News Trading Risks When the Signal Arrives Too Late

Sometimes a news signal arrives after the opportunity has already narrowed. The market has moved, the entry price has changed, and the trader is left trying to avoid a loss rather than capture the initial move. A signal can point in a sensible direction and still lead to a bad trade.

On 2 October 2026, two of our gold purchases closed at a loss: −1,395.10 USD net at IC Markets and −235.95 USD net at OANDA. We are showing the losing trades because they explain a risk that a page of profitable examples cannot. In both saved captures, the highest Bid recorded while the position was open remained below its purchase price.

News trading remains my main strategy and, in my own trading experience, the most profitable of the approaches I have used. That personal assessment does not make an individual signal reliable, remove losing periods or promise the same result to anyone else.

Two gold trades that closed at a loss

Both trades were Buy positions tagged News. The table preserves the broker-recorded prices and millisecond timestamps. Times are shown in the recorded broker/feed clock, not as UTC.

These are two selected positions around the same release window, not independent tests of two brokers. They differ in size, instrument specification, fill price and exit time. The larger monetary loss is not a broker performance ranking, and without the account equity it is not a percentage return or drawdown.

Explore the recorded bid and ask ticks

The solid line is Bid, the quote relevant to selling out of a Buy position. The dotted line is Ask, the quote relevant to buying. The BUY and EXIT markers show actual recorded fills, including fills that do not coincide with a captured quote. The horizontal entry line makes it easier to see whether the exit-side quote recovered to the purchase price.

Use Around entry to inspect the first seconds, While held to isolate the position's lifetime, or Full capture to restore the surrounding market. Hover over the chart or use the quote slider to inspect millisecond timestamps and the spread. The charts contain the saved individual quotes, not one-second averages.

During the IC Markets position, the highest saved Bid was 4227.79 at 15:30:04.006, which was 2.73 below the 4230.52 entry. During the OANDA position, the highest saved Bid was 4225.615 at 15:30:04.001, 1.285 below the 4226.900 entry. Neither saved series shows a gross break-even exit quote while its position was open; the IC Markets commission adds a further cost.

This is why looking only at the direction of the initial move can be misleading. A Buy entered after a sharp rise may need the market to rise further simply to recover the spread and any adverse execution. If the impulse fades, the trader can be left with very little room to exit. A short holding time does not make that exposure small.

A captured quote is an observation, not a promise that an order of a given size could have filled there. The recording is not a complete exchange order book. Lines connect saved observations for readability; missing observations are not reconstructed, and the chart cannot establish an executable price between ticks. Our research methodology for trade records and market data explains how we assemble evidence and distinguish recorded observations from conclusions.

Signal delay and execution delay are different risks

There are several stages between publication and a position: receiving the report, checking its values, evaluating the trigger, delivering the signal, submitting an order and obtaining a fill. Delay at any stage can change the available price. Faster order submission cannot recover an opportunity that disappeared before the signal arrived.

The execution telemetry for these two trades records 251 ms at IC Markets and 1,249 ms at OANDA, with 1 ms broker ping in each case. These are client-observed order-submission-to-confirmation intervals. They are not measurements of the time from publication to signal receipt, and ping is not a substitute for either interval. We explain the measurement boundary in Low-Latency News Trading.

These orders do not have populated signal-receipt or trigger-delay fields. A separate release log records US unemployment data at 12:30:02.801782 UTC and non-farm payrolls at 12:30:03.580082 UTC, relative to a stored 12:30:00 UTC release time. Those are log observations, not proof of which packet triggered either account or when that account received it. The broker/feed charts use a different clock domain; we do not overlay these signal times or infer an exact release-to-fill delay from them.

The trades therefore demonstrate the cost of the actual entries and the subsequent price path. They do not, by themselves, isolate how much of each loss came from signal delay, execution delay, spread, slippage or the exit rules. That distinction matters when deciding what to improve.

Our experience with US and Australian releases

In our experience with US releases since 2020, we have not found a signal supply we can rely on to arrive early enough for consistently good entries. This is an observation about the feeds, releases and execution paths we have used. It is not evidence that nobody anywhere can deliver faster data, or that signal quality was guaranteed before 2020.

As of 6 October 2026, the positive experience we are seeing this year is concentrated in our selected Australian releases. We currently focus on roughly two opportunities each month in that programme. That describes our selection, not the total number of Australian economic announcements. We have been pleased with signal quality in this sample, but two opportunities a month provide a limited basis for generalisation.

We continue developing improvements to our US signals. Until those improvements are demonstrated in real release and execution data, they are work in progress. A good Australian result does not make the next Australian trade safe, and it does not validate a US signal.

Why we still prefer news trading

Our aim in news trading is to interpret new economic information and take market risk at the prices available when we trade. The outcome depends on the report's surprise and revisions, trigger settings, signal timing, liquidity, broker execution, instrument, volume and exit management. Stop-loss, take-profit and trailing rules influence the result, but none can turn a poor entry into a guaranteed profit.

This differs from latency arbitrage that targets stale broker quotes. That approach seeks a price discrepancy before a slower quote catches up. Its viability depends on a particular broker or liquidity route continuing to offer the exploitable discrepancy. Finding such a setup, checking it and replacing it when conditions change can take substantial time.

Another approach places opposite positions on separate accounts before a release. The existence of a winning leg does not guarantee a positive combined result: the losing leg, both spreads, commissions and slippage still count. Opposite positions are not automatically abusive. The contractual problem arises when an arrangement is designed to exploit stale pricing, technical errors, bonuses or negative-balance protection rather than accept the combined market risk.

Broker terms matter. For example, IC Markets Global's order execution policy describes restrictions on latency arbitrage and abuse of negative-balance protection across accounts or client profiles, and remedies that can include cancelled orders or closed accounts. This illustrates a contractual risk; the applicable legal entity and account terms must be checked separately. It does not establish that either losing trade shown here was abusive or subject to such a remedy.

Trying to sustain an arrangement through accounts registered to other people adds another risk: control of the account and money may depend on someone else. Identity and source-of-funds checks, deposit and withdrawal arrangements, and loss of contact can become as important as the trading result. We do not consider that a sound foundation for our strategy.

We also would not build a business around remaining undetected. Brokers can analyse execution patterns across accounts; there is no sound basis here for assigning an AI detection rate or claiming that a fixed percentage of profits will be cancelled. Whether a broker profits from other clients does not create a safe allowance for prohibited trading.

For our news strategy, we can evaluate a broader set of brokers on measured execution, costs and their actual rules, instead of requiring a rare stale-price opportunity. That is an operational advantage in our experience, not a claim that a particular percentage of brokers will accept or execute every news strategy well. News trading can still face restrictions, rejected orders and execution disputes.

What these losses change about risk management

Before a release, the useful question is how much a bad fill and a fast reversal could cost at the chosen size. For these gold contracts, the recorded contract size was 100 per lot. The IC Markets price loss is consistent with (4210.66 − 4230.52) × 0.70 × 100 = −1,390.20 USD, before commission. OANDA's is (4211.170 − 4226.900) × 0.15 × 100 = −235.95 USD. Instrument specifications must be checked before applying that calculation elsewhere.

Exit rules also have execution risk. OANDA's risk disclosure explains that news can produce fast market moves and that a non-guaranteed stop does not protect against price gaps. Its spread guidance also notes that news-related widening can trigger stops or margin closeouts. A stop-loss is a risk-control instruction, not a guarantee of its requested execution price.

In our own reviews, we need to evaluate the full chain: the signal we actually received, the trigger that fired, the quote at submission, the fill and the way the trade exited. A strategy should also be able to skip a release when the signal is stale, the spread is unacceptable or the available price no longer fits its risk limits. That can mean missing a profitable move; it can also prevent chasing an opportunity that has already gone.

We continue with news trading because of our own results and the way we want to operate. These losses belong in that assessment alongside the winners. Every market and every signal carries a risk of loss. Profit is never guaranteed.