Multi-timeframe analysis

A chart interval is a choice about how much price data is compressed into one bar, and nothing else: the same trades produce a daily candle, an hourly candle and a five minute candle, aggregated three different ways. Reading two or three intervals together is a convention that treats the longer one as the setting and the shorter one as the resolution, on the reasoning that a move which looks like a trend on a short time frame is frequently a pullback inside a larger one.

The intervals chosen are conventionally separated by a factor of roughly four to six, so that the shorter chart holds several bars for each bar of the longer one and the two are visibly different rather than nearly the same picture. Practitioners disagree about the number of intervals worth using at once, and the argument against more than two or three is concrete: each additional chart raises the chance that at least one of them disagrees, and a rule requiring agreement across many intervals produces very few occasions on which they all align.

Two errors are specific to the technique. The first is treating agreement between intervals as independent evidence. It is the same price series resampled, so an indicator computed on two intervals is one measurement smoothed two ways, not two measurements. The second concerns the newest bar: on any interval it is incomplete until the period closes, so an indicator reading taken from it can change or reverse before the bar finishes, and a trend line or level drawn on a live weekly bar is provisional in a way the same drawing on a closed bar is not.