In the last two days the market impressed me; an incredible acceleration from 10$ to 17$ with good volume. Many has asked me to make a forecast about a potential important target, well isn’t an easy task but i’ve decided to do it, so i calculated a potential mid term target using weekly data. Usually a swing, in any time frame (hourly, daily, weekly, monthly) may last up to 25-30 bars with a minimum of at least 4 bars. This weekly swing started from 0.56$ in april and so far this market has risen over the last 9 weeks.

*Hypothesizing* that this weekly swing will last for 25 weeks, a *maximum *target would be around ~70$ in october. I’m pretty sure that approaching such an impressive target many big bitcoins holders will be tempted to sell their stake stopping the ride. Can this market go beyond 25 weeks? Yes, but more likley with a weekly correction of at least 3-4 weeks.

Otherwise if the top is now and a weekly correction will start immediately with a minimum duration of 4 weeks, the area around 10$ may provide support attracting long term buyers.

*If someone is interested to a further deep explanation on how i derived these targets, let me know it leaving a comment here. Thank you.*

### Like this:

Like Loading...

*Related*

I am very interested in such an explanation.

Ok, so i’ll try to be concise as much as possible.

These are the steps:

1)I compute natural logarithm of weekly increments since mtgox opening (17 august 2010)

ln(t)/ln(t-1)…ln(t-1)/ln(t-2) and so on for all data

as a first raw input i use the average of high and low of each weekly data bar

2)i compute the mean for all the weekly increments as computed in step 1, i’ll label the result as AVG

3)I compute rms (root mean square) of all weekly increments, squaring each weekly increment and sum all togheter, i’ll label the result as RMS

The term root-mean-square is used in the engineering or physics sense of

noise powerfrom electronic communication theory4)With these two values RMS and AVG i’ll compute then the shannon probability with this formula:

P = (((avg / rms) – (1 / sqrt (n))) + 1) / 2where n is the total number of samples used, in this case the number of weeks of the input, the formula will compensate also the probabilty for possibledistorsion due to the usage of a short data set. this formula is the tendency of the underlying time serie to rise or fall, possible values range from zero to one.

5)with P we can then compute the gain factor

Gwith:G = ((1+rms)^P*((1-rms)^(1-P))

6) at this point we can forecast a potential top or low

for top: e^((t * ln (g)) + (rms * sqrt (t)))

for low: e^((t * ln (g)) – (rms * sqrt (t)))

where e stand for exponential function

where t stand for the number of time period, in this case i used 14 weeks (9 + 14 =25 total weeks for the swing).

so:

avg = 0.124258

rms = 0.293455

P = 0.682355 (unusually high, above 0.5 is persistence, under 0.5 is anti-persistence)

G = 1.0674 we can compute the forecasted price also using only g simply doing for this case g^14(weeks from now) or 2.49 times the initial value of ~15$ leading to 37$ so without taking in account the rms or volatilty

with the formula of step 6 we have

top=72.16

low=8.02 this means that if during this upswing due to noise or volatility we might collapse to 8 in the process:) yes there is a lot of volatility!

We have only 42 weeks of data since mtgox opened the doors, i compensated all the value for such a short data set, with the pass of time we will have more accurate value of avg/rms/p and g and more accurate resistance and support projections.

I’d like to add while i said in the post 25-30 samples as maximun duration for a swing and 4 for the minimum duration

this is because the chance to move in a direction follow this formula

1/square root of time and with 4 we have 0.5 or 50% chance that a swing will last 4 bar of any time frame while 25 gives only 20% chance that the swing will continue.

For further details and explanations of these formulas i recommend this website http://www.johncon.com/ntropix/

Well, i’ve to look better:) i’ll study carefully that website, interesting stuff!

Me too, i’ll like to use that approach for computing daily high and low target.