The ranking system is only useful if it is predictive. Is Baby really the 6th best player in the world at the moment? Would you confidently predict that he would consistently beat MC, Mvp, and Taeja? If not, your ranking system is not really useful at all.

Grovbolle

Denmark3777 Posts

December 28 2012 16:07 GMT

#105

On December 29 2012 01:05 Salient wrote:
The ranking system is only useful if it is predictive. Is Baby really the 6th best player in the world at the moment? Would you confidently predict that he would consistently beat MC, Mvp, and Taeja? If not, your ranking system is not really useful at all.

I agree. But you can't just say that: "Before anyone has beaten at least 100 players, we won't acknowledge what he does".
Let's see if this won't be even more useful once HoTS hits and the board is wiped clean

Conti

Germany2514 Posts

December 28 2012 16:15 GMT

#106

The funny thing is that the ranking system is extremely good at predicting matches. I let the site predict the playoff matches of the recent GSL Blizzard Cup, and it correctly guessed 4 of the 5 matchups, including a 4-2 win of Life over Parting. Go ahead, you can try it out yourself at http://aligulac.com/predict/match/

Yes, there are oddities in the rating, and there always will be. No rating is perfect, especially not in a game with so many random elements as Starcraft 2. But as far as ranking systems go, this one is pretty damn impressive.

Grovbolle

Denmark3777 Posts

December 28 2012 16:16 GMT

#107

On December 29 2012 01:15 Conti wrote:

Show nested quote +

Also did 7/8 at Homestory cup RO16

TheBB

Switzerland5132 Posts

December 28 2012 19:36 GMT

#108

On December 29 2012 01:05 Salient wrote:
The ranking system is only useful if it is predictive.

I'm getting tired of this. The whole idea behind the system was to make it predictive.

Behold. This is a plot of almost 50k historical games. On the horizontal axis you find predicted winrate for the presumed stronger player, using the ratings at the time the game was played. The games were grouped in reasonably small groups, i.e. 50%-53.3% and so on. (Obviously no numbers below 50 since this is the predicted winrate for the stronger player. It also only goes up to about 75 because there are very few games past that.)

On the vertical axis is the actual winrate for each group.

As can be plainly seen cough, cough, the actual winrate is close to the predicted winrate or, in some cases, higher.

Now I'm no statistician, and honestly I don't know how to do a proper prediction test, but it looks fucking damnable good to me.

Things this rating is not:

a ranking of player's past achievements. If you want that, count medals and winnings.
a people's award. If you want that you can make a poll.

Things this rating is:

a ranking of who would probably win in a hypothetical game played right now.

I'm happy that these threads generate some controversy and discussion, but it would really be nice if I didn't have to explain the same thing 500 times.

Salient

United States876 Posts

December 28 2012 19:43 GMT

#109

So you really think Baby (Code A player) is the 6th best player in the world at the moment? Very unlikely.

TheBB

Switzerland5132 Posts

December 28 2012 19:52 GMT

#110

On December 29 2012 04:43 Salient wrote:
So you really think Baby (Code A player) is the 6th best player in the world at the moment? Very unlikely.

Do I really think that? The (presumably) sentient human being TheBB? Does he think that Baby is the 6th best player in the world?

Nowhere did I say that.

What I can say is that if I had no information about Baby other than his past results, and that if I were to pit every Starcraft 2 player in the world against a random and fairly uniform assortment of opponents, Baby might possibly (with a probability that I could estimate, but I won't because this is ridiculous) have a winrate that only five others could beat.

However, the sentient human being TheBB knows more about Baby than that, and he can possibly make a better judgement. Things I know about Baby that the rating system does not include such interesting information as his team status, nationality, probable training regime, friendships with other progamers, and so on.

Do I, the sentient human being TheBB, think that you are underestimating Baby? For sure.

It's like you didn't even read my post, let alone the FAQ on the site.

Edit: I just realized you are literally behind almost every single complaint on the last two pages. That makes me feel a lot better about things.

Yorbon

Netherlands3188 Posts

December 28 2012 19:53 GMT

#111

On December 29 2012 04:43 Salient wrote:
So you really think Baby (Code A player) is the 6th best player in the world at the moment? Very unlikely.

According to the definition above:
There are only 5 players to be expected to win against baby at this moment. Baby can win, he can lose.
Also, being the best player or not is irrelevant, that's in your head. It's about statistical chances of him winning against an opponent.

Conti

Germany2514 Posts

December 28 2012 20:07 GMT

#112

On December 29 2012 04:52 TheBB wrote:
Edit: I just realized you are literally behind almost every single complaint on the last two pages. That makes me feel a lot better about things.

Don't worry, there's always people complaining as loudly as they can without reading a single explanation or answer. They're bound to appear with threads like these, and they definitely shouldn't get you down. You're doing awesome work here, and the community is all the better for it.

dsousa

United States1363 Posts

December 28 2012 20:12 GMT

#113

Combatex should not be on the list. He is the root of all evil.

WCG should be banned for not banning Combatex.

Salient

United States876 Posts

December 28 2012 20:39 GMT

#114

The GSL ratings are superior because they are limited to games between top players. You fail to account for the quality of the games with your statistics 101 class project. More sophisticated models can actually do that (and thus become more powerful), but they become necessarily more subjective. It's a price you need to pay to make the ratings meaningful. It takes a lot of work though. Your system is apparently less subjective and easier to construct, but you get what you pay for.

ROOTdrewbie

Canada1337 Posts

December 28 2012 20:47 GMT

#115

foreigner list seems about right :D just need to add nerchio / + 2 other zergs of the week and remove the non-zerg players

LighT.

Canada4457 Posts

December 28 2012 21:05 GMT

#116

The only thing MVP has on Flash is strategy, and thasts because he's an old gen who started a good 1.5 years ahead of him.
Flash is mechanically better at both the macro and micro aspect than MVP. Dont fool yourselves.

Salient

United States876 Posts

December 28 2012 21:10 GMT

#117

Flash was the "deity" of another game. We'll see if it carries over. I hope so. Grubby was the king of War 3, but that didn't carry over at all, unfortunately.

dabom88

United States3481 Posts

December 28 2012 21:13 GMT

#118

On December 29 2012 06:10 Salient wrote:
Flash was the "deity" of another game. We'll see if it carries over. I hope so. Grubby was the king of War 3, but that didn't carry over at all, unfortunately.

Madfrog was pretty good as WC3 as well. And Moon was GOD of WC3. And we saw how that turned out.

casualman

United States1042 Posts

December 28 2012 21:14 GMT

#119

I'm tired of all these people who know nothing about statistics shitting up this thread because they have some conception of what "should" be the right ranking. It doesn't work that way, and eventually the players that "don't deserve" to be ranked so highly will drop off with decreasing performance.