Skip to main content

Performance, Latency & Slippage

CrossTrade is built for speed, but end-to-end fill quality depends on more than just our turnaround time. This page covers what CrossTrade measures and controls, what it doesn't, and how VPS hosting, volatility, and broker response times contribute to the picture.

How fast is CrossTrade's execution?​

CrossTrade's average alert-to-NT8 turnaround is roughly 34ms. That number measures the time from our front door (when a webhook arrives at the CrossTrade endpoint) to the handoff into your NT8 add-on — which is the portion of the pipeline we own and can measure. We perform our part of the job as fast as possible and measure it; anything upstream (TradingView, your alert source) or downstream (NT8, broker) is outside that window.

What determines end-to-end latency for a trade?​

End-to-end latency is a chain, and CrossTrade is only one link. The main contributors are:

  • How fast TradingView (or your alert source) sends the webhook after the bar condition triggers.
  • CrossTrade's ~34ms alert-to-NT8 turnaround.
  • NT8's internal order processing.
  • Your connection to the broker and the broker's response time.

If there's latency in the overall flow, it's most often either on the alert-source side or the broker's end (slow response time) — not in CrossTrade's ~34ms window.

What causes slippage, and can it be avoided?​

Slippage is part of trading and can't be avoided. The primary factors are:

  • How quickly the alert is sent by TradingView.
  • Market volatility at the moment of execution.
  • Your broker's fill behavior and response speed.

You may also see small amounts of slippage on auto-flatten actions (e.g., hitting a max daily loss target a few dollars past the threshold) — that's expected with fast-moving markets and multiple moving parts.

Does running a VPS in Chicago reduce my latency?​

Yes, meaningfully. If the last network hop before your broker is in Chicago (where CME/Rithmic/most prop broker servers are co-located), your execution latency will be significantly reduced compared to running NT8 on a home PC over a residential connection. Latency from a CrossTrade Chicago VPS to CME/Rithmic/NT servers is sub-millisecond — around 0.5ms one way, or about 1ms round-trip.

All CrossTrade VPS sizes (Small, Medium, X-Large, etc.) ping the same, since they all sit in the same data center.

Besides speed, why else would I use a VPS?​

Two main reasons beyond latency:

  • Connection stability. Automated trading signals require your CrossTrade connection (and NT8) to stay online. A VPS runs 24/7 without your home network or PC being a single point of failure.
  • Execution latency. Co-location in the Chicago data center puts you a fraction of a millisecond from the exchange.

If neither of those matters to you — for example, if you only use the Trade Copier locally — running on your own PC is a personal preference.

What's the difference between bandwidth and latency on the VPS plans?​

They're two different things, and it's a common source of confusion:

  • Bandwidth is the maximum allowable data rate — the size of the pipe. "1 Gbps+" means you can burst up to and beyond that rate, subject to fair usage.
  • Latency (speed) is how quickly data actually moves through that pipe to its destination. From the CrossTrade VPS to the CME/NT8/Rithmic servers, that's under 1ms.

Bandwidth is the size of the pipe; latency is how fast the water moves through it.

What's the latency inside the Trade Copier's Order mode?​

Order mode latency is single-digit milliseconds. It all happens locally inside NT8. There may be a marginal increase with more accounts, because NT8 executes replicated orders sequentially — but the difference between 5 followers and 20 followers is small.

Why does Execution mode have a 1–2 second delay compared to Order mode?​

That's by design. In Execution mode, follower accounts only enter new positions after there's a confirmed fill on the leader — so the copier has to wait for NT8 to report the fill before replicating it. Order mode, by contrast, replicates the order instruction itself the moment NT8 sees it, which is why it's faster but behaves differently around pending/limit orders.

If you want Order mode's speed but only for market orders, that's essentially what Order mode already gives you.

How often does the Account Manager (NAM) poll, and is that the same as execution latency?​

NAM's polling interval is 2.5 seconds — that's how often it refreshes account P&L and evaluates auto-flatten conditions. This is a separate concern from order execution latency. Order execution (alerts, copier fills) runs on its own fast path; the 2.5s cadence only governs how quickly the Account Manager notices a P&L threshold has been breached and triggers a flatten action.

Small amounts of slippage past a max loss target (e.g., closing at -$450 when the threshold is -$400) are normal and reflect the combination of the polling cadence plus market movement between detection and fill.