RE: http://roelandp.nl/bitshareswitnesslog/ (Great site BTW):
Would it be possible to indicate historical reliability of price feeds? Perhaps the avg variance and the avg time between price feed updates over x days?
Thanks
I could also start calculating variance difference for every publication. However variance is a subjective thing: A witness can post a high variance pricefeed because the markets are suddenly moving into a new direction and others have not yet responded with new price feeds, so the median is still reflecting the 'old' market and will gradually change when new pricefeeds come available. Average variance over a rolling time will therefore be all in the same range I think.
Yes I can add the 'publishing frequency' as I log all unique pricefeed entries. Feed frequency could be interesting, but can be a subjective indicator as well, misleading. One witness cannot update pricefeeds as long as markets are not volatile and stable without many changes... Pricefeeds are based on feeds published with 24h window, and if markets are not changing, there is no need to publish new feeds and therefore this influences the average pricefeed publication frequency number. So there are not really conclusions to draw from...
Thanks for your feedback
@Customminer and please take a minute to check out the recent GUI changes of the tool.