Much of this stuff is also being posted on my twitter account btw. twitter.com
Some also goes to stocktwits. Folks over on stock twits complained that the pattern I posted on the weekly XBI is in fact bearish, as it is in actuality a bearish flag, similar to the one that triggered in January. Target off of such a pattern would be the height of the decline into the flag from the breakdown point, or call it $35. I don't necessarily disagree except a flag pattern doesn't normally have an extension of range in back half and a similar pattern, bearish pennant cannot be drawn, because its wedge is getting wider to the right, not narrower. That being said, I do agree breaking the support area would be bad. I am not sure $35 is a real target, but breaking $50 on a weekly closing basis mind you quickly leads to $45 again, if not lower.
But the weekly candle patterns are extremely bullish so I decided to run just the last one - open near the low and close near the high. Running the pattern strictly, within 10 cents each way, yielded very few examples. So I expanded it to within 0.5% of each. That provides 33 examples over the 10 years XBI has traded, so 3 signals a year. This example is too broad in my mind but I will show the results and narrow it down.
This last candle is the third longest, gaining 7.22%. In July 2015 and near the low in March 2009, slightly higher one week gains were reported. In 2015, XBI plunged soon after. In 2009, it quickly gave up part of the gains but closed higher 13 weeks later by 2%. There were 4 other gains bigger than 5% in the candle. ALL returned more than 10% over the next 13 weeks ranging from 10-19%. The range at 8 weeks was 5-14%. At 4 weeks, -2% (sole loser) to 10%. At 1 week, 2 down and 2 up but flattish. Overall results for all samples:
1 week: down .5% vs. up 0.3% all cases (16 down, 15 up)
4 weeks: up 0.3% vs. up 1.2% all cases (17 down, 14 up)
8 weeks: up 3.6% vs. 2.4% all cases (7 down, 24 up)
13 weeks: up 5.5% vs. 3.8% all cases (5 down with 2 crashes, 26 up)
Excluding the 5 cases where Price is well above the 10 Week Average yields flat at 1 week to 7% at 13 weeks.
Restricting the data set to examples with 25 bps for close to high and open to low yields similar results with earlier numbers more negative.
Restricting the data set to 12.5 bps gaps reduces to 6 samples including this one. 4 of 5 were down in 1 week, 4 of 5 were down in 4 weeks, 2 of 5 were down in 8 weeks with all examples having large numbers, and all 5 were up at 13 weeks, 3 with large returns. Returns were -1%, -2%, 2% and 9%.
Jon |
