Yours is arbitrary because it uses all candle wicks but one for the trendline. Either you do it by closes, or you do it by high/lows, not both.
Actually, not sure if that's a 'hard rule' of trendlines.
Anyway, the triangle looks fine to me. Broke out upwards. Don't know what porcupine's problem is with it... I think he just doubts that TA could ever dominate fundamentals, news, sentiment, etc.
If there is an actual glitch in the exchange data or it is due to some form of illiquidity (then again, you don't chart illiquid exchanges), then that's ok, but it's a bad idea to assume you can say which is valid and invalid, which is to be included and which ignored. It's the same thing with news, you don't have a framework of judging that, so might as well draw some dinosaurs.
I really don't think I know this as a 'hard rule'. Can you find me a quote on it?
The way I understand it, there are two ways to go about trendlines, in the most general sense: 1) not allowing any violation (in which case, only candle extrema are possible points of contact), 2) minimizing the no. of violations and only allowing candles that /closed/ above the trendline.
In the latter case, you can mix up extrema and closing points of candles. But like I said, I can be corrected on this.
EDIT: to be clear, you don't declare entire candles as "outliers". you just accept a small number of violations through wicks, and no violations through close.