aminorex
Legendary
Offline
Activity: 1596
Merit: 1029
Sine secretum non libertas
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 14, 2014, 08:58:14 PM |
|
On a log chart, the 90 trend base is fair dinkum. On a log(log(x)) chart it is not. The godmodetrader.de analysis is all predicated on the predictive quality of the log(x) chart. If it is not credible, then the obvious alternative appears to be that the log(log(x)) chart is the correct one for analysis.
|
Give a man a fish and he eats for a day. Give a man a Poisson distribution and he eats at random times independent of one another, at a constant known rate.
|
|
|
segeln
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 14, 2014, 09:04:04 PM |
|
On a log chart, the 90 trend base is fair dinkum. On a log(log(x)) chart it is not. The godmodetrader.de analysis is all predicated on the predictive quality of the log(x) chart. If it is not credible, then the obvious alternative appears to be that the log(log(x)) chart is the correct one for analysis.
could you display the Chart in log(log(x) please,so we can discuss it
|
|
|
|
aminorex
Legendary
Offline
Activity: 1596
Merit: 1029
Sine secretum non libertas
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 14, 2014, 09:06:51 PM |
|
On a log chart, the 90 trend base is fair dinkum. On a log(log(x)) chart it is not. The godmodetrader.de analysis is all predicated on the predictive quality of the log(x) chart. If it is not credible, then the obvious alternative appears to be that the log(log(x)) chart is the correct one for analysis.
could you display the Chart in log(log(x) please,so we can discuss it From another thread:
|
Give a man a fish and he eats for a day. Give a man a Poisson distribution and he eats at random times independent of one another, at a constant known rate.
|
|
|
segeln
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 14, 2014, 09:17:53 PM |
|
On a log chart, the 90 trend base is fair dinkum. On a log(log(x)) chart it is not. The godmodetrader.de analysis is all predicated on the predictive quality of the log(x) chart. If it is not credible, then the obvious alternative appears to be that the log(log(x)) chart is the correct one for analysis.
could you display the Chart in log(log(x) please,so we can discuss it to be honest,I can not see that the log(log(X) Chart [the greenish line] ist more correct.
|
|
|
|
aminorex
Legendary
Offline
Activity: 1596
Merit: 1029
Sine secretum non libertas
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 14, 2014, 09:24:43 PM |
|
to be honest,I can not see that the log(log(X) Chart [the greenish line] ist more correct.
It is more correct in as much as the R^2 value is significantly, substantially, closer to 1. I would if I had more time draw the regression bands on a log(log)-linear chart, but I really shouldn't be doing that right now.
|
Give a man a fish and he eats for a day. Give a man a Poisson distribution and he eats at random times independent of one another, at a constant known rate.
|
|
|
Queeq
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 14, 2014, 09:41:25 PM |
|
The last graph is exactly what I meant by saying "trend change". It does not need to reverse for the change to be counted. It could be accelerating so the double logarithmic scale shows it more precisely.
|
|
|
|
aminorex
Legendary
Offline
Activity: 1596
Merit: 1029
Sine secretum non libertas
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 14, 2014, 09:48:29 PM |
|
The last graph is exactly what I meant by saying "trend change". It does not need to reverse for the change to be counted. It could be accelerating so the double logarithmic scale shows it more precisely.
On the finer time scales, linear analysis seems to fit the trend lines quite well. Zoom out a bit, and logarithmic analysis becomes unavoidable. Zoom out to the limits of history and log(log(x)) analysis begins to become compelling. Each of these changes of scale can be interpreted as encapsulating a larger range of changes in the trading regime, the modality of the trend.
|
Give a man a fish and he eats for a day. Give a man a Poisson distribution and he eats at random times independent of one another, at a constant known rate.
|
|
|
segeln
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 10:30:04 AM |
|
@aminorex, sorry,but i can not follow the more mathamatical Discussion about different log scales. In my textbooks about TA are only log Charts exposed and explained. Do you have any source about your preferred log(logx) Charts just for my understanding what we are talking about ?
|
|
|
|
oda.krell
Legendary
Offline
Activity: 1470
Merit: 1007
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 01:08:10 PM |
|
@aminorex, sorry,but i can not follow the more mathamatical Discussion about different log scales. In my textbooks about TA are only log Charts exposed and explained. Do you have any source about your preferred log(logx) Charts just for my understanding what we are talking about ?
This is not a staple TA technique, as far as I know. Then again, what you seem to consider acceptable (the simple "log line") is also just a mathematical tool that is used across a wide range of fields (biology, CS), not just in TA. What distinguishes the various "all time price graphs" are the starting assumptions about the underlying BTC price function. The graph you posted in here (that I criticized) is a straight line on a log chart with its slope and y intercept motivated by the lower candle wicks. The TA motivation approximately goes "the all time trend never went below those points, so this line marks the lowest boundary of the all time trend". Agreed? Next, the "simple" log-linear model is a line of best fit of BTC price mapped to a log chart. Basically, it is the "best" (i.e. least erroneous) line you can draw through all BTC price points *if* you assume price is generated a) by exactly one, immutable b) exponential function (like, for example, the function "price goes up 5% each week"). By the way, assumptions a) and b) also hold for the "hand drawn" line you posted, just that it's not a line of best fit, but a line that connects the lower extrema of the candles. The log-log-linear model, the green line in the graph above (which was made by user Ducky1 by the way, IIRC), does the same thing, find a line of best fit. Assumption a) above is the same, but assumption b) is different now: instead of assuming that a simple exponential function generated our price data, we are now looking for, i.e. we're approximating, a double exponential function. Hard to believe price could rise that fast? I know I am skeptical about it. Then again, if you accept the possibility of an exponential price increase, I see little reasons in principle to outright reject double exponential growth.
|
Not sure which Bitcoin wallet you should use? Get Electrum!Electrum is an open-source lightweight client: fast, user friendly, and 100% secure. Download the source or executables for Windows/OSX/Linux/Android from, and only from, the official Electrum homepage.
|
|
|
segeln
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 01:38:20 PM |
|
This is not a staple TA technique That`s why i asked aminorex for a source. The TA motivation approximately goes "the all time trend never went below those points, so this line marks the lowest boundary of the all time trend". Agreed? sorry,but in case you evaluate TA like that it is ridicolous. The line says. up to now the Price did`nt go below that line so there ist a strong Support. If the Price goes beneath that line the trend is at least violated and if it continues even broken and will change to another trend
|
|
|
|
oda.krell
Legendary
Offline
Activity: 1470
Merit: 1007
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 01:56:33 PM Last edit: January 15, 2014, 02:13:26 PM by oda.krell |
|
This is not a staple TA technique That`s why i asked aminorex for a source. The TA motivation approximately goes "the all time trend never went below those points, so this line marks the lowest boundary of the all time trend". Agreed? sorry,but in case you evaluate TA like that it is ridicolous. The line says. up to now the Price did`nt go below that line so there ist a strong Support. If the Price goes beneath that line the trend is at least violated and if it continues even broken and will change to another trend
Dude, two posts above it was pretty obvious you needed help understanding basics like linear regression and double exponentiation. I hope you don't mind if I don't take your opinion on TA all that serious, right? lookmaicancounttopotatoe.jpg EDIT: okay, let me be more clear: re: double exponentiation. there is not "official" explanation, just like there is not "official" explanation of simple log linear price models. they tend to work as an approximation, although I would alway point out that they're too imprecise to trade on. If you want the actual source of the graph, I already gave you a name, user Ducky1. re: bitcoinwondertrader.de's graph (or whatever his name was). My attempt to capture the motivation of that log line wasn't meant sarcastic, or anthing. that's roughly the idea behind it: price never dropped below, so until now, this marks the lower bound of the all time trend. what you quote after that doesn't contradict my summary at all -- I never claimed proponents of the log trend line say that price could never fall below this line, just that, so far, it never *has* fallen below it.
|
Not sure which Bitcoin wallet you should use? Get Electrum!Electrum is an open-source lightweight client: fast, user friendly, and 100% secure. Download the source or executables for Windows/OSX/Linux/Android from, and only from, the official Electrum homepage.
|
|
|
segeln
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 02:11:39 PM |
|
I hope you don't mind if I don't take your opinion on TA all that serious, right?
Neither do I take your opinion on TA too serious. You did`nt say anyting about what the line says in my point of view.
|
|
|
|
oda.krell
Legendary
Offline
Activity: 1470
Merit: 1007
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 02:14:26 PM |
|
added a more serious response in an EDIT
|
Not sure which Bitcoin wallet you should use? Get Electrum!Electrum is an open-source lightweight client: fast, user friendly, and 100% secure. Download the source or executables for Windows/OSX/Linux/Android from, and only from, the official Electrum homepage.
|
|
|
segeln
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 02:24:32 PM |
|
I never claimed proponents of the log trend line say that price could never fall below this line, just that, so far, it never *has* fallen below it.
but you have not realized yet i claimed always the same : just that, so far, it never *has* fallen below it and nothing else But you ascribed this the all time trend never went below those points, so this line marks the lowest boundary of the all time trend". Agreed? to me
|
|
|
|
oda.krell
Legendary
Offline
Activity: 1470
Merit: 1007
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 02:35:26 PM |
|
I never claimed proponents of the log trend line say that price could never fall below this line, just that, so far, it never *has* fallen below it.
but you have not realized yet i claimed always the same : just that, so far, it never *has* fallen below it and nothing else But you ascribed this the all time trend never went below those points, so this line marks the lowest boundary of the all time trend". Agreed? to me Which is empirically true. The "all time trend" (up to now) has never fallen below those points. If it ever does, the previously assumed all time trend is considered broken. In fact, that already happened, when the lower bound all time trend had to be updated at the last point (early 2013, I guess).
|
Not sure which Bitcoin wallet you should use? Get Electrum!Electrum is an open-source lightweight client: fast, user friendly, and 100% secure. Download the source or executables for Windows/OSX/Linux/Android from, and only from, the official Electrum homepage.
|
|
|
segeln
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 02:42:33 PM |
|
In fact, that already happened, when the lower bound all time trend had to be updated at the last point (early 2013, I guess).
When was the lower bound broken ? you mean from Jauar 2013 a steeper line ? That would be a "subtrend" within the longterm trend Edit: Hope we discuss more serious and less pejorative
|
|
|
|
oda.krell
Legendary
Offline
Activity: 1470
Merit: 1007
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 04:07:00 PM |
|
In fact, that already happened, when the lower bound all time trend had to be updated at the last point (early 2013, I guess).
When was the lower bound broken ? you mean from Jauar 2013 a steeper line ? That would be a "subtrend" within the longterm trend Edit: Hope we discuss more serious and less pejorative No, the current all-time log trendline is unbroken of course. That's the whole point: it is drawn that way. But I'm asking you to put yourself in the situation of a trader about a year ago. He used the same methods, tried to do the same thing, but he only had the price data up to that point, The following is a very sloppily drawn trend line. Please don't argue about the exact points of contact etc., it's just there to illustrate my point (and it could be made more precise, if necessary), ![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FbJUTZqY.png&t=665&c=hEBBPpT9NxpZEw) RED line is (sloppily drawn!) the trendline you posted, the one from the german website. Now time travel back to late 2012. You just drew the ORANGE trend line, also through the lowest candle wicks, just like we are doing now in early 2014. Maybe you can argue that back then you would have had a slightly less "solid" trend, less points of contact or something, but in the end, it's what you had back then, and what you had to work with. So anyway, then came (roughly) December 2012/January 2013, and that orange line was broken. So it had to be redrawn. The irony is of course that, as the previous AT trend line was broken downwards, we were about to enter a hugely profitable rally ![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif)
|
Not sure which Bitcoin wallet you should use? Get Electrum!Electrum is an open-source lightweight client: fast, user friendly, and 100% secure. Download the source or executables for Windows/OSX/Linux/Android from, and only from, the official Electrum homepage.
|
|
|
segeln
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 04:29:49 PM |
|
So anyway, then came (roughly) December 2012/January 2013, and that orange line was broken. So it had to be redrawn.
I think we are a bit hair-splitting now about drawing the proper line. I never drew the orange line and I dont not know wether the professional analysts from the german webside did it.(I doubt it) If a solid trend is going about to be broken you have also to take in account the time -scale. Breaking a solid trendline is not done in a few hours/days, it takes more time. Let me say a last word about the godmodetrader.de chart:in these nervous days/weeks here on bitcointalk.org I am very confident referring to that Chart since I am longterm-involved. I can`t stand these hour-/days-charts as posted here so many times. BTC is too dynamic to get "fenced" in charts like that. Do you agree ?
|
|
|
|
oda.krell
Legendary
Offline
Activity: 1470
Merit: 1007
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 04:42:52 PM |
|
So anyway, then came (roughly) December 2012/January 2013, and that orange line was broken. So it had to be redrawn.
I think we are a bit hair-splitting now about drawing the proper line. I never drew the orange line and I dont not know wether the professional analysts from the german webside did it.(I doubt it) If a solid trend is going about to be broken you have also to take in account the time -scale. Breaking a solid trendline is not done in a few days, it takes more time. Let me say a last word about the godmodetrader.de chart:in these nervous days/weeks here on bitcointalk.org I am very confident referring to that Chart since I am longterm-involved. I can`t stand these hour-/days-charts as posted here so many times. BTC is too dynamic to get "fenced" in charts like that. Do you agree ?Yes, and no. If your point is to zoom out a bit, look at the big picture, and encourage people to hold instead of panic sell: sure. It makes sense to post a log trendline and point out that price can go higher, can fall down, but in the end it looks like we're in a very long uptrend. But my 'orange line' example above wasn't purely semantic: it is the ultimate problem of what we're doing here -- using past data, analyzing it, and extrapolating to the future. And that alway carries the risk that it can at some point stop. "The Trend Is Your Friend, Until It Ends" holds for the short term view as much as it does for the ultra long term view. We probably agree in a more practical sense: price would have to fall much much deeper before I begin questioning the overall uptrend that BTC saw since trading began. It's just that I would never call it a "bull market" if price would deflate all the way to, say, 90 USD. So in that sense, it's a semantical distinction: what you call "bull market" I wouldn't apply to such a long time frame, but calling it an "uptrend" I'm fine with. That said, *if* we would fall back to 90, there's a good chance IMO we'd break through it. Not saying that BTC is finished in that case, but so much investor's confidence would be destroyed that all previous trends would, probably, be broken. But tbh, I highly doubt it'll come to that. In fact, I'd be surprised if we ever visit sub-500 for longer than a few hours, maybe a few days, again.
|
Not sure which Bitcoin wallet you should use? Get Electrum!Electrum is an open-source lightweight client: fast, user friendly, and 100% secure. Download the source or executables for Windows/OSX/Linux/Android from, and only from, the official Electrum homepage.
|
|
|
aminorex
Legendary
Offline
Activity: 1596
Merit: 1029
Sine secretum non libertas
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
January 15, 2014, 04:49:42 PM |
|
So anyway, then came (roughly) December 2012/January 2013, and that orange line was broken. So it had to be redrawn.
I think we are a bit hair-splitting now about drawing the proper line. I never drew the orange line and I dont not know wether the professional analysts from the german webside did it.(I doubt it) If a solid trend is going about to be broken you have also to take in account the time -scale. Breaking a solid trendline is not done in a few hours/days, it takes more time. Let me say a last word about the godmodetrader.de chart:in these nervous days/weeks here on bitcointalk.org I am very confident referring to that Chart since I am longterm-involved. I can`t stand these hour-/days-charts as posted here so many times. BTC is too dynamic to get "fenced" in charts like that. Do you agree ? Price has many components. Several components are best modeled as random variables in some distribution(s). If your trade frequency --which is limited by friction created by fees -- is high enough, then you can profit from relatively low S/N ratios, over the long run. This is the basic ideology of successfuly high-frequency trading. A great motivator for this approach is that risk is quite low for very short term trades. The result is slow steady gains. The analytic, technical, and market apparatus required is fairly substantial, however. The more frequently you can trade with a given constant edge, the greater your gains over time. Eliciting an edge of a given magnitude for a short term trade is much more difficult than eliciting an edge of that magnitude for a longer term trade. On a scale of human life, the reward/risk ratio for BTC is astronomical, and epistemic problems, analytic problems, market frictions, are inconsequential. A rational maximizer of expected outcome will attempt to scale their time-frames to their epistemic, analytic, and mechanical resources. Most players will fail to assess their correct scale, due to Dunning-Kruger effect. They will trade too frequently, and probably suffer from other deep mistakes.
|
Give a man a fish and he eats for a day. Give a man a Poisson distribution and he eats at random times independent of one another, at a constant known rate.
|
|
|
|