4 replies [Last post]
LociOiling's picture
User offline. Last seen 4 hours 2 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders

I was never very clear on how player ranks were determined. So I did some research and created the Foldit rankings wiki page.

My big discovery is that the main rankings are based on your total "global points" in the last 120 days. I'd previously thought that rankings were based on the last 40 or 50 puzzles. (On the other hand, there were 49 or 50 puzzles completed in the last 120 days....) Feedback 989490 documents the 120-day window, and also the 30- and 7-day "month" and "week" windows.

There were some other discoveries. I found the global points formula is documented in the technical section of the Foldit FAQ. But what's not mentioned is that the number of soloists competing is used for both the soloist and evolver rankings, while the number of groups competing is used for the group ranking. The smaller number of competitors for the group ranking means that the points awarded drop off much more sharply as your rank decreases.

Another finding was that beginner puzzle points do not count toward group ranking.

Players are also ranked by category. Each category has its own rules for tabulating scores, such as "your best 8 out of the last 10 puzzles" in the prediction category.

I've asked Inkycatz to update the FAQ with a mention of the 120-day window used in rankings.

So far, the 120-day rankings I've looked at have checked out OK, in that the total global points reported on fold.it match the results for the individual puzzles. I haven't been able to successfully verify my own "prediction" ranking yet, but the number I get is at least within 5 points of the number on the website.

I also want to look at group rankings. I've noticed that the numbers reported on the groups page don't match the numbers on each individual group's page. For example, the groups page reports 3,754 points for Beta Folders and 3,065 points for Anthropic Dreams, but their individual pages show 3,115 and 2,926 points, respectively.

I've asked Inkycatz to update the FAQ with a mention of the 120-day window used in the main rankings.

LociOiling's picture
User offline. Last seen 4 hours 2 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders
category rankings seem good

Looks like the category rankings are correct, despite my previous confusion. The trick is knowing which puzzles count. The web site shows the list, at the bottom of each category page. For example, on the prediction page, scroll down to Puzzles -> Closed, then click "show all". This shows the 10 puzzles that contribute to the prediction category at the moment.

As it says at the top of the page, the totals shown are for "your best 8 out of the last 10 puzzles".

The last 10 puzzles are the ones shown, and they're the same for everyone. I didn't play puzzle 1203, so I wasn't including it in my count. Once I had the correct 10 puzzles, my spreadsheet total matched what's shown on the prediction page.

LociOiling's picture
User offline. Last seen 4 hours 2 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders
group rankings are mostly OK

The rankings shown on the groups page seem correct, although the current 3,754 for Beta Folders and the 3,065 for Anthropic Dreams include puzzle 1169, which expired 121 days ago at this point. We'll see when the scores get recalculated, but good enough for the moment.

As noted in the original post, the global points shown on the individual team pages don't match the global points on the groups page.

LociOiling's picture
User offline. Last seen 4 hours 2 min ago. Offline
Joined: 12/27/2012
Groups: Beta Folders
group rankings unchanged after puzzle 1218 closed

The group global point totals shown for Beta Folders and Anthropic Dreams didn't change after puzzle 1218 closed. Updates to group rankings must happen on a "lazy" schedule.

Joined: 09/24/2012
Groups: Go Science
Group score mismatch

"The numbers reported on the groups page don't match the numbers on each individual group's page". Indeed.

Is it a bug or is it a (normal) concept mismatch?

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons