Missing bars and the indicators that quietly shift
Some feeds skip a minute when no tick arrives. A chart then has a gap, and indicators calculated over the last N bars suddenly span a longer period than intended.
A 20 period moving average that includes a gap of ten minutes covers a different stretch of time. Volatility measures, range filters, and momentum readings get distorted without any visible sign.
On liquid pairs during busy sessions, missing bars are rare. On exotic pairs and in thin periods they are common, and the effect on indicators can be significant.
Some platforms fill the gap with a flat bar at the previous close, others leave it empty. The difference alters indicator values and therefore signals.
Checking how the platform handles gaps, and whether the strategy trades pairs and times where gaps are frequent, avoids a lot of confusing behaviour.
A useful guard is computing indicators on time-indexed data, not on a fixed count of bars. A twenty minute window is then always twenty minutes, regardless of gaps.
Has an indicator ever misbehaved on a thin pair because of missing bars?
Missing bars and the indicators that quietly shift
Missing bars and the indicators that quietly shift
It’s Fairman 
-
FXS London Scalper
- Posts: 976
- Joined: Sun Oct 18, 2026 12:25 pm
Re: Missing bars and the indicators that quietly shift
A 20-period average spanning 25 minutes because of skipped bars is a quiet distortion. During the London lunch lull on crosses it happens more than people expect. Checking that the last N bars cover the expected time span before trusting the indicator is an easy guard in code.
FXS Official Team | London Session Desk
-
Amanhasnoname
- Posts: 2060
- Joined: Fri Oct 30, 2026 9:45 am
Re: Missing bars and the indicators that quietly shift
This bit me on NZDCHF a couple of years ago. I had a range filter that needed the last thirty one-minute bars to stay inside a band before it allowed a fade entry, and during the quiet hour after New York closed it kept firing on setups that looked nothing like what I had tested. When I finally lined the chart up against the tick export, those thirty bars were stretching across nearly fifty minutes because the feed simply printed nothing when no quote came in.
The fix I settled on was boring but it worked. Before the signal is allowed, the code compares the timestamp of the oldest bar in the window with the current time, and if the span is more than about twenty percent longer than it should be, the trade is skipped. It cost me a handful of entries a week and removed most of the strange ones.
I also stopped trading that cross outside London hours entirely, which made the whole issue mostly disappear.
The fix I settled on was boring but it worked. Before the signal is allowed, the code compares the timestamp of the oldest bar in the window with the current time, and if the span is more than about twenty percent longer than it should be, the trade is skipped. It cost me a handful of entries a week and removed most of the strange ones.
I also stopped trading that cross outside London hours entirely, which made the whole issue mostly disappear.