Econometric model of BTC valuation

[hazenyc]

For your consideration, an econometric model whose coefficients are statistically significant to 99.99% confidence interval, and with an adjusted R² value that explains over 90% of the variation in BTC prices (in USD). The differences between model4 in blue and model3 in red, is that the model in blue only takes data out to Dec24 2011 (685 observations) while the model depicted in red goes back to 1180 observations. It is possible that something fundamentally changed inside of those 2 intervals, so an average of the 2 may be most accurate.

As for what goes into the model and the mathematics, that is proprietary. But take a look at the output:

[kdrop22]

Nice post, it is interesting to analyze these mathematical models.
Could you post the mathematical formula and parameters/assumptions, used to calculate these charts.

[hazenyc]

Nice post, it is interesting to analyze these mathematical models.
Could you post the mathematical formula and parameters/assumptions, used to calculate these charts.

I cannot, but I can tell you that it predicts over $600 by this December.

[mskryxz]

Would it be too much to ask to see what it projects for April 2013?

I just want to see if it compares to mines.

I have mine projecting originally at $500 by the end of the year. But because of November 18, and other factors, I too am projecting $600+ by end of the year and $500+ by end of November.  We are $23 away or $477 at Mtgox.com previous high of reaching that goal.

[NUFCrichard]

For your consideration, an econometric model whose coefficients are statistically significant to 99.99% confidence interval, and with an adjusted R² value that explains over 90% of the variation in BTC prices (in USD). The differences between model4 in blue and model3 in red, is that the model in blue only takes data out to Dec24 2011 (685 observations) while the model depicted in red goes back to 1180 observations. It is possible that something fundamentally changed inside of those 2 intervals, so an average of the 2 may be most accurate.

As for what goes into the model and the mathematics, that is proprietary. But take a look at the output:

That is a fairly unimpressive correlation, your line completely missed the march/April rise. Does it ever predict significant falls?

[DeathAndTaxes]

Remember price =/= valuation.   Not speaking on the merits of the model but if the model is using non-price inputs and speculation drives price up (in short term) faster than the underlying economic activity the model isn't modeling that.  However as you see price eventually corrects to value. 

[hazenyc]

That is a fairly unimpressive correlation, your line completely missed the march/April rise. Does it ever predict significant falls?

you obviously have no experience with modeling and forecasting. Everything is probabilistic, it cannot predict crashes or spikes. It cannot predict forces outside of the model. What it can do is explain and give insight into the drivers of phenomena. The correlation itself is 93.6% which is actually super impressive when it comes to fitting data. It is statistically significant to .001 of significance.

[PrymeTyme]

Remember price =/= valuation.   Not speaking on the merits of the model but if the model is using non-price inputs and speculation drives price up (in short term) faster than the underlying economic activity the model isn't modeling that.  However as you see price eventually corrects to value. 

Price + Time = Value

Economy per se  , has no direct Impact in that , the History of BTC speaks for itself

[DeathAndTaxes]

Remember price =/= valuation.   Not speaking on the merits of the model but if the model is using non-price inputs and speculation drives price up (in short term) faster than the underlying economic activity the model isn't modeling that.  However as you see price eventually corrects to value. 

Price + Time = Value

Economy per se  , has no direct Impact in that , the History of BTC speaks for itself

That is about the dumbest thing I have heard today. 

[PrymeTyme]

Remember price =/= valuation.   Not speaking on the merits of the model but if the model is using non-price inputs and speculation drives price up (in short term) faster than the underlying economic activity the model isn't modeling that.  However as you see price eventually corrects to value. 

Price + Time = Value

Economy per se  , has no direct Impact in that , the History of BTC speaks for itself

That is about the dumbest thing I have heard today. 

at least u heard something ....



what economy existed in the beginning of BTC ?

tell me ? what economy right now  supports the current value of BTC ? .. gambling ? Drugs ?

[DeathAndTaxes]

Once again.  PRICE =/= VALUE.

[hazenyc]

Once again.  PRICE =/= VALUE.

I would say that the VALUE of a bitcoin is the labor power required (computing power + energy costs + labor costs of building hardware etc.) to produce a bitcoin. It's a very Adam Smith/David Ricardo labor theory of value argument.

The PRICE is what a buyer and a seller agree to transact at.

Time has nothing to do with either (except for the labor time part of value).

[Impaler]

OP's lines look like they are hash-rates due to the rather smooth exponential look of them, some additional data may give them the more fine grained squiggles.  Hash rates ultimately have nothing to do with creating value (though some idiots think they do), thus I conclude OP's model is worthless.

[PrymeTyme]

Once again.  PRICE =/= VALUE.

I would say that the VALUE of a bitcoin is the labor power required (computing power + energy costs + labor costs of building hardware etc.) to produce a bitcoin. It's a very Adam Smith/David Ricardo labor theory of value argument.

The PRICE is what a buyer and a seller agree to transact at.

Time has nothing to do with either (except for the labor time part of value).

Beyond understanding the importance of knowledge of the present and the importance of arranging transactional data into the bell curve,
the approach to markets advanced incorporates two basic equations. The first is used to break down and define market reality.
The equation is simply: price + time = value. This statement is known and even obvious to most intelligent persons; subconsciously we accept
a known product 's price as value as we repeatedly are charged that price over time. The equation also holds true for the most complex
markets. It captures and defines a market's range of activity, creating a distinction between prices which are accepted as fair value and those
prices which the market rejects as temporarily unfair ly above or below value in the current timeframe.


 Market activity generates information:

  A) A readable preference of variant time-price opportunities which
        occur at different times, at different or the same prices, at
        different or the same levels of volume. This is an expression of
        accepted, rejected or changing value by participants as a whole.

           I) Time-price opportunities (and hence prices) can appear with a
               large or small relative amount of time.
 
           a) Those time-price opportunities resulting in a large amount of
               time must result in large transactional volume and display
               market acceptance or value for both sides in that timeframe.

            b)Those time-price opportunities resulting in a small amount
               of time must result in small transactional volumc and display
               market rejection or unfair value for one side in that timeframe.

  B) This amounts to market distinction as to price.
 
           I) prices accepted versus prices rejected

  C)  A further distinction of the type of market activity can be made within
       the framework of acceptance/rejection.

           I) prices in a market-created opportunity which are responded to
             versus prices in a market-created opportunity which are
              overtaken by other initiating forces. Therefore...

Price + Time = Market Acceptance (responsive or initiating)
                 =Transactional Volume
                 =Market Value (definition of all free markets)

[mescalito]

It looks like an output of some Monte Carlo simulation.

[SuperHakka]

models schmodels. I is dumbfk alright but even I can tell that depending where you cut off yur data points for modelling lead you to entriley different outcomes, if you get my point.

[NUFCrichard]

That is a fairly unimpressive correlation, your line completely missed the march/April rise. Does it ever predict significant falls?

you obviously have no experience with modeling and forecasting. Everything is probabilistic, it cannot predict crashes or spikes. It cannot predict forces outside of the model. What it can do is explain and give insight into the drivers of phenomena. The correlation itself is 93.6% which is actually super impressive when it comes to fitting data. It is statistically significant to .001 of significance.

So if it turns out that we are now in a bubble, your model doesn't work?

[hazenyc]

So if it turns out that we are now in a bubble, your model doesn't work?

A model is meant to describe reality, not MIRROR reality. A model is only as good as its inputs, so if there is a bubble as a result of irrational exuberance that cannot be modeled. The model doesn't exist to be an accurate representation of the world, nor should it be interepreted as such. It is meant to gain insight into what is going on. If price action occurs that the model cannot describe, then it is safe to say, at least, that it is not due to the assumed inputs in the model causing the price action and that it must be explained by some exogenous variable(s).

But the point of making a model is to keep it as simple as possible, as complexity bogs down the insight gained from the calculations.

For a real good book on financial models and why most people don't understand their purposes you should read Models Behaving Badly:

http://www.amazon.com/Models-Behaving-Badly-Confusing-Illusion-Reality-Disaster/dp/1439164991/ref=sr_1_1?s=books&ie=UTF8&qid=1384710333&sr=1-1&keywords=models+behaving+badly

 

[Adrian-x]

OP's lines look like they are hash-rates due to the rather smooth exponential look of them, some additional data may give them the more fine grained squiggles.  Hash rates ultimately have nothing to do with creating value (though some idiots think they do), thus I conclude OP's model is worthless.

This is just an unsubstantiated personal opinion.

Difficulty affects ones ability to mine coins verses buy them, thus the higher the difficulty the more one adds to the demand side of a finite supply. So it should be expected that the law of supply and demand will impact the price.

[hazenyc]

OP's lines look like they are hash-rates due to the rather smooth exponential look of them, some additional data may give them the more fine grained squiggles.  Hash rates ultimately have nothing to do with creating value (though some idiots think they do), thus I conclude OP's model is worthless.

This is just an unsubstantiated personal opinion.

Difficulty affects ones ability to mine coins verses buy them, thus the higher the difficulty the more one adds to the demand side of a finite supply. So it should be expected that the law of supply and demand will impact the price.

Well after trying to reconcile the auto-correlation in the model it turns out that after careful analysis, hash rates are indeed caused by high prices and not vice-versa according to the data,.. this, for me, only makes it more interesting!