maqifrnswa's picture
By: maqifrnswa, Scott H
Apr 05 2009 9:19am
4.5
Login to post comments
3150 views


First off, welcome back! This is the second installment of a series that examines which games we play on MTGO. Since there are many high quality deck analysis and format review articles, I'm taking a try monitoring the player base's choice of what games to play and report interesting data and trends. Last week we looked at the average ratings for AAC limited events, the expected placement as a function of rating, and expected return. This could be a useful tool to help people figure out which queue to aim for a given week. This week we apply a similar analysis and cover many more formats including the free CAW events and 8 player constructed queues.


This week (March 29 - April 4, 2009) MTGO had a lot going on: free 256 person constructed and sealed tournaments firing every 8 hours, the return of 8 person queues, the start of the MTGO Championship Season, and the start of "Nix Tix Weeks." This means lots of data . . . so let's get to work!

 

 

Limited

Queue Avg. Rating Error +/- Weekly Change
AAC Swiss Draft 1590 18 -31
10E Swiss Draft 1598 18  
10E Swiss Draft (nix tix) 1598 17  
10E Sealed (nix tix) 1633 38  
10E 4-3-2-2 1638 33  
AAC Swiss Sealed 1645 21  -18
AAC 4-3-2-2 1654 21  -6
MED2 Sealed Swiss (nix tix) 1717 17  
AAC 8-4 1763 16 +10

(Since there were so many queues this week, I couldn't get enough data for some of them, such as nix tix 10E 8-4, for example.)

  

Constructed

All data is for 8-player queues.

Queue Avg Rating Error +/-
ALA Block 1678 20
Extended 1683 14
Standard 1692 14

 

CAW

CAWs are a very interesting animal... besides playing with the economy and giving those who got in a chance to play for free, it also offers us an amazing chance to actually get data on the entire cross section of the MTGO player base.

 

Event Avg. Rating Error +/-
Constructed  1611  13
Sealed  1626  16

 Top 32 were about 1694 and higher for Constructed, and 1740 and higher for sealed.

 

Some analysis:
First off, we see some queues with averages below 1600 in limited while constructed tends to have higher ratings. Felorin brought up a great comment last week - can we use this type of analysis to see if higher rating players are "scaring off" lower rating players. This would be seen as an inflation in the overall ratings of everyone actively playing MTGO as new players "give up" on tournaments. It's hard to deduce that from the 8 player queues (since higher rated players are more active and would artificially inflate the ratings), but the CAWs give us a peak into the truth that there may be a small amount of inflation going on . . . we'd have to follow the trends in the future.  Looking at the CAWs, we see that the limited player base has a wider spread of ratings than constructed (standard deviations of 99 and 73, respectively).

  • "First Prize" - Having 1488 or higher rating in AAC Swiss Draft will net you a pack.
  • "Going Infinite" - Thanks to nix-tix, 1680 or higher will go infinite in a 10E Swiss Draft (nix tix). For AAC, going infinite kicks in around 1770-1780 for 4-3-2-2 and Sealed Swiss
  • "Opportunity of the Week" - check out the MED2 (nix tix) and 10E Swiss Sealed (nix tix) for players between 1750 and 1850. The MED2 payouts are high because: only 5 packs to enter and nix tix

In closing, I'd like to take a look at a quote from Hamtastic's "State" this week:

"They've also brought back the release event championship events that have really big payouts for the top players.  They've done all this while also giving players low-end competition opportunities as well.  I think that this double effort has had a tremendous impact to the tournament scene on MTGO."

While I haven't looked at the big events yet, we are beginning to see more <1600 players in the 8-player queues. And, as Hammy notes, even he is trying for QPs by getting into more competitive events. (His experiment with 8-player queues showed up as a data point in one of the constructed deck tournaments this week). This will be interesting to watch, we can see how the curves move over time as WotC try different formats ( Kaleidoscope, nix tix, Championships.)

Thanks for checking in! I hope to continue expanding coverage in the coming weeks, and soon we'll get some figures looking at week to week variation in some formats.

Work in progress: daily events, premier events, queue volume, plots of average rating versus week, synchronizing data collection with MTGO weeks (Wednesday to Tuesday)

March 29 - April 4

6 Comments

Additional info by maqifrnswa at Sat, 04/04/2009 - 10:54
maqifrnswa's picture

Hello again, after reading http://puremtgo.com/articles/qualifying-2009-magic-online-championship by "one million words," I thought it would be interesting to post the ratings where you are most likely to earn QPs for MTGO Championships (this is assuming you are equally specialized in all formats):

limited
AAC Swiss Draft 1692
10E Swiss Draft (nix tix) 1674
10E Sealed (nix tix) 1758
10E 4-3-2-2 1744
AAC Swiss Sealed 1761
AAC 4-3-2-2 1774
MED2 Sealed Swiss (nix tix) 1811
AAC 8-4 1851

8 constructed
ALA Block 1768
Extended 1748
Standard 1770

I love this article im really by Brush (not verified) at Mon, 04/06/2009 - 06:17
Brush's picture
5

I love this article im really looking forward to next week.

I have angered the WotC gods... by maqifrnswa at Mon, 04/06/2009 - 22:14
maqifrnswa's picture

I may have angered the WotC gods...

It looks like this will be my last article covering this topic:

http://forums.gleemax.com/showthread.php?t=1172837

"As part of this Wednesday’s build (April 8th) a change will be made in the MTGO ELO system. After this build, players will not be able to view any other player’s ELO rating. Players will still be able to view their own.

This change is being made as part of our ongoing review of our overall organized play system for Magic Online. None of our current tournament and championship programs use this rating in any way. In addition we have observed unhealthy behavior based on assumptions of other players and their rating. As with all of our systems, we will be observing the effect of this change and may make future changes as needed."

While this series might not have caused this change (maybe the ego of too many 1800s were hurt by losing to 1400s...), but indulge my ego that I had something to do with this!

My thoughts:
Ratings data mining can guide players towards events that they would feel would be competitively appropriate.
However,
***1) this information could be used so higher level players "optimize" their performance by attacking lower rated queues to get more QPs. That would ruin the intention of "entry level queues" such as 10E swiss draft. (see one million word's article from this week, he mentions that exact phenomenon)***
2) lower rated (i.e. <1500) may get discouraged seeing this data and stop playing.
3) this model assumes a minority of players use this data. If a large majority of players read it and act on it's data, WotC will have a very unstable player base from week to week.

All of those, I agree, are "unhealthy" and take away from the overall magic playing experience.

I'd like to ask whoever reads this from WotC if it would be possible to publish, at least through official channels, some indication for new players about where to start when it comes to tournaments (which formats, queues, etc.). However, I see there is some fun in trial and error (that is, try 4-3-2-2 and see you'd like swiss better, or the other way around). This series started during the Conflux Release events when hundreds of new tournament players were overwhelmed by choices, and some of which entered Conflux release event queues well over their competitive ability level.

I will continue to think about some of the more esoteric facets of the game, and might come up with another series in the future (I'll check with Wizards first next time to avoid any problems). Until then, I'll see you in game - gl hf

Going Infinite by Ryan (not verified) at Mon, 04/06/2009 - 22:49
Ryan's picture
4

It's a bummer that you can't continue the series.

I'm having trouble seeing how one can go infinite in Swiss Drafts, though.

This probably urged them to act sooner rather than later by IvoMV (not verified) at Tue, 04/07/2009 - 10:26
IvoMV's picture

They were probably already pondering about this stuff. Then they pondered more when they made queues "pay out" QPs. I really think your articles urged them to act sooner though, I can really see this sort of study as undesirable to WotC.

So basically I agree with you that you had something to do with it - not the cause, certainly, but VERY likely one of the enablers for the decision to have taken place now.

Honestly, my hope is that by Metalman (not verified) at Tue, 04/07/2009 - 15:27
Metalman's picture

Honestly, my hope is that this is an efficiency upgrade. Eliminate unnecessary information transfer. Like, for example, when a big tourney starts and a bunch of the people click on their opponents names to see what their ratings are. Lagg spike?

Also, isnt it obvious that better players will play in tournaments with higher overall payouts, and with payouts skewed to the top, and players who dont trust their ability to consistently win three matches will play in the tournaments that still give them product even if they lose before the finals?

The really interesting thing IMO is that the value of a QP could change the whole setup. If you value a QP the same as lets say a pack, then the 4322 actually pays 5322. This makes the total payout more in line with the 8-4 tourneys (8-4 has a tougher crowd which equals a lower chance to win it and the QP payout is the same as the potentially weaker player filled 4322). I'd certainly expect more higher rated people to be getting into the 4322's to capitalize on that.

-M