Wall Observer BTC/USD - Bitcoin price movement tracking & discussion

[Erdogan]

626 is 1 12 hr high. Is it the ignition?

[oda.krell]

more useful than coindesk's latest article?  
http://www.coindesk.com/trend-spotting-identify-trends-bitcoin-price-charts/

One of the things that make me skeptic of Technical Analysis (TA) is their general reliance on high and low points to determine trends and such.

Highs and lows are very noisy statistics, that may depend on a single decision by a single trader.  So much so, that even for exchanges that track each other very closely  (like Huobi and OKCoin), trend lines that connect highs or lows may be very different depending on which chart one uses -- even increasing/converging on one chart, while decreasing/diverging in the other.

Basic statistics says that the weighted mean a much more reliable parameter than high, low, open, or close; and a weighted least-squares-fitted line is much better than a line connecting any of those points.  To quantify the size of deviations from the mean or the fitted line, the standard deviation is much better than the high-low difference.

Median (and averages in general) are used in TA, but you are right, a lot of trendlines are largely based on extrema. I disagree however with your claim that the latter are too noisy to be useful: you seem to apply the insight from other fields were a noisy, sparsely populated parameter is best avoided. Two arguments why this insight can't just be applied 1:1 to trading: a) price extrema are the result of trades "eating" through the order book first, i.e. they are the conclusion of a longer, motivated process, not a randomly generated "blip", b) the knowledge what the exact maximum/minimum price some human (or his algo) was willing to trade for is not irrelevant or meaningless at all. I understand that from a purely statistical point of view, your argument makes sense, but taking into account that we are trying to model human behavior, not, say, bacterial growth, not everything can be carried over as straightforward, I believe.

[aminorex]

Highs and lows are very noisy statistics, that may depend on a single decision by a single trader.  So much so, that even for exchanges that track each other very closely  (like Huobi and OKCoin), trend lines that connect highs or lows may be very different depending on which chart one uses -- even increasing/converging on one chart, while decreasing/diverging in the other.

Basic statistics says that the weighted mean a much more reliable parameter than high, low, open, or close; and a weighted least-squares-fitted line is much better than a line connecting any of those points.  To quantify the size of deviations from the mean or the fitted line, the standard deviation is much better than the high-low difference.

Moving averages and z scores are t.a. indicators as well.  You are describing features.  Your job is to find a utile way to combine features to compose control decisions.  Some features are more naively utile.

[edwardspitz]

more useful than coindesk's latest article?  
http://www.coindesk.com/trend-spotting-identify-trends-bitcoin-price-charts/

One of the things that make me skeptic of Technical Analysis (TA) is their general reliance on high and low points to determine trends and such.

Highs and lows are very noisy statistics, that may depend on a single decision by a single trader.  So much so, that even for exchanges that track each other very closely  (like Huobi and OKCoin), trend lines that connect highs or lows may be very different depending on which chart one uses -- even increasing/converging on one chart, while decreasing/diverging in the other.

Basic statistics says that the weighted mean a much more reliable parameter than high, low, open, or close; and a weighted least-squares-fitted line is much better than a line connecting any of those points.  To quantify the size of deviations from the mean or the fitted line, the standard deviation is much better than the high-low difference.

Fair point.

That is where experience and judgment come into play, also seeing TA for what it is, an indicator to aid judgement calls - it really would not be possible to have a market at all if there was an indicator that was right 100% of the time.

Maybe another good reason for you to start trading BTC Jorge I agree with what Empowering says. I don't know much about statistics, but it is useful to consider that the market is moved by two opposite forces (bulls and bears). When price fails to move below a certain point it is because bears have run out of steam. And when price fails to go higher is is because bulls have run out of steam. It doesn't matter if it is a single player who prevents further movement in either direction. All we can say is that price failed to go beyond that point and that is always significant. So maybe it is better to think about the market and TA as forces in physics instead of statistics. If price fails to go further in one direction it will bounce and move in the opposite direction. Like physical objects bouncing off each other when colliding.

It is not enough to consider one exchange alone, but since they all go together you can still use TA from one exchange as a general indicator for market direction.

[HarHarHar9965]

It looks to me like Huobi is forming a nice base at the 50% retrace level off the recent top. Bears can't get any momentum going.... launch soon? Well, I hope so.

[empowering]

[ChartBuddy]

Explanation

[edwardspitz]

On the topic of TA and indicators does any of you use the CMF indicator also found on Bitcoinwisdom and Bitcoincharts?

[ShroomsKit]

It would be nice if we could make this place less about JorgeIdiot and more about Bitcoin.
Eventhough i have him on ignore his bullshit still fills most of the page.
I wish that guy would get a life. But i guess this is what narcissistic sociopaths do.

[theonewhowaskazu]

On the topic of TA and indicators does any of you use the CMF indicator also found on Bitcoinwisdom and Bitcoincharts?

CMF indicator?

I was disappointed. I thought you may have made a CCMF indicator.

[edwardspitz]

On the topic of TA and indicators does any of you use the CMF indicator also found on Bitcoinwisdom and Bitcoincharts?

CMF indicator?

I was disappointed. I thought you may have made a CCMF indicator.

LOL. The CCMF indicator is very good, but usually only works after the fact 

I can't make much sense of the CMF indicator. Maybe it is useful for longer time frames and for divergences?!

[aminorex]

LOL. The CCMF indicator is very good, but usually only works after the fact 

Monkey recommends commodity channel money flow used with mixed moving correlation stochastics.

[BTCfan1]

is the dumping convention still in town or what?

[Chef Ramsay]

is the dumping convention still in town or what?

[Rub3n]

On the 25th? 630$

[edwardspitz]

LOL. The CCMF indicator is very good, but usually only works after the fact 

Monkey recommends commodity channel money flow used with mixed moving correlation stochastics.

I have not heard of moving correlation stochastics before I will have to look into that.

[ShroomsKit]

is the dumping convention still in town or what?

We went up 3 dollars so people have to take their 20 bucks profit you know.

[ChartBuddy]

Explanation

[dropt]

We went up 3 dollars so people have to take their 20 bucks profit you know.

I was just reading some garbage over on reddit that a self-described "non-professional" market trader with 25 years of experience posted.  It blew his mind at how the people trading this market had no idea what they were doing.  One point in specific, he noticed that people continue to want the price to crash so that they can "buy back in".  That, coupled with the apparent inability for people to let their winners run caused him to speculate that if/when Wall Street level traders do show up, that the "day tradurrz" of BTC right now are going to get chewed the fuck up.

I can't wait.

[adamstgBit]