Skip to content
VectorhelmRISK-MANAGED AUTOMATION
Join the waitlist

The risk engine

Two layers of limits,
checked pre-trade

A notification after the account is gone is not risk management. The engine sits between your strategy and the venue, evaluates every order against the limits you set, and refuses the ones that breach them.

1.8 ms
median signal-to-order in internal testing
Two layers
per strategy, and a master guard per account
Pre-trade
checked before the order leaves, not after the fill

How the limits nest

The account has the last word. Each strategy has its own limits.

One account can run many strategies. Each carries its own caps, and all of them sit under one account-wide ceiling that can stop every route at once.

Account layer

The master guard

Set once per account and applied across everything running on it. When this ceiling is reached the engine stops opening on that account — every strategy, at once — and flattens if you have told it to.

  • Account daily loss
  • Account drawdown
  • Total exposure
  • Kill switch
Strategy layer

BTC Momentum

  • Daily loss
  • Position cap
  • Max drawdown

EURUSD Reversion

  • Daily loss
  • Position cap
  • Max drawdown

Grid · SOL

  • Daily loss
  • Position cap
  • Max drawdown

Per-strategy caps

Each running strategy carries its own limits, so an experimental one can be kept on a short leash while a proven one runs with more room. A strategy hitting its own cap pauses alone — the others keep running.

In the product

Both layers, on one screen.

The risk centre shows each strategy against its own caps and the account against its ceiling, at the same time. Headroom is the number that matters day to day — how much room is left before something stops.

When a limit is close the bar reads as warning; when there is room, it reads with no colour at all. Nothing about the display is decorative.

  • Per-strategy headroom, per-account headroom, one view
  • Change a limit while a strategy is running
  • One kill switch flattens everything on the account

Why in front

A limit checked after the order is a report.

The difference between the two is the whole product, and it only shows up on the day something goes wrong.

Checked after the fill

  • The order reaches the venue, then you are told it breached
  • The position exists while the alert is still in flight
  • A bad sequence completes before anything intervenes

Checked before it leaves

  • The order is refused and never reaches the venue
  • There is no window where the position exists unnoticed
  • The sequence stops at the cap, mid-sequence

How it behaves

What the engine does when a rule is hit.

The interesting part of a risk layer is not the happy path.

  • It refuses, it does not warn

    A breaching order is rejected before it reaches the venue. There is no state where the position exists and the alert is still in flight.

    Pre-trade

  • Manual orders are not exempt

    Discretionary tickets pass the same checks. The rulebook does not care who sent the order.

    Same rules

  • Every decision is written down

    What was refused and which rule fired. The event log is an audit trail, not a notification feed.

    Logged

  • Your funds never move

    Vectorhelm is non-custodial. Your money stays in the account you already hold, and you revoke access at the venue.

    Non-custodial

Breach a limit on purpose.

Set a daily cap on a demo account, push a strategy past it, and watch the engine refuse the order and halt the day. That is the only way to trust a risk layer.

Join the waitlist