I came across an interesting video by Rob Mitchell (link here). He confirmed something that I have felt for a long time on reading too much into intraday bar patterns.
You can find many resources or educators that tell you things such as
- Pay attention to what the individual bars tell you (e.g. where the close is in relation to the bar range)
- Get confirmation of a signal when the close of a bar is in the top 1/4 of the bar
- Reading individual candlestick patterns work just as well for intraday bars
From my own intraday trading experience, I have found that those advice above don’t work. Logically thinking, I don’t see why the points above will work in the first place. Think about it, the close of an intraday bar is essentially a sampling at periodic intervals of a whole stream of transactions. When you start your sampling process can totally change the interpretation of any single bar. It is very common for two people to have slightly different charts, even from the same broker, depending on how the data is pulled or delivered. If one guy looking at his ‘samples’ can find a ‘hammer’ and another guy looking at his ‘samples’ of the same stream can find a ‘doji’ at the same time spot, it does not make sense at all to apply individual bar interpretation to intraday bars.
Note that if you are looking at daily or above bars, the situation is different because the close of a daily bar conveys much more information as the big boys fight to the close the market day.
So I agree with the conclusion from Rob Mitchell’s study (see below). To address some of these issues, he went on create a new chart type called Ultimate Tick Bars that appears to combine the benefits of both range bars and tick/volume bars. Someone on Big Mike forum tested them and posted a good review. Pretty interesting stuff.
Effect of Chart Type of Pattern Identification
- The same pattern (in substance), happening at two different times, will not look the same when it is shown on the chart. There is an element of randomness that is introduced, which is also affected by what chart type and parameters are used.
- A simple study measured what percentage of bars closed at a directional extreme (i.e. close of the bar is at the high of the bar, or 1 tick below the high of the bar) in the same direction as the trend (a stretch of prices from low to high).
- Using 510 volume bars, 32.20% of bars.
- Using 295 tick bars, 25.00% of bars.
- Using 132 second bars, 11.76% of bars.
- Using 10 tick range bars, 72.50% of bars.
Implications of the Study
- The close of an individual bar in the intraday timeframe is basically random, it has no directional bias.
- Any indicator that computes on the close of a bar, or a candlestick pattern that is based on the close of the bar, will be very inconsistent over time.