time:
-
Permalink:
chen-star/net_alpha@57bb98d9fc6b7f66174f43a23ace464fc210a989 -
Branch / Tag:
refs/tags/v0.64.0 - Owner: https://github.com/chen-star
-
Access:
public
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
release.yml@57bb98d9fc6b7f66174f43a23ace464fc210a989 -
Trigger Event:
push
-
Statement type:
File details
Details for the file wash_alpha-0.64.0-py3-none-any.whl.
File metadata
- Download URL: wash_alpha-0.64.0-py3-none-any.whl
- Upload date:
- Size: 1.4 MB
- Tags: Python 3
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/6.1.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
a28303ac1fe28181b86d69aa00ac1371aedc784e7fa8029b75842fc397d16373
|
|
| MD5 |
35679a6b3fc3915c3e2d8e47a2ebcbc4
|
|
| BLAKE2b-256 |
7fdce0a172b30d4b59804f6a5675833316598a7b066440bc36de94b6a41fb9fe
|
Provenance
The following attestation bundles were made for wash_alpha-0.64.0-py3-none-any.whl:
Publisher:
release.yml on chen-star/net_alpha
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
wash_alpha-0.64.0-py3-none-any.whl -
Subject digest:
a28303ac1fe28181b86d69aa00ac1371aedc784e7fa8029b75842fc397d16373 - Sigstore transparency entry: 1515278705
- Sigstore integration time:
-
Permalink:
chen-star/net_alpha@57bb98d9fc6b7f66174f43a23ace464fc210a989 -
Branch / Tag:
refs/tags/v0.64.0 - Owner: https://github.com/chen-star
-
Access:
public
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
release.yml@57bb98d9fc6b7f66174f43a23ace464fc210a989 -
Trigger Event:
push
-
Statement type: