Yesterday Seeker asked about not being in the newbie league anymore, this is consistent with the 18-month period, but:
- in the season standings (http://zak.stunts.hu/index.php?page=season&season=2019), he has been wiped out; Heretic is listed as taking 12 points in C210, while in fact he took only 11 (http://zak.stunts.hu/tracks/ZCT210);
- a similar phenomenon also happened to me in 2015, but there is more: Cas is listed as participant in the season standings (http://zak.stunts.hu/index.php?page=season&season=2015), while he does not count as a newbie in any of the races! This can have something to do with the fact that he registered in 2005, but he posted his first replay only in 2015, and maybe the 18 months are counted starting from a different date in the two cases.
How do we have to count correctly the points? Note also that Marco joined in January 2015, but according to season standings, he took points in the newbie league until the end of 2016 (that is, for a period longer than 18 months), and FinRok joined in March 2016, while taking points in that league until the end of 2017 (also here, for more than one year and a half); in both cases, the results in the single races are correctly counted only up to 1.5 years. Maybe the latter ones make more sense...
To be honest, I have no idea how the newbie league code works, but it seems to cover all the pipsqueaks for which...
pipsqueaks.joined + INTERVAL 18 MONTH > season.year . date('-m-04')
The code refers to the 4th day of the current month, whenever the season scores are recalculated.
I suppose that was meant to say "all the pipsqueaks who joined within 18 months from the start of this race", but races don't start on the 4th of the month anymore, so that can be wrong of up to 4 weeks. And if anything changes and we recalculate the scores for an earlier race, the set of newbie pipsqueaks could be disturbed... leading to a messed up scoreboard.
So I'll likely have to fix the date filtering, then fix the whole "recalculate for a previous race/season", then recalculate the past five seasons and hope that scoreboards don't change too much :'(
Hmm, I see. As for winners, Marco would step down in 2016 since its last races would not count, but there should not be other changes; more differences would arise in the podium, though.
These should be the right computations for the period 2014-2019. I think it would make sense to update the wiki with these, along with a brief explanation.
Nice work investigating this, afullo. On a related bit of context (that you might be aware of already), IIRC the window is 18 months long so that a newbie who doesn't miss races will have at least one season with nine newbie league entries (9 races + 3 discards = a full season).
Thanks, Duplode. Yes, I saw that the idea was something like that; as a newbie, I raced even 19 times, since the full month of August 2015 was considered part of the period. This should be possible also with 28-days races and four hiatus weeks, even if I don't know whether you can be not considered a newbie anymore during a race, being the majority of them spanned over two calendar months (see also dreadnaut's post).
2020 update.
2021 update. This year the seasonal scoreboard and the racewise constructed one do agree.
Good one :) And you're back to "ods", better than "xls" ;D
As part of the Amateur league work, I am rewriting most of the scoring code. Hopefully Afullo won't have to do this manually anymore.
At the same time, these tables will be essential to double check I've not messed up the old data :D