Everything a desk does.
Nothing a desk hides.
Every feature below ships in the app today, and each one is here because it changes what you can do — not because it made a list longer.
Research that comes to you
Nightly whole-market screener
Every night after the close, Greenwick ranks 10,000+ US stocks and ETFs on momentum and volume — pure math, zero AI tokens — and rotates the best into your watchlist. Hand-added names are never touched; leveraged and inverse funds are excluded on purpose.
The opportunities you'd never think to type in find you instead.Morning & midday scans
Before the open and again midday, the desk scans your whole watch set for momentum and volume leaders and writes a brief you can actually read — with the day's earnings landmines flagged.
You start the day already knowing where to look.On-demand analysis of any symbol
Type a ticker, get the full pipeline: five analyst reports, the bull-vs-bear debate, a sized decision with confidence, and the exact reasons. "Analysis only" mode stops short of any order.
Deep research on your terms — including research that never trades.Charts that carry your thinking
Candles from daily down to 2-minute, SMA/EMA/VWAP overlays, your own levels and trendlines saved per symbol, price alerts that ring once when hit, and an overlay showing the exact stop and target prices the safety machinery would act at.
The chart shows what the desk will actually do, not just what happened.Decisions with receipts
The bull-vs-bear debate
Before any decision, one agent builds the strongest case for the trade and another builds the strongest case against it. The portfolio manager must weigh both and say which points decided it.
Greenwick actively looks for reasons not to trade — most tools only look for reasons to.One-click approvals
With "Ask me first" on, the desk writes the complete ticket — entry, size, stop, target, reasoning — and waits. Edit it, approve it, or ignore it. Nothing reaches your broker without you.
You get the leverage of automation while keeping the final say.A risk-checked manual ticket
Trade your own ideas through the same protections: live quote under the symbol field, estimated cost as you type, and every manual order passes the same risk engine as the AI's.
Your fat-finger risk gets the same guardrails as the AI's judgment.Assistant with web search
One chat for questions about symbols, terms, or what the desk is doing — with optional web search through your own key, always citing sources. It has no trading authority, by design.
Answers when you want them, and a hard wall between chat and your account.Safety that isn't a setting
A deterministic risk engine
Position caps, daily-loss lockout, confidence floors, no averaging down, session cutoffs — in plain code outside the AI loop, checked on every single order. Presets set the envelope in one click: Conservative, Standard, Aggressive.
The one part of trading that should never be creative, isn't.The STOP button
One click cancels every order Greenwick placed, switches every strategy off, and blocks all Greenwick trading until you press Resume — in under two seconds, from the app or the menu-bar tray, even if the window is frozen.
However automated your desk gets, the brake is always yours and always works.Safety exits & reconciliation
Stops, take-profits, time exits, and end-of-day rules run on a fixed cadence for every position Greenwick opened — even for strategies you've turned off, even on a lapsed subscription. A reconciliation loop treats your broker as the source of truth every minute.
Protection doesn't depend on anything being enabled — or paid for.Paper first, and yours stays yours
Every account starts in your broker's paper environment — real prices, zero dollars. This release trades paper only; live trading arrives in a future update, once our legal review completes, behind its own deliberate enable ceremony. And positions you held before Greenwick, or bought yourself, are never touched: no exits, no strategy trades, ever.
You can evaluate everything with nothing at stake — and nothing of yours is ever assumed.Strategies you can read
Plain-English strategy cards
Momentum, scalp, and earnings strategies each say what they do, when they act, and what limits them — with an on/off switch and a Conservative / Standard / Aggressive dial. No configuration files, no code.
You can't trust automation you can't read.Scale-out profit taking
A winning scalp banks 25% at each of three rising targets, then a trailing stop protects the runner. Levels are fixed strategy settings — nothing about the exit is decided by AI in the moment, and no scalp ever holds overnight.
Exits are where discipline pays, so exits are mechanical.Cold-streak probation
Five consecutive losers automatically dial a strategy down to Conservative; eight pause it and tell you. A deterministic rule, not an AI mood — and your own settings always win.
The system de-risks itself exactly when tilt would do the opposite.Auto-Invest
Recurring dollar amounts into a small ETF portfolio on your schedule — it only ever buys, never sells what you hold, tracks dividends, and benchmarks you honestly against the same dollars into SPY on the same days.
The boring, proven strategy runs itself — with a benchmark that can't flatter it.A desk that remembers
Journal & postmortems
Every completed round trip is logged with its outcome and exit reason, then written up by a postmortem agent. The distilled lessons feed straight into future analysis runs.
Most traders never review; your desk reviews every single trade.Honest reports
Evening and weekly reviews pair AI narration with deterministic scorecards computed from your real numbers — day P/L, per-strategy attribution, equity vs. the S&P — with data gaps marked rather than papered over.
The narrative can't drift from the numbers, because the numbers come first.Track records the AI must face
Each strategy's analysis runs include its own real hit rate, average win and loss, and current streak as plain context — history, honestly labeled, not cheerleading.
The AI reasons over how it's actually been doing, not how it imagines it has.Model benchmarking
Compare models and prompts against recorded runs with frozen inputs — replay a real analysis byte-for-byte under a different model and see what changes, with paired statistics.
"Which AI should I use?" becomes an experiment, not a guess.Bring your own everything
Eight AI providers, or fully local
Anthropic (Claude), OpenAI (GPT), Google (Gemini), xAI (Grok), Mistral, DeepSeek, Azure OpenAI, and any OpenAI-compatible service — or run Ollama models entirely on your Mac. Keys are validated live before saving, then stored only in your keychain. No AI configured at all? Screens, scans, charts, guardrails, and reports still work.
You pay your AI provider directly — no markup — and can leave any of them tomorrow.Your broker, your data feed
Connect your own Alpaca account with your own API keys — paper environment first, always. Pay for a Massive (Polygon) data subscription? Plug it in and it powers the charts, scans, and screener, while order safety deliberately stays on broker data.
Greenwick orchestrates accounts you own — it never sits between you and your money.Ready to put a desk on your Mac?
Join the private beta and run every one of these features free while it lasts.
Join the beta