calseda

Trading desks

See what a trade does to your risk before you make it.

Prices move every second, and your value at risk moves with them. Calseda keeps each portfolio's valuation and risk current the moment a price changes, and checks a proposed trade against them in private. Every figure shows the prices and rates behind it.

A demo on real public market data. No sign-in.

What you can't see today

Risk as it was at last night's close.

Risk at this morning's prices

Value at risk is calculated overnight. By the afternoon it describes a market that has moved.

A trade before it is booked

A trader checks a new position against the morning's figure, or not at all.

The busiest minute

Screens that queue every price fall behind in a sell-off, exactly when a trader needs them.

The spreadsheet in between

Traders paste figures into Excel and adjust them by hand. Nobody can trace the changes.

What you'd see instead

13:30, a sharp fall.

Today

  1. 07:00

    The overnight run finishes. A portfolio's value at risk is well inside its limit.

  2. 13:30

    Prices fall sharply. The risk report still shows the morning's figure.

  3. 13:40

    A trader adds to a position, checked against the 07:00 figure. It passes.

  4. Next day

    The report shows the portfolio over its limit. Risk asks the desk what happened.

With Calseda

  1. 13:30

    Each price change revalues the portfolios that hold it, and only those.

  2. 13:31

    Value at risk moves with the market, and the portfolio shows near its limit.

  3. 13:40

    Before trading, the trader checks the trade privately. It would take the portfolio over its limit.

  4. 13:41

    The trader resizes it, with the same live figures in front of the desk and risk.

An example with simulated data.

Valuations, current all day

Each portfolio revalued when one of its prices or rates changes, and only then.

Risk before the trade

A proposed trade is tested privately against the live portfolio. Nobody else sees it, and nothing is booked.

Calm at the busiest minute

When prices move faster than a screen can redraw, Calseda skips the values in between and sends the newest. Nothing builds a backlog.

Excel with live figures

Traders keep their spreadsheets. The cells hold live figures, and overrides go back through checks, logged with who posted them.

See it in a demo

Client portfolios valued live on real market data.

In a demo, a small trading desk values simulated client portfolios, each in its client's own currency, from real public prices of US equities and crypto, bitcoin options included. Each portfolio's value at risk is checked against a limit of 4% of its value. Try a trade before making it, or value a portfolio that doesn't exist.

A demo's pre-trade check showing a portfolio's value at risk before and after a trade A demo's pre-trade check showing a portfolio's value at risk before and after a trade
From a demo running on public data: a trade checked against a portfolio's value at risk before it is made. The clients' portfolios and their 4% limit are simulated, and second-by-second exchange rates move around the ECB's real reference rates.
A demo's what-if valuation of a portfolio that doesn't exist A demo's what-if valuation of a portfolio that doesn't exist
From a demo running on public data: a portfolio that doesn't exist, valued live from the same prices as the real ones. Only the page that asked sees it.

What changes for your desk

One set of current figures for the desk and for risk.

Fewer next-day surprises

A trade that would breach a limit is seen before it is made, while it can still be resized.

New questions without a project

“What would this portfolio be worth with half the bitcoin sold?” is a private question on live prices, from a browser or Excel.

What front-office risk gets

Every figure can show the prices behind it.

  • For each valuation as it stands, the exact prices and rates it was calculated from, with where each came from and how fresh it was.
  • A flag on any figure that rests on a price older than its source should allow, trading hours included.
  • How long a price change took to reach the figure.
  • Every override logged with who posted it, and when.

For your quants

Models in Java or Python, in the same small shape.

A model tried from a notebook against live data is a draft: it cannot touch production figures until someone moves it there. The detail is on For technical teams.

How it starts

With one book, next to what you run today.

  1. Pick one figure

    For example, value at risk for one desk's book.

  2. Connect the data you already have

    Positions and market data, read from the systems that already produce them. Nothing is replaced.

  3. Write the calculation

    Your quants, or ours, write it, in Java or Python. Calseda does the rest.

  4. Run it side by side

    It runs alongside the overnight run, so the desk and risk can compare the two before relying on it.

See your own book live

Become a design partner: a pilot on generous terms, in exchange for a case study published only with your approval.

For your colleagues: Trading desks, one page (PDF) and the executive briefing, 10 slides (PDF).