Topic: [PAID] Publish result of Days Destroyed calculation - page 2. (Read 11617 times)

Well, looking at the Block Explorer for block 65711, it seems that the previous block 65710, which holds the redeemed output, actually has a timestamp that is later than block 65711. So, apparently the calculation is "correct", but I honestly don't know why the dates are backwards...

maybe I'm missing something but are there some errors in these numbers for BTC days destroyed? Some blocks have negative values for days destroyed.

For example see block 65711 has -151.03 BTC days destroyed. There are 48 other blocks in this csv file with negative numbers for days destroyed.

Very interesting data and analysis by the way. thanks to all concerned.

The downside of days destroyed is that it is releasing more funds into what is presumably, an active economic exchange pattern.  Unless whoever receives those coins continues to hoard them, it will ultimately DECREASE the overall value of bitcoins, since there is now more "active" supply of them.

I'm working on some ways of detecting which transaction was the change. In many cases this is trivial.

This BitcoinDays Destroyed number still doesn't provide a perfect picture, it is just a much better measure than looking at raw spending totals.  Here's an example:


Consider a slight variation on that ... example 5:
5. Someone who mined 50 bitcoins exactly one year ago spends 1 BTC.

In Case 5 only 1 BTC was spent and 49 BTC was returned as change.  There would still be 18,250 Bitcoin-days destroyed, even though the coins representing all but 365 of those Bitcoin-days remained in the same wallet.

I'll try to explain by examples.  Here are several transactions.

1.  Someone who mined 50 bitcoins exactly one year ago spends them.
2.  Someone who mined 50 bitcoins yesterday at this time spends them.
3.  Someone given 50 bitcoins today spends them 12 hours later.
4.  Someone moves 50 bitcoins from one wallet to another (but they own both)

In case 1, you can examine the block and see that the last time this particular money was touched was 1 year ago.  So that's 50 coins, times 365 days that have been "hoarded", but now are spent.  So that's 50x365 => 18,250 Bitcoin-days destroyed.  That's good.

In case 2, the coins weren't hoarded as long.  50 Bitcoin-days destroyed.

Case 3 might be the "same coins" as case 1 or 2, only they circulated again.  But since they recently circulated, it's not nearly as good as 1, and only half as good as case 2.  25 Bitcoin-days destroyed.

Case 4 is the odd case, because money didn't really change hands, but you can't tell that.  It still counts as "circulated".  In one sense, it's an indication that the bitcoins haven't been lost, so it's a good metric.  On the other hand, moving money from one pocket to another really isn't a good measure of economic activity!


Does that help?

If it's not too much trouble, could you explain how and why?  I'm still very confused, but I like the concept (or at least the vague notion of it that I have, which you've stated is clearly incorrect).  I very much want to understand, and I'm sorry for being a pain.  I just want to understand but I don't.  I feel terrible and idiotic and I don't want to be wasting your time like this, but please, please help me understanding how this metric works and what it is.

BTCDD is better precisely because of it's difference with something like you describe. It gets at the difference between Satoshi unloading and noob shipping coins around all day.

That's not possible to know, and it doesn't really matter anyway.  If I'm understanding it correctly, it's just a velocity of money metric to gage market spenditure versus hoarding.

I feel like BTC Days Destroyed should be able to be roughly calculated (not exact) by the average number and size (in btc) of transactions per block and the average interval between blocks, over the entire life of bitcoins.  Like couldn't I just find the total amount of BTC traded per block, and get how many "bitcoin blocks" instead of "bitcoin days" by using the summation forumla:

(Blocks^2 + Blocks) / 2

And then just figure out the average number of blocks per day.  Those figures should give a basis of measurement of btc days destroyed should they not?  Is that true, or am I wrong about that?

If I am correct, could somebody post the (roughly accurate but not completely accurate) formula for btc days destroyed, since I can't read python!..?  Thank you!

I didn't invent the Bitcoin-days-destroyed metric, someone else did.  I was just trying to interpret it.
My addition was to make it a cumulative value.  All I did was take the spreadsheet that other people produced, and summed it up, and calculated a percentage off of the total Bitcoins in circulation since the beginning of time. 

Regarding a bubble - that's my interpretation.  I could be wrong, but it seem like when people start hoarding, that will cause prices of Bitcoins to go up irrationally, bubble-style.  And the Cumulative Bitcoin Days Destroyed metric seems like a decent metric to determine the level of hoarding.


Your concern is definitely a limitation of the metric. 

Another interesting consideration is this:  Since the spending algorithm always favors spending smaller "denominations" first (if I understand it correctly), this means that as the old-timers who are active in the system will generally be spending "new money" until they get some coins in big chunks.  For example, an old miner has a bunch of 50 BTC units in his wallet, from his mining success.  And as long as he has tiny units (1 BTC, 5 BTC, etc), the old mined 50 BTC units never get spent.  But if he gets a large chunk of money that's been actively circulating, say 100 BTC, then this will be hoarded in his account until he spends all his 50's.

So I think cumulative bitcoin days destroyed is a pretty decent metric, but certainly not perfect.

How can you know that most Bitcoin days destroyed aren’t just people moving old coins between their addresses?

Maybe, maybe not. I'm working on a related project right now (which unfortunately nobody's offered any significant bounty for) which may provide some further insight into this.

One can argue (and it probably works perfectly on the stock market) that a large value of bitcoin days destroyed may mean that the Bitcoin "founding fathers" (insiders) start to run for the exits.

Yes, but we don't live in the block chain; we measure time by hours and days, not blocks.

You know, it would probably be far more convenient to work in units of bitcoin-blocks, instead of bitcoin-days.  So if B is the current block number, and the total number of bitcoins is 50B~6 million, then every new block adds another 50B bitcoin-blocks.

Then you can trivially see if a block's net effect is to add bitcoin-blocks to the total, or subtract.

This metric is weighted by the amount of time someone held a bitcoin before spending it, something that it's never before been possible to measure with any other currency. If people hold more Bitcoins over time, then it will go down; if they spend, it will go up. If they hold for a long time and then spend, it will go WAY up.

jerfelix - how did you figure this out, and how did you figure out that is when a bubble is forming?

Interesting metric - thanks for the explanation!

Perhaps this stat could be added to this page? http://blockexplorer.com/q