On October 31 2013 02:32 slowbacontron wrote: That's good to hear! Except that, um, I don't think it's updated as far as I can tell? + Show Spoiler +
Also, if you click on the results tab, for some reason a ton of results are unrated even though they were entered many many hours ago. Is there something up with the updating?
Yes, we know. It's an unrelated bug that causes a division by zero while trying to maximize a function. TheBB is on it.
In the "Highest ratings ever achieved by X" sections, the ratings shown seem to be mirror matchup ratings, not overall ratings. The actual ranking seems to be right, though.
In the "Highest ratings ever achieved by X" sections, the ratings shown seem to be mirror matchup ratings, not overall ratings. The actual ranking seems to be right, though.
On November 03 2013 23:30 slowbacontron wrote: Ok, I'll admit it. By "a bit", I meant "a lot. a loooooooooot". Maybe someday I'll understand it. somedayyyyyyy~~~
The charts are plotted using data from not only the past, but also the future. The ratings in the lists use only past data (so that they never change). It makes the charts smoother, prettier and nicer to look at. Sparkles and ribbons and such.
On November 03 2013 23:30 slowbacontron wrote: Ok, I'll admit it. By "a bit", I meant "a lot. a loooooooooot". Maybe someday I'll understand it. somedayyyyyyy~~~
The charts are plotted using data from not only the past, but also the future. The ratings in the lists use only past data (so that they never change). It makes the charts smoother, prettier and nicer to look at. Sparkles and ribbons and such.
Cool, that makes sense! It seems to me that it makes the ratings inconsistent with the graphs, though.
I was wondering something : if we take the MC number as a probability distribution, then what does the probability distribution look like?
I am pretty sure that 90% of the players have a MC number of 2 or less, some poor guys stand at 3, and you have at most a couple of progamers at 4. (My guess would be SEA players that have never gone abroad and have never faced moonglade)
Or do you have an incredibly isolated community (African/South American competitors?)
I assume that you guys can use this for multiple purposes, especially in write-ups and during LR-threads
Looks nice When you guys have time left you could also add the ability to filter on an ongoing tournament (e.g., WCS), so they can use this for example when writing the preview of the final and show the performance of both players in that tournament alone. Road to the final basically.
I assume that you guys can use this for multiple purposes, especially in write-ups and during LR-threads
Looks nice When you guys have time left you could also add the ability to filter on an ongoing tournament (e.g., WCS), so they can use this for example when writing the preview of the final and show the performance of both players in that tournament alone. Road to the final basically.
On November 07 2013 16:29 dani` wrote: Looks nice When you guys have time left you could also add the ability to filter on an ongoing tournament (e.g., WCS), so they can use this for example when writing the preview of the final and show the performance of both players in that tournament alone. Road to the final basically.
This would be nice, I agree.
Note that I also fixed two bugs:
The problem above reported by slowbacontron. (Records wrongly showing mirror match-up ratings)
The ability to access the short urls generated by the button at the bottom of the page.
I assume that you guys can use this for multiple purposes, especially in write-ups and during LR-threads
Looks nice When you guys have time left you could also add the ability to filter on an ongoing tournament (e.g., WCS), so they can use this for example when writing the preview of the final and show the performance of both players in that tournament alone. Road to the final basically.
This is an interesting web site. I will certainly follow it avidely as a good prediction system is worth money.
Having been invovled in offshore wagering outfits for more than 10 years, I know that prediction is always a tricky and difficult thing to do. Those who find a good system end up making a lot of money (millions). They are rare.
The current market prices for the matchups of BlizzCon don't all agree with the predictions of aligulac.
Dear/Taeja, aligulac predicts that Dear will win with a probability of 46.98% (Dear an underdog!!). Bookmakers have set the price of Dear winning to be around 65%.
Bomber/MMA, aligulac predicts that Bomber will win with a probabiliy of 43.73% Bookmakers have set the price of Bomber winning to be around 59%.
Innovation/DuckDoek aligulac and bookmakers are about the same.
Jaedong/MVP, aligulac predicts 65.75% Bookmakers Jaedong wins at 65%
I do wager on these, started recently, and so far so good (against the bookmaker). It's a short run so far, so I could be luckyier than I should have been. On the other hand, I'm not sure I would have won using this prediction system.
Keep in mind that some characteristics of the SC2 scene make it extremely difficult to portray some players' skills both accurately and objectively. If a player almost always plays among a very limited set of players, for example GSL or Proleague, there isn't a very good way to standardize his skill against players outside that pool. Additionally, some players' skill levels are just so close that it's impossible to really predict a match between them.
EPT![[image loading]](http://i.imgur.com/IDDSoPp.png)
