Neighbourhood Pool Watch: Weekly network forecast 1st July 2013

1st July, 2013

Last post

Important:
These posts are a test of the methods used to create the forecasts, and definitely do not forecast non-MTGOX US$ / BTC price related changes - for example the current massive decrease in the cost of a hash. Please do not use the forecasts here for financial forecasts - these posts might be useful in that respect.

0.Introduction

The network hashrate continues to increase even as the MTGOX US$ BTC price falls, which should decrease the accuracy of the forecasts. However it seems that the stability of the network hashrate increase over the last four weeks has lead to improvements in the model forecasts, regardless of the decrease in price. Part of this is that the model is based on the log of the median price and the average hashrate, and partly due to some of the slight increases in price over the past weeks.

So, for the past four weeks the one week hashrate forecast has been quite accurate, with very low percentage and absolute average errors. The other models have been less accurate due to the lag in their responses, but the network hashrate has still been inside their confidence intervals for error.

Difficulty modelling has been somewhat less accurate - the method 2 is much less flexible than method 1, but method 1 doesn't have confidence intervals for error so an estimate of its relative accuracy is not possible.

If the network hashrate continues to rise in a steady fashion, future average weekly network hashrates are also likely to be within current forecasts' 95% confidence interval for error - perhaps at least the one week forecasts might become useful once again.

1. Models and datasets:

The model datasets have been collected into one paste to save time. Model estimates have been likewise aggregated.

Forecast and canary model analysis

All datasets

All estimates

Canary model: An estimate of the current hashrate estimate based only on current price and previous network hashrates. Used to detect changes to the network caused by agents other than the direct action the MTGOX US$ / BTC price as determined by the model.

Models f1, f2, f3 and f4: These are the one, two, three and four week forecasts of the network hashrate, based on the modelled relationship between the MTGOX US$ / BTC and the network hashrate.

Difficulty modelling: Method one calculates a difficulty estimation using the network hashrate estimation and is recalculated each week. Method 2 calculates difficulty estimation based only on previous difficulties and the MTGOX price, and is only recalculated after each difficulty retarget.

2. Results

3. Charts

Chart 1:

A graphical history of the one, two, three and four week forecasts of the weekly average network hashrate compared to the actual network weekly average hashrate.

Chart 2:

A graphical history of the Canary model estimates of the weekly average network hashrate compared to the actual network weekly average hashrate.

Chart 3:

A graphical history of the last twenty six weeks of retarget date and difficulty estimates, actual retarget dates and actual difficulties. It should be noted that each retarget will often be forecast by two consecutive weekly forecasts, hence the multiple points per retarget on the new chart.

Chart 4:

A graphical history of the one, two, and three retarget forecasts of the Difficulty, modelled directly as a function of previous Difficulties and MTGOX US$ / BTC prices.