Talk Fantasy Football

mubo wrote:But I guess from your perspective you'd rather see mu/phi than last month's change in mu/phi?

That’s right. Since it’s changed to last month, I totally lost the feeling for comparison with “dead” teams. For example Floridaboy (NAF #6636) has1396.4 Elven Union Ranking with a decay of -0.8. He played 6 matches March 2007! Over 12.5 years ago! Shouldn’t his rating be rock bottom?

And could change “Elves” to Elven Union to separate them from the 3 other Elven teams?

Anubis wrote:

mubo wrote:
That’s right. Since it’s changed to last month, I totally lost the feeling for comparison with “dead” teams. For example Floridaboy (NAF #6636) has1396.4 Elven Union Ranking with a decay of -0.8. He played 6 matches March 2007! Over 12.5 years ago! Shouldn’t his rating be rock bottom?

And could change “Elves” to Elven Union to separate them from the 3 other Elven teams?

Just to be clear- the rankings themselves haven't changed- just what I choose to display on the NAF site. The decay is points lost in the last month.

Re: your example- it is rock bottom. The player is unranked (phi > 100).

And, yes I could change elves to Elven Union, will do that next release.

sann0638 wrote:Great - thanks. Worth a line in the page that says top 25000 are shown, or similar? Saves people like me asking the queson

I don't know if this was actioned, I didn't see it, but I was going to ask the same question.

Can we expect an update on this ranking, post WC?




(and Nick; If Jim is keeping you captive in his basement to avoid any ranking updates, just give the secret signal so we can get you out of there!)

Yes- apologies. There was a delay in uploading the NAF results to the database, and I have to lag a few days behind that. Should be with us very soon.

Done! Congrats by the way

Sorry to poke my nose in, but for those glicko enthusiasts among us, mubo and latterly Ketil have fixed it so that [see kyrre post below].

Well done, chaps!

I really don't understand how Glicko ratings work compared to the ELO ratings. An example close to my own heart is in the High Elf rankings where Dirold and I are the top two in both rankings at present, in ELO I am top by the slimmest of margins, in Glicko I am second by over 100 points, we have similar phi ratings, I would really like to understand why this is different?

I thought that if the phi was similar and the ELO similar it would produce a similar glicko ranking, but this doesn't seem to be the case, is this a basic misunderstanding on my part?

Hi Jim,

There are three explanations as to how glicko works (and hence why you are where you are) that we could offer:

First - Nick's answer, which will doubtless follow. This should be considered correct and render this post moot.

Second - the overview page on thenaf.net.

Third - my pigeon / simplistic understanding. In short, glicko seems to care more about who and when your numbers have come from than ELO does.

Under ELO, a 200 coach is a 200 coach. An active, good player doing well over years with a couple of recent whiffy games. A six game, one hit wonder at a NAFC. Maybe someone that hasn't played since 1985. Who knows how they got 200. You will get the same numerical boost for beating all three, as the formula doesn't care. Glicko is a touch more discerning, in that phi (confidence in how well you're doing lately) and mu (how well you're doing) of you and your opponents all matters when it comes to rolling up your final figure. If you're playing the one hit wonder or the old hand, glicko won't be too sure about him, and his poor (high) phi will reduce his overall number and your respective gain relative to beating the good player, who will have a lower phi (and likely higher mu, I suppose). The formula also makes it harder to amass a big number over time by beating a series of (in ELO money) 150-170 players over a long period. Conversely, it is fast to reward you if you go on a run against the big cheeses; reacting relatively quickly to hone in on what it considers your real ability. If you are an ELO 250 player, glicko will get you to the equivalent number far more quickly than ELO will.

So - without checking the games in question, I would gamble that the difference you're seeing is one / more of:

- Your ELO performance being boosted by major championship multipliers (not present in glicko. It cares about opponents, not tournament gravitas)

- Your glicko suffering decay (appreciate that's unlikely if you're interested, but that's present in glicko but not ELO)

- Your opponents not being considered as good as dirold's according to the formula (not a criticism, people are just judged differently as noted above. And of course, you have little choice in who you play unless you're deliberately travelling larger distances to meet better opponents. A particular issue in less densely populated areas, where glicko is a bit meaner to coaches than it would ideally be. Watch this space for 'something else' that may help there)

Er - that's a probably nonsense ramble. Perhaps should have left it to Nick.

The cut off for the calculation is still the last day of the last month. So only games played before January 1st 2020 when this reply was written. In that way, it is still a monthly thing. Late tournament results could still change the rank from one day to the next.

Aha. Understood, good clarification!

Thanks Phil, I guess my issue is that I can see the movement in the ELO rankings, while we just have the numbers for the Glicko ranking with no detail available, so can't see where the difference is created.

A further issue for me is that having played all races, I have a target to get them all to certain rankings, difficult in ELO where you are constantly being overtaken, impossible in Glicko when you add in decay.

Sure. It’s absolutely fair to say that there are pros and cons in any ranking system, especially when you add in meta goals. I like the idea that the membership has more now and as such can choose what they like / what they want to shoot for, but I appreciate that it’s possible to be frustrated because it’s impossible to play enough BB to satisfy your meta goals in every system.

Glicko ranks have been changed to include every game played. So no monthly cut-off with daily updates like elo.

Ranking column seems to have a problem, many people ranked 1,769 throughout the column of data regardless of their value in the rating column.