I’m trying to follow IRI, and something keeps coming up that really bothers me. It has been a problem at every competition I’ve ever been at. Here it is:
Every team’s “Rank” is set by “Rank Points”, rather than “Rank Points / Match Played”. I know it makes no difference come alliance selection, when everyone has played the same number of matches, but until then, it’s really confusing to see teams slowly drop down the ranks between matches, and then shoot up to the top temporarily. Bother!
What would it take to change the “Rank” value showed by the official stats to the match-normalized calculation?
This method is a lot better for seeing how many points you need to get in order to jump in the rankings. “We need 3 points to jump Team XYZ” is very useful.
Having two fields, one for actual rank, one for rank average, might work I guess.
I tried to adapt MLB’s “games back” metric in the spreadsheet I used for my announcing. I’m convinced that it can work, though not that I did it yet.
I used RP since wins isn’t as valuable a metric.
If PRP is the total number of ranking points possible (4 * number of matches played) for a given team, I computed (2 * PRP - RP) for each team. I subtracted each team’s total from the leader’s and divided by two.
This might be entirely equivalent to dividing a ranking score by the number of games, or there might be more happening there, but it does make those teams that are trailing in the rankings because they’re a match behind pop out without a lot of searching.
How about instead of changing the ranking system we color code each line on the rankings live. There are plenty of ways to do this but I would think something like Green colors on the board are the teams that have played the “latest” number of matches (say 5 matches), orange is for teams that are down below that a match and are not current (4 matches played for example), and red for teams that are down two matches from the current (3 matches played) or have been carded so that their ranking is considered already.
Once all matches are played for that team turn the color to solid black.
I think it would be an easy change and a visual indicator that would fit the ranking style of any FRC game. There may be more intelligent ways to color code the rankings to mean something different that you guys can figure out I am sure.
Even if this is not implemented in FMS, it may be worth requesting from the folks at TBA or Spyder. This would make figuring out how far we’ll move in rankings a lot easier.
There’s some great feedback in this thread! One reason we’ve been hesitant to modify the ranking display too much in TBA is that we don’t want to cause any confusion about which columns are “official” and which ones aren’t. However, I definitely agree that there are things we can tweak to make the display much more informative (we just have to clearly differentiate which numbers come from directly from FMS and which don’t)
I think we can add a notion of “derived columns” to the rankings display on desktop: we can show a “match difference” outlining how many matches away from the max each team is (so it would show {-- (even), -1, -2, …}). This could go in parentheses after the official “played” column. And we can also add RP/match.
This all gets a little trickier on mobile, because information density is hard.
If anyone would like to help build this for TBA, let me know and I’ll help you get started! I’ve filed an issue to track the feature
That’s awesome. Thanks Phil! My web programming skills are not up to the task myself, but I hope someone rises to the challenge. If the derived column(s) prove useful, we will have a clear example to show FIRST HQ to request an official change.
What if there were two rankings? One would be the current ranking system, while the other would be teams ranked by how much they score in matches? It could just be based on their average score/alliance contribution in the matches they’ve played.
OPR estimates a team’s own contribution, minimizes the residual in match scores, and relates to the point totals on TBA by a linear transformation. But not quite the same thing.
FIRST regularly reports the sum of your alliance scores across your matches, broken down by area (auto, different ways to score in teleop), though this is not normalized by matches played. However, this is not typically used for ranking (with the notable exception of 2015). Even as a tie breaker, the component scores are used (usually auto first) rather than a total.
In a way, but making it more accurate, and sort of a combo of OPR and DPR, so if a robot played defense, that team’s ranking/contribution could be calculated by the number of points they stopped the opposing alliance from scoring. Both could be used in one match if a team scored for their own alliance and also played defense.
For what purpose? To try to estimate what a teams ranking will be or to try to assess how good teams are? If it’s the latter, TBA already as OPR somewhere for the top n teams. If it’s the former, OPR (or whatever modified version you like) has absolutely no correlation to the current ranking system. It will also never become an official stat, let alone ranking method, by FIRST because it completely goes against what FIRST believes in.
Yeah the OPR/DPR/“any Rating” does not correlate to ranking for many FRC games at all and it makes it useless from year to year usage. Especially games like 2015 and 2001 where your alliance or average scores come into play or any games like 2012 where you have coopertition considerations.
Any adjustments to how it is presented visually I think will take away most of the confusion associated with how your ranking will swoop after a good win on your last match in a situation where the rankings are really close and nobody else has played their last match.
Then, you get to watch in awe as your rank dips until everyone has played that last game. I have seen kids celebrate first place for one moment and then get immensely upset 5 minutes later at being 9th seed suddenly at the end. Those of us who are accustomed to it don’t see the problem but for people new to FRC it is confusing and often upsetting to the point it has probably ruined some people’s experience. It also makes scouting more difficult and more important for the draft order.
No amount of teaching or training is going to necessarily fix that confusion so I am open to hearing more suggestions.
Being able to view RP/match is something I’ve wanted to be able to do for a while, so I decided to take action on it.
On The Blue Alliance, I implemented ranking points per match as an additional column in the rankings for all years 2007 - 2016 (except 2015, which already ranks using an average). It may take up to 24 hours for the update to appear on all events. Here is an example if you’d like to see it. I hope this helps!
