Are there mechanisms other than scoring corrections that could cause this behavior? In this matchup (Ottoneu Fantasy Baseball - game - Fantasy Aceball Invitational), I looked through the week’s scoring corrections, and all I could find was a change to Griffin Jax that would have lowered the score of the team that now has more points (so the opposite of what I would have expected).
I have a sneaky theory here, and it seems to hold up based on reality.
- There was a suspended Mets-Nats game on 5/13. It resumed and completed on 5/14.
- As a reminder, this is how suspended games work
- The team that scored more points as of today had Hunter Harvey in their lineup on 5/13, who scored 14.33 points in the resumed portion of the suspended game
- There is some confluence of a race condition between (and try and follow along here, this gets stupider):
a. When the complete 5/13 data came from our data provider, including the final bits of the suspended game that ended on 5/14
b. When we incorporated that data (going back and fixing 5/13 with the newly final Mets-Nats game stats)
c. When standings are updated because the week is ‘over’
So this required a suspended game to be resumed on precisely Sunday and have players in this H2H matchup play in the resumed portion of the game (since we had the 5/13 stats pre-resumption already counting) and score enough points to bridge the point gap between the two teams. That is a very specific thing to happen!
In the post I linked to about suspended games, I wrote “Any final H2H results will not be changed.” This caveat exists because sometimes games are resumed months after they were suspended rather than the next day. So what I’ll do here is as follows:
- Look at the timing of these various processes and see why they are happening out of order, i.e. the standings stuff should happen after we run stat corrections for the previous week.
- Probably, if the timing of these processes holds up, change who won and lost this particular H2H game.
So for now this is a theory, but it is a pretty strong theory with some pretty strong evidence, and I’ll report back shortly.
3 Likes
Alright, confirmed confirmed.
We finalize the week around 4am ET on Monday. However, FanGraphs’s data load happens between 4am and 530am ET. So we need to update stats after this data load and then finalize the week after we update stats.
The reason we were finalizing the week this early is because people start visiting the site and wondering why they are seeing last week’s games and standings haven’t updated yet. I think getting all these stats correct is obviously more important than a little extra clicking on Monday morning, so I’ll change the timing of our final stats update and when we update standings to accomodate the FanGraphs data load timing (which I have no control over).
Oh, and I’ll flip the result of this game as well.
3 Likes
Fantastic. Thanks Niv! You will have made one manager very happy (one less so).
1 Like
Should be all set now!
1 Like