One pair specific line worth adding for EURUSD is a reopen rule based on the spread rather than the clock. A fixed five or fifteen minute timer treats every release the same, while the spread shows when liquidity has actually come back.
Say the normal London spread is 0.2 pips. After a big US print it jumps to 2.0 and takes six minutes to settle, so the rule is no new trade until it has stayed at 0.4 or less for three straight minutes.
The made up chart shows that case, with the first M1 candle spiking above the pre news range and closing back inside it. Twenty minutes after the release price breaks the same way, which fits the point about a close back inside carrying no information yet.
Under the spread rule the earliest entry here is minute 9, after the worst of the cost but well before the real break. Writing the threshold as double the normal spread keeps the rule usable on days when the base level is wider.
ECB days need two windows under the same rule, one for the statement and one for the press conference. The spread often widens again when the questions start.
News process: pair-specific notes for EURUSD
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: News process: pair-specific notes for EURUSD
Price on a tick chart often alternates between two values, the bid and the ask, even when nothing is happening in the market. That alternation is called bid-ask bounce, and it can fool indicators into seeing patterns.
Imagine a chart that plots every trade price. A buyer lifts the ask, then a seller hits the bid, then another buyer lifts the ask. The line zigzags, with a width equal to the spread. Any strategy that treats each zigzag as a swing will generate false signals.
The result is especially noticeable on pairs with wide spreads and on very short timeframes. A moving average crossing might fire repeatedly, simply because price is bouncing within the spread.
Solutions are straightforward. Chart mid-price, apply a spread-based filter, or require price to move several spreads beyond the signal level before acting.
The effect also inflates backtest results when the data ignores the spread, which makes testing on bid-only bars even riskier.
Using a mid-price chart is available as a setting on some platforms, and on others can be approximated by an indicator that averages bid and ask. Comparing signals on bid-only and mid charts for a week shows how many were created by the bounce.
Have you noticed a strategy produce suspiciously many signals on very short timeframes, and was bid-ask bounce the cause?
Imagine a chart that plots every trade price. A buyer lifts the ask, then a seller hits the bid, then another buyer lifts the ask. The line zigzags, with a width equal to the spread. Any strategy that treats each zigzag as a swing will generate false signals.
The result is especially noticeable on pairs with wide spreads and on very short timeframes. A moving average crossing might fire repeatedly, simply because price is bouncing within the spread.
Solutions are straightforward. Chart mid-price, apply a spread-based filter, or require price to move several spreads beyond the signal level before acting.
The effect also inflates backtest results when the data ignores the spread, which makes testing on bid-only bars even riskier.
Using a mid-price chart is available as a setting on some platforms, and on others can be approximated by an indicator that averages bid and ask. Comparing signals on bid-only and mid charts for a week shows how many were created by the bounce.
Have you noticed a strategy produce suspiciously many signals on very short timeframes, and was bid-ask bounce the cause?
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: News process: pair-specific notes for EURUSD
Correlation tables look reassuring. EURUSD and GBPUSD move together, AUDUSD and NZDUSD move together. Then a UK inflation number lands and GBPUSD heads in a direction EURUSD ignores entirely.
During pair-specific news, correlations can drop sharply or flip, and positions that looked diversified turn out to be unrelated bets. That cuts both ways. A hedge based on correlation might stop working, while a double position might suddenly behave like two independent trades.
Scalpers who rely on correlation for hedging or confirmation need to treat news windows differently. A rule like "ignore correlation signals for fifteen minutes around releases affecting either currency" prevents a lot of mistakes.
It also helps to check which currencies each upcoming release affects, since the trade's exposure may include currencies that aren't obvious from the pair name.
One quick check before each session: list the day's high-impact releases and write down which currencies each one touches. A UK release hits sterling pairs, a US release touches all the dollar pairs, and a Japanese one moves the yen crosses. Any open positions that share a currency with the release deserve a second look before the clock hits the minute.
Does correlation get rechecked before news, or assumed constant?
During pair-specific news, correlations can drop sharply or flip, and positions that looked diversified turn out to be unrelated bets. That cuts both ways. A hedge based on correlation might stop working, while a double position might suddenly behave like two independent trades.
Scalpers who rely on correlation for hedging or confirmation need to treat news windows differently. A rule like "ignore correlation signals for fifteen minutes around releases affecting either currency" prevents a lot of mistakes.
It also helps to check which currencies each upcoming release affects, since the trade's exposure may include currencies that aren't obvious from the pair name.
One quick check before each session: list the day's high-impact releases and write down which currencies each one touches. A UK release hits sterling pairs, a US release touches all the dollar pairs, and a Japanese one moves the yen crosses. Any open positions that share a currency with the release deserve a second look before the clock hits the minute.
Does correlation get rechecked before news, or assumed constant?