By coincidence the CDF value is the same for the last 10 and last 25 blocks at 0.0159.
That's getting down there but we aren't at discovery level yet for the existence of "Luck".
I want to see 5 sigma on multiple independent sets of blocks, so you should be able to work out limits on my prior for PLuckExists
kano, is it possible to change the calculation of CDF for the last n blocks to include diffs submitted since the last block?
With the current implementation, the last 5 blocks for example would stay bright green even if we never found another block...
Yeah I'm betting on that not happening ... ... ... I see the pool differently to you.
From an accounting/profit analysis perspective it's a better representation of how a miner has done from (now) to (now-nblocks) if we include the non-productive time between (now) and (now-1block). edit: The non-productivity of inter-block time is a correlate of the independence of future outcomes from past.
If your accounting matters on the luck of a single block, I suggest you need a new job.
That table is there for that exact reason, to see how the pool is performing over time, not guess based on a single block.
Single block luck is so random that you'll be out of business pretty quickly if you base your decisions on that.
Hmm back on 24-July last year I can see you closing down your business when we found that 666.666% block
Edit: oh and of course a 666.666% or worse block is expected, on average, once every 785.8 blocks, so you gonna be going broke pretty regularly on larger pools.
edit: From another perspective, this also has the benefit of making the CDF values over time roughly continuous and differentiable and eliminates the variable phase shift of the current filter. For visualization and analysis this is beneficial.
Thoughts?
No, since that is including the completely unknown 'future block' into the calculations.
That's a table of facts, not predictions.
The header of every page on the site shows the current Diff% of the future block at "Shares:" and the time since that last block at "Pool:"
Regarding being continuous ...
That future prediction will also change with every single share submitted, so currently about every 714 nanoseconds ...
The blocks table information is generated once every time any code requests it and finds that it needs updating.
That's (normally) twice per block, once when a block is found and once again when a found block stats are corrected.
(when the ~ goes away)
alfonso just a short rental ....
