In May a thread on the NinjaTrader forum turned into the whole latency debate in twenty-two posts. It opened with a claim that traders blow funded evaluations in the first thirty seconds of the New York open because the chart shows “a ghost price from 200 milliseconds ago”. By post ten the author was pricing the fix: a custom execution thread that “usually runs between $5k to $15k” (EduardoT, NinjaTrader forum, May 2026). The room split immediately.
“No one is going to pay you $15,000 in this nt community forum,” wrote one regular (Maverick, NinjaTrader forum, May 2026). Another put the whole camp in one line: “Most people, probably 99.999999999% here don’t care or even notice” (WaleeTheRobot, NinjaTrader forum, May 2026). And then a quieter voice from the other side: “the information is fascinating and explains the slippage I’m always avoiding as best possible” (A_BeV, NinjaTrader forum, May 2026).
Everyone in that thread was right about their own trading and wrong about everyone else’s. That is the shape of every latency argument. In this article I’ll show you where half a second actually goes between your TradingView alert and your broker fill. Then a matrix by trading style that tells you whether it matters to you at all. Then the factor that costs more time than the network, which no faster server fixes. You’ll leave with a yes or a no for your own setup, and the numbers behind it.
The first camp says latency is overhyped. If your entry logic has a real statistical edge, half a second of slippage is noise inside your edge. This is the swing trader’s position and the pattern-system builder’s position, and their P&L supports it. The 99.999999999% line above is the same argument in forum dialect.
The second camp says the first is naive. A home internet connection typically sits 150 to 300 ms from a broker’s data centre. During NFP or CPI the price can move materially inside that window, before your order has even landed (For Traders, “Slippage impact on prop trading performance”). A scalper with a six-tick target watches that math eat a third of the trade. The forum member who has “used DOMs (purely for their speed/low latency) alongside charts since I began trading” is in this camp, and his P&L supports him too (brucerobinson, NinjaTrader forum, May 2026).
The debate never resolves because it is framed as “does latency matter” when the only answerable question is “does latency matter for this setup”. Nobody reframes it, because reframing it ends the argument. So here is the reframe. Most people selling low-latency execution are selling it to traders who don’t need it, and a few of the people dismissing it are the ones who do.
Two different numbers get called “latency”. Ping is how long a packet takes from your terminal to the broker’s server and back. From a VPS next to the broker that is 1 to 5 ms. From your home broadband to a broker three continents away it can be 200 ms or more. Execution time is what happens after the packet arrives: the broker’s server queues the order, checks liquidity, matches it, confirms. Calm markets, tens of milliseconds. A news release, and the same queue can hold for whole seconds. Neither number is the one that matters for your TradingView setup, because the alert-to-fill path has four legs, and the biggest one is not yours.
Those ranges come from our own production logs and they add up to about half a second from alert to fill under normal load. The number that matters in that list is the first one. When a vendor advertises sub-100 ms execution, they are measuring leg 2, and leg 2 is a quarter of the trip. If you are budgeting for latency, budget for the whole path, and notice that the part you can buy is the smallest part of it.
Here is the honest breakdown by trading style. Find your row, and read only your row if you’re in a hurry.
| Trading style | Impact of ~0.5 s | Why |
|---|---|---|
| Scalping, tight targets | Critical | A few ticks of slip is a large share of a small target. Consistent negative slippage can take a positive-expectancy system to breakeven; the arithmetic is simple. |
| News trading (NFP, CPI, FOMC) | Critical | Liquidity vanishes at the release and price moves inside the 150 to 300 ms window. Half a second late is the top of the impulse. |
| Copy trading, several accounts | High | Every delay widens the gap between master and copies. Over a month it becomes systematic drift, not noise. |
| Prop-firm evaluations | High | Not edge erosion, rule violation. One bad fill during news can push drawdown past a daily loss limit that the strategy never would have. |
| Intraday mean reversion | Moderate | Depends on the instrument’s pace. A calm EUR/USD Asian session, negligible. NASDAQ mid-morning, measurable. |
| Swing trading, multi-day | Negligible | Entry slip is statistical noise against a target measured in hundreds of ticks. |
| Position trading, multi-week | Irrelevant | If you are reading this to check whether you need low latency, you have your answer. |
If you are in the top four rows, keep reading. If you are in the bottom three, the rest of this article is about other people.
Traders arguing about latency argue about the network. The bigger factor sits inside your copier, and no VPS on earth fixes it for you.
Architecture costs more milliseconds than distance. A copier that processes webhooks one at a time queues alerts two and three behind alert one when they fire together, and that queue has nothing to do with ping. A copier that routes every order through a symbol-mapping table pays for the lookup on every trade. A copier hosted on the wrong continent from your broker adds hops that no setting removes.
The last point cuts against us, so it belongs here. Our routing sits in Frankfurt. For European brokers and most prop firms that is close to ideal. For a trader scalping E-mini futures against a Chicago-hosted broker, Frankfurt adds latency, and a copier hosted in Chicago would beat ours on leg 3 every time. We’d rather you know that from this page than from your fills. The other two points are design choices, and you can test any copier for them the same way. Fire three alerts within a second on a demo account and read the three timestamps in your log. If the second and third orders are late by more than the first, you have found a queue.
Latency is your problem if you scalp with tight targets, trade through releases, run copies across accounts, or trade a funded account with a daily loss limit that a single bad fill can breach. In those four cases the half second is real money, and the place to spend attention is the whole four-leg path, starting with the biggest leg you can’t buy and the architecture you can test.
It is not your problem if you hold for days or weeks. No execution speed you can buy will show up in your results over that horizon. And if you scalp US futures against a US-hosted broker, a Frankfurt-hosted copier, ours included, is the wrong tool.
Find your row in the matrix first. If yours isn’t one of the top four, the half second isn’t where your money goes.
If you are in the top four rows and route TradingView alerts into MT4, MT5 or NinjaTrader 8, the leg-2 numbers above are measured on Nordman Connector. The spread ceiling and disconnect alert that protect a funded account from the news row are described in TradingView Webhook Fails During News? Here’s Why.
Nordman Algorithms provides software infrastructure for trade automation and does not offer financial advice, trading signals, or managed trading services. This article is for informational and educational purposes only. Forum quotations are reproduced verbatim from public threads for illustration and do not constitute endorsement. Latency figures are internal measurements under normal load and will vary with your connection, broker and market conditions. Trading leveraged instruments such as Forex, CFDs and futures carries a high level of risk and may not be suitable for all investors. Only risk capital should be used. Full Risk Disclosure: https://www.nordman-algorithms.com/risk-disclosure/
