Topic: . - page 3. (Read 9137 times)
Way cool, what I see is lots of activity leading up to a peek when new demand (cryptocurrency converts) enter the Bitcoin economy. Once in, it looks like they locking in and save. (an attitude of I have my Bitcoin's I'm going to hold tight and let deflation play out its course) then Activity / Demand decreases.
As demand decreases the price falls, people start to use / sell their coins and activity increases, as activity increases the demand increases, and the price increases with demand activity starts to decrease (people start saving)
So it looks like to me, new cryptocurrency converts saving there coins, are the cause of the decrease in activity that causes the price to fall. The falling price of Bitcoin is akin to inflation in the Bitcoin economy and should spur innovation and growth, and disperse in bitcoin as a store of welth investment (and a decrease demand for Bitcoin.)
That will in turn drive activity and so we start the cycle again.
The graft can't make a prediction as it is driven by cryptocurrency converts, but it is cool it confirms that Bitcoin is distributing and the nature of its use / value.
As demand decreases the price falls, people start to use / sell their coins and activity increases, as activity increases the demand increases, and the price increases with demand activity starts to decrease (people start saving)
So it looks like to me, new cryptocurrency converts saving there coins, are the cause of the decrease in activity that causes the price to fall. The falling price of Bitcoin is akin to inflation in the Bitcoin economy and should spur innovation and growth, and disperse in bitcoin as a store of welth investment (and a decrease demand for Bitcoin.)
That will in turn drive activity and so we start the cycle again.
The graft can't make a prediction as it is driven by cryptocurrency converts, but it is cool it confirms that Bitcoin is distributing and the nature of its use / value.
typical price*bitcoin volume = an estimate of USD volume
If I plot USD volume it looks pretty much the same as net money flow. I think you are taking lag 1 differences and then adding them back when you do the integral (get the running cumulative "net money flow") so in the end it is a wash. Perhaps I am misunderstanding, but the charts are definitely very similar.
I still have no intuition for what total money flow means, at least the way I calculated it (and got similar charts to yours except using daily data rather than the 12 hr you must be using for so many data points). I did it as the cumulative sum of lag 1-day differences in USD volume magnitude. I think it may be interesting though.
If I plot USD volume it looks pretty much the same as net money flow. I think you are taking lag 1 differences and then adding them back when you do the integral (get the running cumulative "net money flow") so in the end it is a wash. Perhaps I am misunderstanding, but the charts are definitely very similar.
I still have no intuition for what total money flow means, at least the way I calculated it (and got similar charts to yours except using daily data rather than the 12 hr you must be using for so many data points). I did it as the cumulative sum of lag 1-day differences in USD volume magnitude. I think it may be interesting though.
Even though the data looks noisy there is a conservation of value at play here, that is why I use the whole integral from the very first trade, even very small imbalances add up over time. Using USD volume as an approximation also does not help you with the incident *direction* of money flow, and the integral does not add up the same.
Total money flow does not care about direction though, it always goes up because it uses the absolute value of money flow for each sample.
Hmm. I think there is some miscommunication. If you posted some code or part of a table that would be great. I'm also unsure why the closing price gets so much attention. It is pretty arbitrary with respect to bitcoin. VWAP instead of typical price would make more sense I think. But even VWAP*BTC volume is just estimating the actual USD volume which seems to be the measure that includes the least compression... I dunno.
To get total money flow I integrate the absolute value of those terms. It looks like this:
The net money flow integral looks like this:
The net money flow integral looks like this:
And this represents the volume under the curve of the money flow?
typical price*bitcoin volume = an estimate of USD volume
If I plot USD volume it looks pretty much the same as net money flow. I think you are taking lag 1 differences and then adding them back when you do the integral (get the running cumulative "net money flow") so in the end it is a wash. Perhaps I am misunderstanding, but the charts are definitely very similar.
I still have no intuition for what total money flow means, at least the way I calculated it (and got similar charts to yours except using daily data rather than the 12 hr you must be using for so many data points). I did it as the cumulative sum of lag 1-day differences in USD volume magnitude. I think it may be interesting though.
If I plot USD volume it looks pretty much the same as net money flow. I think you are taking lag 1 differences and then adding them back when you do the integral (get the running cumulative "net money flow") so in the end it is a wash. Perhaps I am misunderstanding, but the charts are definitely very similar.
I still have no intuition for what total money flow means, at least the way I calculated it (and got similar charts to yours except using daily data rather than the 12 hr you must be using for so many data points). I did it as the cumulative sum of lag 1-day differences in USD volume magnitude. I think it may be interesting though.
Nice to see you posting interesting things again, chodpaba.
My personal analysis seems to converge with yours - though I'm reluctant to post a chart with the indicators I've been working on for the last ten years, for obvious reasons. Right now, we're in a bull phase that has some indications of continuing until at least May 1st. (Using my timing indicator, that is.)
Around that point, plus-or-minus a few days, we're likely to start a new bear phase. The only problem is - I can't estimate magnitude. So it may be your classic 'panic' decline, or a relatively soft correction that looks more like a sagging plateau. Its a very difficult problem to solve.
Keep us posted, it is fascinating to look at.
My personal analysis seems to converge with yours - though I'm reluctant to post a chart with the indicators I've been working on for the last ten years, for obvious reasons. Right now, we're in a bull phase that has some indications of continuing until at least May 1st. (Using my timing indicator, that is.)
Around that point, plus-or-minus a few days, we're likely to start a new bear phase. The only problem is - I can't estimate magnitude. So it may be your classic 'panic' decline, or a relatively soft correction that looks more like a sagging plateau. Its a very difficult problem to solve.
Keep us posted, it is fascinating to look at.
Intuitively, I'm leaning more toward a relatively soft correction. We have one data point (2011) showing that a massive/drastic drop following a parabolic move will not kill overall appetite for bitcoin in the long run. Even if you bought at the peak of $32, and held, you have made massive returns by now. I think that would add a lot to dampen the emotional panic. Perhaps that may explain the major bounces yesterday.
Nice to see you posting interesting things again, chodpaba.
My personal analysis seems to converge with yours - though I'm reluctant to post a chart with the indicators I've been working on for the last ten years, for obvious reasons. Right now, we're in a bull phase that has some indications of continuing until at least May 1st. (Using my timing indicator, that is.)
Around that point, plus-or-minus a few days, we're likely to start a new bear phase. The only problem is - I can't estimate magnitude. So it may be your classic 'panic' decline, or a relatively soft correction that looks more like a sagging plateau. Its a very difficult problem to solve.
Keep us posted, it is fascinating to look at.
My personal analysis seems to converge with yours - though I'm reluctant to post a chart with the indicators I've been working on for the last ten years, for obvious reasons. Right now, we're in a bull phase that has some indications of continuing until at least May 1st. (Using my timing indicator, that is.)
Around that point, plus-or-minus a few days, we're likely to start a new bear phase. The only problem is - I can't estimate magnitude. So it may be your classic 'panic' decline, or a relatively soft correction that looks more like a sagging plateau. Its a very difficult problem to solve.
Keep us posted, it is fascinating to look at.
Is this MtGox data? Can you show me the actual equations? I'm having trouble understanding your definitions.
Yeah, it's Gox data.
The 'typical price' or 'pivot point' for a candle is often calculated as the average of the High, Low, and Close. Multiply that by the volume of that candle. For me it makes more sense to use the Bitcoin volume than to use the Dollar volume for this. The difference between that calculation for each candle is supposed to be a measurement of money that has flowed in the market from candle to candle, and the direction it flowed. That is what I am calling net money flow. For what I am calling total money flow I just take the absolute value of that money flow calculation. Each one of those is summed in an integral from the beginning of the time series, and the ratio taken from those integrals. (total money flow/net money flow)
So TMFn = sum(i=1, n) |NMFi| ?
Thanks, if I understand correctly then the ratio is a measure of the quantity of money moving in/out of bitcoin relative to the total amount ever moved to that date.
Sorry to be dense, but basically what you are saying is, that there is a serious resistance point at around $189 and another at $1274. What about support levels, though, or since there has only been one arc before is there not yet enough dips to draw any conclusions about that?
This is fascinating to look at, thank you for posting. So the price could potentially get to the $180-$190 range before reversing, eventually hit its bottom somewhere during the next arc, and then get to the $1270 during the next rally in the coming years?
That is pretty much my speculation at this point. But since the data is very noisy I would not want to bank on those values being very precise, if my speculation is correct it would be somewhere in that ballpark.
Regarding the current peak: If the $32 peak is any indicator, the value of the money flow arc does not quite converge with price. And once they start to diverge it is all over. The divergence of the money flow ratio currently has already started to look more decisive. I suppose enough cash could change that but everything I am seeing points to a drop in the price in the short term.
What I do not have a good bead on is how low to expect it to drop. While the money flow ratio seems to clearly show extremely overbought peaks I have not really come up with a way of reading oversold conditions from it or valuing them.
For sure its all ballpark speculation for those exact prices. But yeah the peaks of the arcs clearly don't match with the bottoms. Oct-Nov 2011 bottom was only a 1/4 into the second arc followed by a year long climb to connect with the arc again. Again this is all fascinating!
Please forgive my ignorance but could you define money flow for me, is this total capital coming into bitcoin in USD? It looks interesting though, I too think that bitcoin may be bubbling and have positioned myself for the time being into fiat, but my decision was based on search volumes being excessive compared with historic norms.
Average(H+L+C)*Volume for each candle. Add those up for net money flow, and take the sum of the absolute values for total money flow. Some analysts use dollar volume, but I use Bitcoin volume. Since price is $/BTC and volume is in BTC that yields money flow dimensioned in $ for each sample. The money flow ratio I was talking about is technically dimensionless as it is just $/$ (total money flow/net money flow).
(edit) Sorry, I should clarify. Net money flow is the *difference* of that calculation from candle to candle.
Daily?
Jump to: