Advertisement IC Markets

Candle patterns depend on when the bar starts

Share, develop, and backtest custom MQL4/MQL5 Expert Advisors, Python data-scraping scripts, trading bots, and automated market alert systems.
Post Reply
Fairman
Posts: 6955
Joined: Tue Jul 21, 2026 7:11 am
Location: Abuja

Candle patterns depend on when the bar starts

Post by Fairman »

Candle patterns depend on when the bar starts

A famous reversal pattern appears on a five minute chart on one platform. On another, set to a different server time zone, the same market produces a completely different set of candles.

The reason is that bars begin at fixed clock times. A five minute bar starting at 10:00 contains different ticks from one starting at 10:02 on a platform with an offset. Every pattern built from those bars is therefore a function of an arbitrary starting point.

On the daily chart, the problem is bigger. A broker that closes the day at 5pm New York produces different daily candles from one that closes at midnight server time. A pattern on the daily chart in one place may not exist on another.

This raises uncomfortable questions about patterns that rely on exact candle shapes. If shifting the clock by a minute or two makes the pattern disappear, how much of it was ever real?

One test is shifting the chart's offset by a few minutes and checking whether the strategy still works. A robust edge survives. A fragile one often vanishes.

This also touches on time zones for session-based patterns. A pattern defined by the first hour after the London open uses a clock-based definition, and shifting server time by an hour changes which ticks belong to that hour. Defining the window relative to UTC or an exchange clock keeps it stable.

Does the strategy hold up when the bar start times are shifted?
It’s Fairman :geek:
Recommended broker for automated trading & scalping IC Markets
FXS London Scalper
Posts: 976
Joined: Sun Oct 18, 2026 12:25 pm

Re: Candle patterns depend on when the bar starts

Post by FXS London Scalper »

Fairman wrote: Candle patterns depend on when the bar starts A famous reversal pattern appears on a five minute chart on one platform.
The same market producing different five-minute candles on two servers is a strong argument against relying on exact candle patterns. Patterns that survive a change in bar alignment, for example testing on bars shifted by one minute, are more likely to be real.
FXS Official Team | London Session Desk
Fairman
Posts: 6955
Joined: Tue Jul 21, 2026 7:11 am
Location: Abuja

Re: Candle patterns depend on when the bar starts

Post by Fairman »

A one-minute offset test is a useful robustness check, but I would keep the raw price sequence fixed and change only the bar boundaries. That separates sensitivity to aggregation from differences in feeds, spreads, or missing ticks. I would also predefine a small set of offsets—say the chart’s native alignment and a few nearby shifts—rather than search many alignments and report whichever performs best. The strongest-looking candle can move simply because a high or low falls on the other side of a boundary; that does not make the underlying move disappear, but it can make a rule based on the candle shape unstable.

For a session rule, define the clock explicitly, including the time zone and daylight-saving treatment. “First hour after London open” should refer to a time-zone-aware market event, not a broker’s current server clock, otherwise the window can drift when the server changes offset. Compare the strategy’s net results and trade count across those fixed alignments, with identical costs and execution assumptions. If the edge exists only at one arbitrary bar start, that is a warning; if it survives several sensible definitions, that is encouraging but still needs out-of-sample validation.
It’s Fairman :geek:
Amanhasnoname
Posts: 2060
Joined: Fri Oct 30, 2026 9:45 am

Re: Candle patterns depend on when the bar starts

Post by Amanhasnoname »

I noticed this when I compared the same EURUSD five-minute session on two brokers with different server offsets. A clean engulfing on one chart was a messy pair of bars on the other, built from the same ticks with a two-minute shift in bar starts. The market had not changed; the clock boundaries had.

That pushed me to test candle-shape rules with the raw price sequence fixed and only the bar start shifted by a minute or two. Patterns that vanished under a small offset left my playbook. Session rules I now define against UTC or the exchange open, not the broker's server clock.

Daily charts are even more sensitive. A New York afternoon close and a midnight server close produce different daily candles on the same pair, so a pattern that looks obvious on one platform may not exist on another. If an edge only survives at one arbitrary alignment, I treat it as fragile.
Post Reply