What ticks arrive on weekends and holidays
Some brokers publish quotes when the market is technically closed, such as a few ticks on Saturday or during a holiday. These indicative prices are often stale or placeholder values.
In a chart they show up as tiny bars or odd spikes at strange times. In a backtest, they can trigger signals or fills that could never occur in practice.
Cleaning them is generally straightforward. Remove data outside the real trading hours, usually from Friday evening to Sunday evening New York time, and check holiday dates for thin sessions.
The risk is more serious when data from different time zones is merged, or when daylight saving shifts push the weekend boundary by an hour, leaving ghost bars at the start of the week.
A look at the first and last bars of each week in the dataset exposes many of these issues quickly.
Brokers also sometimes put out one or two quotes at the Sunday open before the real market starts, with spreads that look absurd. Those ticks belong in the bin for backtesting purposes.
What oddities have turned up in the weekend or holiday data when it was inspected?
What ticks arrive on weekends and holidays
What ticks arrive on weekends and holidays
It’s Fairman 
-
Amanhasnoname
- Posts: 2060
- Joined: Fri Oct 30, 2026 9:45 am
Re: What ticks arrive on weekends and holidays
The worst one I found was a single Sunday tick on AUDUSD that sat about forty pips away from where the market actually opened. My breakout test happily treated it as a range break, entered, and then recorded the rest of the gap as a profit. One phantom trade was making up a surprising share of the annual result, which is why I started looking at the first bar of every week.
Holidays were sneakier. Around Christmas and New Year the data had real ticks, but so few of them that the bars looked like flat lines followed by sudden jumps. Nothing was wrong with the data as such, yet no sensible trader would have been active, so I now exclude a short list of thin days entirely.
The daylight saving issue hit me when I merged two sources, one on server time and one on UTC. For three weeks a year I had an extra hour of empty bars on Monday morning, and my session filter was quietly offset.
Holidays were sneakier. Around Christmas and New Year the data had real ticks, but so few of them that the bars looked like flat lines followed by sudden jumps. Nothing was wrong with the data as such, yet no sensible trader would have been active, so I now exclude a short list of thin days entirely.
The daylight saving issue hit me when I merged two sources, one on server time and one on UTC. For three weeks a year I had an extra hour of empty bars on Monday morning, and my session filter was quietly offset.
-
FXS London Scalper
- Posts: 976
- Joined: Sun Oct 18, 2026 12:25 pm
Re: What ticks arrive on weekends and holidays
Filtering weekend and holiday ticks from backtest data avoids false signals. These ticks are usually indicative quotes rather than tradeable prices, and a single one can trigger an entry or a stop in the tester that would never have happened live.
FXS Official Team | London Session Desk
Re: What ticks arrive on weekends and holidays
The stop side is the one that gets overlooked. Everyone thinks about phantom entries, but a stray Sunday tick can just as easily close a real position in the tester. Picture a long carried over the weekend in a backtest with a 25-pip stop. If one indicative quote prints 30 pips lower before the open and the real first price is only 5 pips down, the tester books a full -1R loss that could never have happened live. Get that a few times a year and the drawdown figures are wrong as well as the returns.FXS London Scalper wrote: Fri Nov 13, 2026 3:12 pmFiltering weekend and holiday ticks from backtest data avoids false signals. These ticks are usually indicative quotes rather than tradeable prices, and a single one can trigger an entry or a stop in the tester that would never have happened live.
A quick filter that catches most of it: drop ticks between the Friday close and the Sunday open, roughly 21:00 or 22:00 UTC depending on daylight saving, then flag the first five minutes of each week where the spread is more than three times its median. Check those bars by hand rather than deleting them blindly.
Holidays need their own list. Those ticks are real, just thin, so a time filter alone won't catch them.
It’s Fairman 