A bar tells you the extremes but not their order. A day that hit its high at 09:45 and its low at 15:50 produces an identical bar to one that did the reverse, and any strategy whose fills depend on which came first cannot be tested honestly on bars alone.
This is the source of a common and flattering bug: assuming both a stop and a target that were inside the same bar's range were reached in the order that favours you. The conservative convention is to assume the stop hit first, and to resolve genuinely ambiguous bars by dropping to tick-data.
Volume in the bar is also less simple than it looks. Consolidated volume includes off-exchange prints you could not have interacted with, so using it as an execution capacity estimate overstates what you can actually trade.
Related: tick-data, bar-types, fill-assumption, resampling