BuzzerBeater Forums

So are we going to look at this or is it normal?

Something IS f***ING broken, and as usual nothing will be done. I won against first team last match and this evening i have a very low attendance against 2nd... Not Logic at all.

Are you still looking at this issue? I'm certain something is up. I've won 8 of my past 10 games and my attendance for the past 3 home games has been 80k lower than usual, and it's still dropping. The games were all against highly seeded teams and in the same period, league averages have been dropping too.

Is it possible that the missing Fan Surveys from earlier in the season are still having an affect, because that's when this problem started happening.


Last edited by Slug at 2/8/2020 9:29:16 PM

Can you help mate, what is wrong with the 3 last income from home games?

(http://prntscr.com/qzuffx)

I still can't figure out what's causing the discrepancies.

Full disclosure, the statistics are not helping much, so I've decided to open the issue to suggestions. I've added a attendance averages table to the (/community/overview.aspx) (might move it, probably not the best place for it). You can also see the merchandise averages there too.

To me, it looks like matchday 2020-02-04 definitely needs to be recalculated, unsure about others. Also, merchandise for the last two economy updates need to be recalculated. Please take into account that matchday 2020-01-11 is probably too high due to a bad recalculation on my part. I'd love to hear some thoughts.

Also, I have the same averages in more detail (by division) but it's probably better not to list them publicly as they might disclose too much about teams that are alone or almost alone in their divisions, but I can look into them if needed.

Last edited by BB-Marin at 2/17/2020 9:53:20 AM

I am pretty sure you had already done this, but just in case.

I would pick the most extreme case or cases, and look everything really really neetly manual calculation. Somehow i have a feeling there might be mismatched integer types or some other silly stuff, maybe bad memory allocations. When it is random then it is extremelly hard to try find any hooks, maybe we could try to look for the earliest that type of report and look what changes had been done before, maybe they had some influence.

Games are calculated in batches, grouped 15 minutes before each starting time slot, at most 1000 at a time. Human vs human matches are calculated first, followed by human vs bot and finally bot vs bot. Rarely is there a possibility of some bugs or other deterring factors affecting less than a full batch. This might shed some light, if I can find which timeslot/batches have been negatively affected.

As for merchandise, most of the merchandise related data is calculated before the economy updates even start (economy updates are calculated country by country, following the same order as the matches). These updates (merch related) are global, however it is quite possible that they break mid-calculation, that is not the most likely scenario. Usually, if something breaks, it breaks for everyone. This then results in quite obvious merch drops that can then be quite easily identified (especially now, using the new table).

I'm explaining all this so it would be clearer how the backend works, enabling you to understand the mechanisms better, ultimately, hopefully, with a larger set of eyes we can find the cause faster. At least, it might help identify the problematic areas.

As for your specific suggestion, I will do something of that sort, considering all of the written above.