Latency is the number people ask about. It is rarely the one that costs them.

Where the clock starts and stops

A copy is a chain of steps, and a latency figure is only meaningful if it names the two ends.

The honest span starts when the leader's order fills and ends when the follower's order is sent to the broker. On Quanify that is usually 50 to 200 milliseconds. Most of it is the broker's round trip: the broker reporting the leader's fill, and the broker accepting the follower's order. Quanify's own step, from receiving the execution report to sending the follower order, takes microseconds.

A figure that starts later, for example when the copier received the execution report, will look better without being faster. A figure that ends earlier, for example when the copier decided to send, leaves out the part you actually wait for. Ask which span you are being quoted.

No copier is faster than the broker

The follower's order cannot exist before the leader's fill is reported, and it cannot fill before the broker accepts it. Both of those are broker time. A copier can add delay to that, through slow code, a queue, or a desktop machine that was asleep, but it cannot remove any.

So a claim to be faster than the broker is not a claim to take seriously, and comparisons between copiers on the same broker come down to how little each one adds. Once a copier's own step is down in microseconds, the remaining differences are mostly where it runs and how its connection to the broker is held.

What 100 milliseconds costs in price

In a quiet market, price often does not trade a tick in a tenth of a second. In a fast one, several ticks can trade in that time. The follower's market order fills at whatever is there when it arrives, so the gap between the leader's fill and the follower's fill is mostly a function of how fast the market was moving, not of the copier.

Put a number on it. An MNQ tick is worth $0.50 a contract. A one-tick difference on ten MNQ is $5. On NQ, a one-tick difference on two contracts is $10. Those are real costs and they add up over a month, but they are small next to a follower that trades the wrong size or keeps a position the leader closed.

Measure it over a week of trades rather than one. Single fills are noisy; the average difference between leader and follower prices tells you whether there is a cost worth chasing. Slippage in futures trading covers where the rest of that gap comes from.

The failures that cost more than milliseconds

These are the questions that separate copiers. None of them shows up in a latency figure.

Questions to ask any copier

Six questions get you further than a latency figure.

1. 01What does your latency number measure? Where does the clock start and where does it stop.
2. 02What happens on a partial fill? Does the follower get the filled quantity or the order quantity.
3. 03How are duplicates detected? Per execution, per account, inside what window.
4. 04What happens to a fill that arrives late? After a disconnect, is it copied, refused, or reported.
5. 05Where does it run? And what happens to copies when your own machine is off.
6. 06What happens to open positions when you stop it? Quanify does not close positions on your behalf: new orders stop, open positions stay at the broker.

What you can actually control

You cannot shorten the broker's round trip. You can choose when that round trip matters.

Trade when the book is deep. The same copy costs less in the middle of the session than in the first minutes after the open or around a scheduled release. Keep follower sizes modest relative to what rests at each price, so a follower's market order does not walk the book on its own. And on TopstepX followers, remember that Quanify sends market orders only, so each copy takes the price that is there.

Then check the things that matter more than speed: that every follower is armed, that every multiplier is right, and that a flatten on the leader leaves every follower flat. Run it on demo or paper accounts first; the configuration stays the same when you move it to funded.