Neighbourhood Pool Watch: 9.11 ASIC choices: Avalon and BFL earnings

0. Introduction
Bitcointalk.org forum member SolarSilver mentioned to me that Avalon had updated the shipping dates, and thought it be a good idea to update the earnings estimates, and I had to agree.

Since the simplified earnings charts from the last post seemed to most readers to be preferable to the previous charting methods, I've extended that idea somewhat. I decided to make the earnings estimates agnostic so they could be used for any mining device. To do this, the daily and cumulative BTC earnings charts are per Ghps.

To obtain the btc amount earned for your device, take the daily or cumulative BTC earned and multiply by the hashrate (in Ghps) of the device in which you are interested. The amount you'll need to subtract for electricity costs are given by the final two charts.

Another change I've made is to provide historical data as well, so the amounts earned by Jeff Garzik and Chinese Avalon miners can be estimated. Since I'm have trouble obtaining this data from blockchain.info, I've used the nethash data from http://blockexplorer.com/q/nethash. The disadvantage of this is that data is only available for every 144 blocks, and the charts show this.

In order to estimate the network hashrate, I use a Gamma family generalised smoothing spline. I obtain better results when using network hashrates with constant time intervals (rather than constant numbers of blocks) and Poisson family smoothing splines, but the results are sufficient for this analysis. The chart of the estimate and the actual per 144 block hashrates are shown below for the interested reader.

I've also provided a dataset of the earnings estimates if you have trouble reading values from the charts.

1. Assumptions
These have again been changed to reflect the current guesses about ASIC manufacture. Avalon and BFL shipping dates have been adjusted, and ASICMiner has only 8 Thps left to add to the network hashrate. I'm using the not yet confirmed Avalon "batch 4" again in order to extrapolate data further.

Arrivals are constant. Rather than modelling arrival times, I'm assuming a constant rate of hashes per second added to the network for each of the batches.
Batch 1: 250 units arrive from 11th March to 17th March.
Batch 2, 3 and 4: 1.2 Thps per day delivered from 1st April to 1st July.
BFL delivers 2000 Ghps per day (~35 SC Singles per day) from 31st March to 1st July.
ASICMiner delivers 8 Thps evenly from 4th March until 31st March (the hashrate increase for ASICMiner has slowed recently)
Initial difficulty = 4 367 876
Initial network hashrate = 33 975

2. Charts

Chart 1: As mentioned above first chart is for readers who are interested how I estimate the network hashrate. The dots are the 144 block average hashrates, the block line is the hashrate estimate, and the white shading is the standard error.

The method I use here is not as accurate as if I had easy access to the timings of every main chain block solved, and can only be updated every day or so. If I did have a simple way of obtaining block timestamps then I would be able to provide a realtime (actually hourly)network hashrate estimate with standard errors - which would be very handy if you want to know what the network hashrate is right now.

Chart 2: This group of charts show the expected hashrate, difficulty and blocks per day given my assumptions.

Chart 3 shows:

Expected cumulative BTC earnings per Ghps, when starting mining on the Friday of each week, from 1st February.
Daily BTC earnings from 1st February.

Charts 4 and 5 show the cumulative electricity costs and daily electricity costs for all known ASIC devices, separated into 220V and in 110V countries.

3. How to estimate earnings

Decide when you think your ASIC will arrive and what the US$ exchange rate will be. Read the btc earned from the cumulative earnings chart, multiply by the hashrate of the device you wish to assess. Multiply by your estimated USD$ exchange rate to find an earnings estimate in US$.
On the appropriate cumulative electricity costs chart, select the on the x axis of the number of days mining of which you wish estimate the cost, and on the y axis your electricity cost per kWh. The estimated cumulative cost of mining can be read from the colourbar on the side.
Subtract the electricity costs from the US$ earnings.

Example: You hope your Avalon ASIC will arrive on the 5th of April. The cumulative btc earned by 1st July (the maximum x axis value) is ~ 2 BTC / Ghps. You're thinking of purchasing a BFL Single SC, so you multiply that by 60 Ghps to get 120 BTC. You're bullish about BTC, and you estimate the exchange rate will be US$40 per BTC at that point, earning you a total of US$4800 in 58 days.

You live in California, so you have 110 V mains and your electricity costs are US$0.35 per kWh. From the upper cumulative electricity chart, you mark the intersection of 58 days (x axis) and US$ 0.35/kWh (y axis) and by matching the colour from the colour bar, estimate a total cost of ~ US$200 (click on the chart for a closer view).

In this case, the amount earned will be US$4600.

If you dont want to use the charts the cumulative earnings can be found in this dataset, and the cumulative cost will be:

cumulative daily electricity cost = device watts * cost per kWh / 1000 * days * 24

4. Conclusions

ASICMiner's "slow and steady approach", fewer Avalon arriving than I estimated and BFL ASICs still unavailable, far fewer hashes than expected were added to the network over the last fortnight. This means that the network hashrate estimates and difficulty estimates are lower than previously, and earnings estimates higher than previously. This is a boon for anyone who already has an ASIC, and for anyone who has a current order for one purchasing. Avalon's announcement about the batch 2 ASICs was welcome.

At the end of the email was:

For clarification the shipping priority is as such:

Sorted from Highest to Lowest.

Developers - e.g. bitcoin core dev and other bitcoin contributors we feel deserve this treatment.

Trade-Ins/Old Customers - your unit will ship soon as we receive the FPGA and you've paid the difference.

Regular Batch #2 Customers - Unfortunately you are not special.