cluster is getting tighter. nice. to remove the (obv?) noise, perhaps consider: a) a z-score filter, say 3 sd or b) colormap each value with its z-score, or c) just cut it off at 8 weeks
Topic: . - page 4. (Read 35863 times)
Quote
I actually do count the default minimum value for volume-per-bin as slowly decreasing. It is a line that is slowly decreasing based on a calibration at the $32 peak and a calibration at the $266 peak.
Yes, it does get messier the farther out you go. Dependencies that are easier to see in the actual data start to break down for the projected variables, and it starts looking less and less like 'natural' data. Going out as far as I am trying to go with this method is a real stretch.
do you think adding some more calibration points to the default min vol-per-bin line would affect things? the gox market has definitely changed a couple of times since $266, imo
i understand that you bin in volume, not time. you have made that point quite clear 
by "inverting values" do you mean take the absolute value = | d$/dt | ?
and what if trade vol is not a constant over time? lets say that it is decreasing slowly at some rate. in that case, counting trade vol bins would cause bigger errors as "volume required for reversal" or "time" of projection is increased. i guess you see that as it starts getting really messy the farther out in time the projection is.
by "inverting values" do you mean take the absolute value = | d$/dt | ?
and what if trade vol is not a constant over time? lets say that it is decreasing slowly at some rate. in that case, counting trade vol bins would cause bigger errors as "volume required for reversal" or "time" of projection is increased. i guess you see that as it starts getting really messy the farther out in time the projection is.
interesting stuff as always. my analysis highlights $125 if we go that way soon too.
some Q's if u dont mind. can you try to describe where you see noise entering into your model. or maybe just describe what are the largest contributors to variance?
is it possible that you have either 1) two overlapping distributions, or 2) 1 primary distribution with some artifactual noise?
some Q's if u dont mind. can you try to describe where you see noise entering into your model. or maybe just describe what are the largest contributors to variance?
is it possible that you have either 1) two overlapping distributions, or 2) 1 primary distribution with some artifactual noise?
sr. member
Activity: 294
Merit: 250
This bull will try to shake you off. Hold tight!
There is no risk for selling now, but there is a risk for holding. Selling everything is always a safe bet.
the risk you perceive for holding bitcoin is greatly exaggerated, and you fail to recognize the risk with holding everything in USD
lol
its true dude
bitcoin is happening now regardless of risk
USD is going down regardless of economic forecasts.
cash is king until its not.
bitcoin is gold! gold i tell you!
There is no risk for selling now, but there is a risk for holding. Selling everything is always a safe bet.
the risk you perceive for holding bitcoin is greatly exaggerated, and you fail to recognize the risk with holding everything in USD
lol
There is no risk for selling now, but there is a risk for holding. Selling everything is always a safe bet.
the risk you perceive for holding bitcoin is greatly exaggerated, and you fail to recognize the risk with holding everything in USD
There is no risk for selling now, but there is a risk for holding. Selling everything is always a safe bet.
It wasn't for the poor buggers who sold everything @ $30 before the $240 spike. Not that I am saying we are in the same situation now, but you never know...
Still a safe bet, any profit is a good thing. It's when you start stretching when you have a real problem.
There is no risk for selling now, but there is a risk for holding. Selling everything is always a safe bet.
It wasn't for the poor buggers who sold everything @ $30 before the $240 spike. Not that I am saying we are in the same situation now, but you never know...
A lesson that I learnt the hard way: you should never dump all.
Risk to hold outweighed risk of selling.
Assume I sold 250 BTC @ 138
If the price goes up to $170 I will probably buy back in and still have the same amount of $.
250 * 138 = $34,500
$34,500 / $170 = 202 BTC (same USD value)
At this point, I feel the risk is higher that the price is going to drop significantly. I think I'll be buying in somewhere around ~100, which will give me a gain of 100 BTC.
There is no risk for selling now, but there is a risk for holding. Selling everything is always a safe bet.
A lesson that I learnt the hard way: you should never dump all.
Lessing I learned the hard way: you should never go all-in.
Yeah, and that too
I am now trying to keep a balanced holding of BTC and $, with a slight bias towards one or the other, depending on the prevailing trend.
A lesson that I learnt the hard way: you should never dump all.
Lessing I learned the hard way: you should never go all-in.
Jump to: