Effective stat file times and the consequences

[AJMSmith]

I know that the files generated by Stamford are timed at or about 0:00, 3:00, 6:00, 9:00, 12:00, 15:00, 18:00 and 21:00 California local time (hereafter CLT, which is either PST or PDT depending on the time of year) ... but with the source data for the file being only updated ever two hours this means that every other update file contains data that is an hour older than the ones in between (the difference between the files is alternates between the effect of the addition of 2 hours of results and 4 hours of results.

The updates by Stamford start just after the even hours of the day (ie 0:00, 2:00 etc CLT) and wind up with the data for the period up to almost an hour later ... results arriving while the update is progressing are included ... so the stats on the site are those as of 1:00, 3:00 CLT.

The means in turn that the 3rd party update files should be regarded as being timed at 23:00, 3:00, 5:00, 9:00 etc CLT.

There is an obvious consequences of this with catch up times being more accurate (the present way has catch up times shifting back and forth by an hour on each update) ..

Posted to the general list as this is not IMHO a bug but more in the "food for thought" category ...

Allan

[Kakao]

The means in turn that the 3rd party update files should be regarded as being timed at 23:00, 3:00, 5:00, 9:00 etc CLT.

That would be more precise. Now how can i be sure Stanford will not shift their update routine? Say there is some hiccup with the Stanford system and then they change from 2/4 hours to 4/2 hours or to 3/4 hours or whatever. Should i check every three hours? Shouldn't it be simpler if the Stanford file timestamp reflected the real stats gathering time so the 3rd party systems would be correctly fed? No need to change a single line of code on this side of the fence.

[MoneyGuyBK]

Stanford 'Used to' update their stats once 'Every Hour' up to last year.
Then (Probably because of work load increase) they changed the updates to 'every other hour'

As of late with a flood of server issues, they have had difficulties keeping up with this.  At times I see them updating up to 45 minutes later than their own schedule, and at times with servers down no updates are given (although the last instace is still very rare)

I used to check the stats every hour myself, now I may check them once every 3-6 hours... this way I could (hoping) keep my hair from going gray






Peace

[AJMSmith]

Possibly is could be down to Stanford to show the true time in the file ... at present they show the file creation time ... it might be that the update time would make more sense ... that would solve the problem of what the real time of the update was ... as it would now be in the file ...

I'll put a suggestion on their forum ...

A

[Miller855]

I Don't have a problem with 3 hour updates, what i have a problem with is, sometimes my WU's  posted b4 20:00 UTC get credited for that day, but sometimes, they get pushed to the next day, why is the daily cutoff time not consistent?
Everything else is great !!! i like all the detail reports, it's just some details get me wound up.
THANKS FOR ALL THE GREAT THINGS YOU DO !!!