Now's as a good a time as any to point out a VERY common mistatke, particularly among quant type analysts. Just because a program finds a pattern doesn't mean the pattern is valid. Put another way, the pattern doesn't mean anything. For any given data set, some appropriate algorithm will find some random pattern. BEFORE you look for a pattern, you need to examine the market logic that would dictate the pattern. For example, my philosophy is that all price movement is based on pure supply and demand. So, the first thing I do isn't looking for random patterns, I first figure out what should happen based on logic. Then I look for a pattern to support that logic. Then I run the normal gammit of statistical studies to see if it's real or illusion.
The fact is that in Crude Light we had (in last 400 days) 8 patterns well established in theory before, and all concluded with a bearish day. We call this 100% of pattern concluding bearish. We cannot define this in a different way. Now the nineth pattern is going to confirm. If the closing will be be below 62.62 we have nine patterns out of nine that was followed by a bearish day. If today the closing will be above 62.62 we will have 8 patterns bearish and one bullish: how can we define in a percent way the situation? Second argument. The markets are not normally distributed. The tails are high and the real model is not clear. And, a fortiori, we demonstrated with a random chart that, also if markets were normally distributed, Head and Shoulder and other classics can be perfectly random. If you will or not. To examine the market logic that dictates the pattern can be very useful, but it seems that 8 out of 8 are becoming 9 out of 9. If so, when the pattern will exhibit itself for the tenth time, what to do? Go long? It is foolish. Moreover: we have not well understood why this pattern is ending in a bearish way, nor we are concerned with. This is philosophy. We are in a trading forum and we should help heach other with suggestions. We think that today our suggestion was good. Or at least the intention was good. We are not here for other secret purposes. Excuse the english that is not our mother language. Hasta la vista.
Your entire argument is falacious (excuse spelling) due to one thing. As others have pointed out, having only 8 or or even a dozen instances to support your hypothesis is NOT statistically significant. This isn't my opinion, it basic statistics. I am delighted for you to have what appears an interesting pattern, but it's just that until the data set of occurences gets to be into the double digits, preferably at least 30 to 40 occurences. Best of luck to you in this
This is instead true and we agree perfectly. But consider please that this is the true double hedging sword. As you know, we limit to last 400 days the observations, due to the Chi square problem (old data are meaningless...) so we shall consider that in our reality eight patterns are already a lot, also if they teached us that 30 or 40 occurences are needed and so on. But really 30 occurences in 400 days will not succeed, Remember moreover that we ar covering ourselves with stops. Another thing, le cas echeant, in this particular case, WE NOT OPENED THE BEARISH POSITION BECAUSE THE FIRST PRICE WAS 61.95, BELOW THE RANGE PRE-ESTABLISHED. So you can see that we will use other instruments to protect ourselves. The range pre established was 64.082 | 62.082 and it opened instead at 61.95. (at 10:40 am New York time, other safety) Last but not least: the suggestion is to close at 2 pm New York time, to avoid sudden reactions. Per aspera ad astra.