Bitcoin Forum

:

I appreciate your forecasts. Keep up the good work!

Thanks, hopefully it's accurate.

mining is over  :'(

Mining 10000 btc is over with the actual $/btc rate. Mining, however, is not over AT ALL.

Oh damn.
If the next dificult is really 847013 i'll send my rig back :S

Hm i dont have that much moneey the rig was about 740 Eur for 1 5870 and two 5850.
I thinck i would make about 1 Ghash.
The only thing i want that it pays fot itself  and later keeps above the power price0.28 per kwh ( 20 er cent)

Hmmmn.  To come about, those numbers would have to be met by a surge in ATI card production (surely they know us and what we want by now), coupled with a constant and steady rise in the value of the BTC, but as no one can really predict either of those things, it's nice to have a worst case scenario charted out.  Thanks.

When the difficulty gets to 2Million mark price for BTC will have to be $120 / coin to maintain mining profitibility..

Actually, it will only have to be enough to offset electricity - as most miners in this stage of the game have already paid off their rigs. And in some cases, some don't even have to worry about that! I think your statement may be more true for someone just purchasing a $1,000 rig, given the reduction in payout. Otherwise, its pretty much full profit for the early-adopting miners.

Even at 2.85M difficulty level, bitcoins only need to be worth about $8.5/ea for it to remain profitable for me to mine.

To remain profitable, it just has to be high enough to exceed electricity and depreciation. Right now, that is ~$6 for me. At 2 million it will be only ~$20. In other words, the difficulty/price ratio must be below 100000. Anyone with cheaper electricity than me (a lot of people), or a more efficient setup (more than one graphics card, not an overclocked quad core CPU) can profit on even less.

You make it seem like purchasing BTC on the exchange is easy. It's been a whole week now, and I still haven't successfully transferred any $$$ into a exchange.

If I had invested the $1100 I did directly into bitcoins instead of mining equipment, I would have had over $20k in bitcoins now.  Unfortunately, I did not.  But I did not for good reason - that $1100 was an investment I could not afford to lose.  By "investing" it into mining equipment, I hedged the risk to a reasonable degree (since I could resell the equipment I bought for 90% of what I paid for it, easily), and thus was able to make the investment.  If I had only been able to invest in coins themselves, it would have been something along the lines of $100 or so.

So ~50-60% increase, per difficulty.

Can you also guesstimate when those difficulties will come, in days, given current growth/etc?

Understood on going out too far : )

2-3 Difficulties should be reasonably extrapolate-ble though, I think.

I'm curious to know the meta-concequences of how the difficulty will affect the going price of the BTC.

Would a higher difficulity mean prices get driven higher? Would that mean that if one were to invest into BTC, now would be the time before this happens?

The price will stay below $22 for the next week, followed by a slow creep up to 28-30 in two weeks.  Use those numbers and tell me what happens :)

I'm also very curious to see if total tera-hash/sec capacity increases, or if outright buy-and-hold does. Perhaps in the absence of large sell blocks, we can look forward to a bit of a plateau?

Appreciate the numbers as always, chodpaba!

Wow, this thread was very insightful, thank you for the numbers. I was shocked by the degree of the increase today but I didn't realize that the difficulty could potentially ramp up THAT much in just two weeks to a month. I had to revise the calculations I made of my new rig's probable output at 1.0 GHash/sec... geez, only a third of a coin a day at 3 million+ difficulty! :o

..........

For the last year, difficulty has grown at almost exactly 4%/day on average.  This corresponds with the growth in computing horsepower joined to the network.

see
 http://bitcoin.sipa.be/ for the graphs.

I'm confused, you're projecting a jump from 1,379,223 to 1,684,194 in the next jump, a 22% increase.  But the jump after that is to 2,930,547 - a 74% increase?

chodpaba, thanks again for the numbers.

Been running a few, and I noticed that bitcoin charts actually didn't do the usual "Double the T-Hash rate" thing it usually does when we transition to another difficulty level. In fact, the suggested target of 1,378,416 in 1,976 blocks @ 7.85 blks/hr sounds downright reasonable at a slight decrease of -0.06%!

Wonder if they fixed their estimation windows/averaging.

chodapa, what do you think the odds are of hitting 7m difficulty before the end of August?

Can you edit first post with your updated numbers?


UPDATE 6/24/2011 4:46PM
Here is one way to keep the original record and still update the first post.

If you are interested and want to know what I am doing read on

What he said.


UPDATE 6/24/2011 4:49PM
Hey look my comment was even updated after being quoted

The stock price has been tanking lately.  You know, they've been CEO-less for quite some time now and rumor is, the new guy is from IBM: culture clash.

I bet the cards had an impact on their sales already, no doubt.  But with the increases in difficulty, only gamers will buy new cards now.  Mining to pay for your hardware is a losing proposition now.

The stock market tanked after Sept. 11 and also around 2007/2008... did video cards stop getting better at an exponential rate? Nope.

Nice work.

Around ~3M for block 137088 seems reasonable if the price goes back up to ~$30 before then.

Longer-term prospects of Moore's Law look good.  7000 series AMD GPUs rumored to start shipping later this year, and next-gen Xilinx Artix-7 FPGAs in 1Q 2012 might be cost-competitive with the AMDs.


And then the reported hash rate / difficulty estimate can spike way high just after the readjustment.  I've seen this on bitcoinwatch and bitcoin.sipa.be too.  I think its because they're using the hash rate estimates on blockexplorer.  There should be some method to smooth out the estimate across the readjustment (a short-term estimate with no use of price).  I'd like to use it for my charts.

I think difficulty will level off at the next recalculation, and it will probably take near 14 days to get there. Potential miners are already cancelling their video card orders or trying to return or restock their hardware bought in the past week. We've learned that the prices on exchanges don't have to always go higher. In the next month ATI won't sell more Radeon HD GPUs than have ever been made. It will take a big bump in the exchange rates for people to go gung ho in buying new mining hardware again.

Yes, that was truly incredible accuracy on your original forecast. I did not believe it when I first saw it. It seems to hard to predict. I think it would have been right on the nose if not for the mt gox incident.

chodpaba if you can't get wealthy somehow with your knowledge something is wrong with the world and efficient information markets theory

Indeed. When bitcoins take over the world, chodpaba will be elected supreme ruler.

Very interesting. Ironically, it seems like a good time to invest in mining equipment, as the difficulty shouldn't be going up much in the next month, but if people did that, then the difficulty _would_ rise.

Then again, it is probably an even better time to just buy if you are interested in investing. The price:difficulty ratio indicates that compared to the past, it is a good time to buy, not mine.

I'm guessing this means you don't predict a big price rise soon. Late August?

.... even a Lord of the manor has a price that will get him out of bed ....

yep, the market is digesting, the flat periods are essential, the rallies (or dumps) are just the visible evidence and final outcome of all the mulling that has gone on for some period beforehand ..... a trigger makes them finally make the move on a price that they had already decided on days, weeks or months ago, trigger dates are an interesting angle of analysis also, seasons, moon cycles, tax dates, etc.

... think a drop to 9-10 area is on the cards before the big roller gets up steam again.

After the flurry of news articles the quietness is interesting, lots of normally chattery people are mulling what the heck bitcoin means for them, if they haven't got an answer they like they'll be buying, "can't beat 'em join 'em" types ...

Holy, awesomeness.. =)
If I may, I'd like to make a very uneducated perdiction;

1.65-1.7mil at next retarget and continuing a bit up from there to 1.71-1.85 at the following retarget...

wow, .9963 corelation

Have the methods you are using now brought that any closer?

Blocks 133056 -> 135072: 12 days 8.8 hours

New difficulty-equivalent hashrate from first 48 hours of this round: 1581572 -> 1563027, -1.2%

Current trading price: $14.25

Buy energy hit stiff sell order resistance at $16, and stagnating at that price was enough to trigger some more cashout. I don't see publicity driving more get-rich-quick money in, as markets have settled in since gox 2.0 and there is not much to report. Mining with new hardware is a break-even proposition. We are approaching the 'viable currency' phase of bitcoin's life.

If difficulty can be constant for a while, miners won't see exponentially diminishing returns looming on the horizon, and may gain confidence in investing in hardware equity that won't turn a profit for three months. I see the difficulty growth barely on the positive side in the next two cycles, with about 50% profit-taking, 40% holding, and 10% reinvestment after, and new mining stronger in October when energy costs are free instead of double (heating your home in winter instead of making your air conditioning run in summer).

nice nice. I see it brings it closer to my original guesstimations. Though I see you now show a drop for 139104. Care to elaborate?


  One other factor I am having a hard time with is the botnet factor. and no, not the pools trouble of last week for anyone else thinking that. Just that keeping track of the total amount they are adding into the net is a pain since they are much more erratic than the pools. Or, atleast they were until they decided again it was more profitable to find a better way to 'fit in' than it was to ddos everyone.

  I am sure you watch the network total hash like I do. I noticed a decline after the last retarget had normailized from its usual dip. It was down by about .8 - 1.2 Thash that I could not account for with normal fluctuations, checking the big pools were not temp out and casual miner dropouts. This was from about 12 hours after retarget up to about 16-18 hours ago, where at which time it grew from the back to normal rate to what it is now.

  I'm having a hard time justifying that with gains from new onlined hardware, etc. Obviously some percentage up is expected and we know even with the difficulty even small ops are still onlining. But, if I may risk putting a guess on it, it seems about 20% of the new growth is not user generated....  I know some of you are gonna see 20% and think holy crap, the zombies are gonna eat our brains but let me repeat I am only refering to 20% of the 800-1200Ghash growth or roughly 160-240Ghash worth that I suspect of currently being botnets....


edit; just wanted to point out that I say 'guestimations' because thats just what mine are. I am certainly not doing any where near the lvl of calculation you are to come to your numbers.

edit2; same with my nethash numbers. Though I feel confident about my bot figures they are derived from semi-casual observation and math done in my small brain.

I will disagree on both predictions by chodpaba. Next difficulty will have a very hard time to get as high as 1.79 mill. Difficulty after that will have very hard time to drop as much as 1.64 mill. I think a more natural growth of the network will take over and we will go higher at a slower pace.

I predict 1.7 mill next difficulty and 1.8 after that followed by close to 2.0 mill.

Sorry, you are right. I was thinking too much that your median prediction is more of a hard target.

I missed by less than 1%.

chodpaba, I think your formulas are skewed because of the major disruption in June. I will stay with my prediction, next difficulty will be 1,8mill, to be more exact 1,798.900 :) I think network will have a very hard time hitting even your lower quartile at 1,82 let alone median at 1,98. There is almost no way network could grow 18% in the next 11 days. I understand that my prediction at 1,8mill again falls within your range, but your full range from 1,75 to 2,22 mill is just way too wide to have any meaning at this point. We are only looking up to 14 days ahead and your full range from lowest to highest is 30%.

If your next target has a range of 30% there is no way EVER you could predict next re-targets with any accuracy at all, let alone 4 targets out.

This is positive and negative criticism.  All this statistical terminology and formulas are nice but not very effective. Not so far back you had negative number for 139104 block. Around 1.56 mill, now you have it up to 1.93 mill. You keep tuning your formulas but the results are just way off target.

Natural growth of the network is between 6% to 12%. Unless something major changes in bitcoin usage which would of course dramatically change its price, network will grow at its natural pace of 6 to 12%. For major disruption for this to change price needs to raise or drop more than 50%. If it would drop 50% network hashrate might fall within -5% and +6% in case of price raising 50% network could grow from 6% to 25% depending on what the actual major change would be about. It's that simple.

Everything else is just some mumbo jumbo useless statistical formulas that so far look as accurate as throwing a dart. Your NEW 4 targets will all miss by a lot  but will still fall well within your widest range possible. Anyone can predict that. My prediction of 1.7  1.8 and 2.0 will be closer and it took me less than a minute to figure it out.

I don't stop by and post much..... I'm one of those long time readers that does not get around to posting much.

Anyway, just wanted to say that I like looking over your extended difficulty forecasts. Its definetly not a perfect system but at least your taking the time to try and refine your approach and even willing to admit your mistakes.

5.4% variance on a market as volatile as bitcoin is pretty good in my book.

I was talking about natural growth of the network, at times where there is no major price change. You did the math for the most volatile month and a half bitcoin ever witnessed. This is why 6 to 12% in that case wouldn't work at all. Still way better than your crazy 3 +mill  prediction in post #48 and later on network drop.

Anyways, I'm not here to argue or anything. My initial post in this thread was that I disagree on predictions, and I stated my own 3 predictions, of 1.7mill 1.8 and 2.0 mill.

I don't have a beef in this discussion, but I really wanted to comment on the way you present the results of your statistics. You are doing this in a really dilettantish way. What skyhigh was proposing was something like:

1.84mil (give or take 0.03)

Skyhigh is actually even more clever, since he talks about only two significant digits.

Your constant talk about quartiles and other advanced stuff combined with your unwillingness to share a model puts you in one of two classes: (1) you either deliberately lying or (2) you completely misunderstand what the statistics is all about and doing an arithmetical equivalent of making predictions from the tea leaves left in a cup.

Statistical theory is one thing, presenting it properly is the other. And I wrote here because I want to care about your results.

my prediction is more accurate than yours.

It's going up. 

Behold the power of the austrian school of economics with its unflappable qualitative predictions!  ;D

bro, you are feeding the trolls on your own thread. you'd be better to ignore them I believe...

Chodpaba, I'll make one more post. From the picture above I can't figure out whether you are controlling the gnumeric or the gnumeric is controlling you. I see a lot of "tea leaves" on that picture, but some significant digits too. I noticed that you are most likely using Linux, so I have the following one-liner for you to run:
I hope you'll recognize which digits are significant and which are "tea leaves".

I don't know why you wrote this thread and what is the purpose of your computation. All I can suggest is to get some better tools: (1) http://www.r-project.org/ (2) go to CRAN (3) download the main release (4) also from CRAN get the package "Zelig: Everyone's Statistical Software" (5) from the Zelig's website download its introduction: http://gking.harvard.edu/files/z.pdf .

Zelig will be as objective as it takes, and the knowledge you'll gain will be yours for the rest of your life.

lol I never thought of it that way but I guess the austrians are the rolls of the economics branch of science.

Why you bother with 4 re-targets, if you always change them at least once before they happen. My dart throwing 1.7 1.8 2.0 mill will be closer on average than any 3 you will ever be able to calculate and NOT CHANGE at least once. So far you predicted / calculated re-targets but you soon discovered a better model or something else came up and you came out with new and new numbers. Calculate 3 re-targets and stick to them. I predicted my 1.7 1.8 and 2.0 on July 9th and I'm sticking with it. You changed your July 9th prediction at least 6 times by now. If I didn't write this or if you won't change the way you do your thing, you will keep adjusting and changing your future predictions.

Seriously, you've made your point a dozen times. You disagree with OP's method. Fine, move on.

I apologize of this was asked before, but are you combining NMC and BTC total power for the forecasts? It looks like there was a spike when people switched back from NMC a few days ago when the difficulty increased.  Its possible that the combined rate is not significantly changed, just the focus of which blockchain people are on.

There has been some controversy about your predictions, but I, for one, apreciate them.

So you say, but we see how accurate you are :D :D :D

j/k

Or maybe it just looks that way at the moment because of coincidences. If you check for enough patterns you are bound to find some that fit, but that doesn't mean they'll continue to fit in the future.

I dont really believe in the statistical methods, but I too apreciate your threads. They are fun to read. Its all good speculation, and sometimes you can find interesting correlations.

Btw, I think you have a very difficult task because in general you are trying to predict future behaviour with the experience of past behaviour, but since Bitcoin is so new and growing so fast the patterns can change quickly and radically.

So what's your guess now? I'm 100% certain it wont be 2.38 mill next difficulty change.

my next two re-targets

1758900  followed by
1706000 - 1746000

Because of a massive selling and price dropping below $6 miners started leaving or are being replaced by less and less new miners. If the price will stay above $10 difficulty will most likely stabilize around 1.7mill to 1.8mill for the next few re-targets, but I think its very unlikely that price would stop being so volatile all of a sudden. If price would drop below $9.00 I expect network to keep losing 3% to 5% miners which would result in difficulty dropping at that same pace. If we get a price to rally above $16 I'm expecting 3% to 6% increase in difficulty.