Product
One desk, five parts, no hidden levers.
Paper Desk runs on a computer next to the broker's own trading software. This site is its public face and its remote control. Here is each part and what it is allowed to do.
Scanner and gatekeeper
During market hours the desk looks at a short watchlist every few minutes. Each idea has to pass a fixed set of checks before it can become a paper order. The rules are plain and the same every time, so a stand-down is as explainable as a trade.
- Deterministic rules, no model in the loop
- An optional AI filter can only veto, never create or enlarge a trade
- Every verdict carries its reasons
The book
Sizing comes from a small software book, never the broker balance. The paper book starts every Monday at $2,000, sizing never uses more than $2,000, and it has a hard floor at $1,600. Hit the $60 daily loss cap and the desk takes no new trades until the next day; hit the floor and it stops until the Monday reset. Stock only, and no profit target.
- 1% of the book at risk per trade ($20 on $2,000)
- Limits enforced in code, not only shown on screen
- Self-review may only tighten a limit
The journal
Every scan, call and fill is written to a journal. A correction is logged as its own event with the old figure kept, never quietly overwritten. It is the one source of truth for the book, the record and the weekly reset.
- Closed trades only on the scoreboard
- A typed figure can't raise the book
- Exportable, checkable, boring on purpose
The swarm
Many small, token-free bots each watch for one kind of setup and vote. A single risk manager owns the order path. Bots that underperform are benched automatically and can only be put back by a person.
- Runs locally, no paid model calls
- One risk manager, one order path
- Benching is automatic; un-benching is human
The backtest gate
What a strategy has to prove before real money.
Real price history, estimated trading costs, walk-forward and out-of-sample. Then weeks of forward paper results on top. A strategy that passes earns the right to be considered, not a promise that it will make money. Backtests are hypothetical results with real limits; the disclosures explain them.
- Closed trades
- ≥ 60
- enough to mean something
- Profit factor
- ≥ 1.3
- gross wins over gross losses
- Expectancy
- ≥ 0.15R
- per trade, after estimated costs
- Max drawdown
- ≤ 8R
- worst run, in units of risk
Paper trading does not need the gate. Forward paper results are evidence gathered, not evidence skipped.
The dashboard
Every screen behind sign-in, and what it is for.
Dashboard
The desk right now: mode, last call, book and the figures that matter, each with a one-line explanation.
Paper record
Every closed paper trade, the weekly view, win rate, expectancy, profit factor and drawdown.
Scan log
What the gatekeeper saw and decided, scan by scan, with its reasons.
Risk limits
The rails as the desk enforces them. The site shows them; it does not change them.
Backtest gate
Whether a strategy has passed the tests needed before real money is even considered, and by how much it fell short if not.
Markets
Live prices for the watchlist, movers and news, and the desk against simply holding each fund.
Connect
A personal token that lets a desk report to its own dashboard and nothing else.
Controls
Start, stop and scan once from your phone. Never an order, never a limit.
Each account sees only its own desk. The site never places, changes or cancels an order. Sign in if you have a desk; otherwise the roadmap says when that opens up.
Want to know when it opens up?
One note per milestone. No weekly email, no selling of your address, and you can leave whenever you like.
By signing up you agree to get occasional emails from Paper Desk (flash.me.uk) about product news. Each email will say who sent it and include a way to unsubscribe, and you can ask to be removed at any time. See the Privacy Policy.