Most of us are probably aware that candle patterns can look very different on different time frames; for example a bull trend on a 5 minute chart may show up as a bear trend on a minute chart or vice versa. However, it is also possible that candle patterns may look very different on the same time frame if the data is slightly offset. For example, if your five minute patterns are on the 0-5/5-10/10-15 etc minute bars, they may tell a very different story than those on the 2-7/7-12/12-17 etc bars. I discovered this by accident (though it is obvious with a moment’s thought) when backtesting a strategy based on five minute bars, because the five minute bars of my program did not match the 5 minute bars of my broker package (in this case IG). Now this may not matter too much in practice, as at the tick level the market is the market, but if you are basing trading decisions on the actual bars, to me there might be a problem. For example, some strategies (one by Al Brooks happens to be the one I am testing) suggest putting an entry order just above the top of the previous bar, with a stop just below the previous bar in certain circumstances. However, these values might be very different if the bars are offset by a minute or even by just a few seconds. Moreover, he seems to suggest that there it is very important to wait for the end of the bar, as, for example, a bull bar may turn into a bear bar in the last few seconds of the bar. While I agree that may well be the case, does it make sense to base your stop and entry points so formally on something that might change again within a few seconds? I know you have to put your stops and entries somewhere, but this seems to make it sound less arbitrary than it actually is. As a further point, presumably different brokers, may have slightly different cutoffs for their various bars (IG five minute bars seem to be around the 00/05/10 minute marks, but by no means exact) so this too would also affect your strategy. Anyway, I just wondered if anyone else had thought about this, or if I am missing something.
The potential snag is pareidolia. The classic example is looking at a cloud and seeing an elephant. Pareidolia - Wikipedia If you can do proper backtesting of a given pattern and determine it provides a positive expectancy return, then go for it. Otherwise, don't trade the pattern, because of random walk (yes, I know I'm setting myself up for the naysayers.)
You got to realize "candles" are very course representations of higher frequency data that throw away a lot of information. They can show something real (e.g. a dragonfly could every well indicate a turning point in a dip) but you generally got to zoom in plus apply other context to make use of that. When it comes to trading solely on candle patterns, do realize an immense amount of computing power actively searches for obvious edges in raw price data, hence making it highly unlikely you will find anything with significant alpha. Feel free to try to repeat the search but IMO you can spend that time better elsewhere. So to summarize, I do think learning candles is useful for a visual aid for identifying potentially interesting situations while you're glancing at charts, but it's pretty useless for making final trade decisions.
Candlesticks etc. are not the DNA or the fingerprints or the chemical analysis of what the market has been doing or will do next. They are a reasonable guide to your decision-making but that's it. They don't predict what the market will do, only what you wish to do about it.
%% Tmorton; ,exactly + charts are just data with a picture. Bjoker ; exactly + thats one of several reasons many consider 5 minutes noise. Its a common problem + noted in some trading books. I remember Don Bright Daytrading preferred bid /ask............................................................................{Personally i some times do not record@ end of day settlement price\because i seldom trade @ last minute anyway. AS a % that end of day price/settlement difference is not much % difference, not much ;\like a 5 minute error can be + is.}