Convergence

[Jacob Orlove]

[15:03] <BR4M> you'll all no doubt be pleased to know that only 100 cards make up 75% of all the cards (including the SB's) of an average T1 top 8.
[15:03] <BR4M> that kinda surpised me
[15:03] <j_orlove> that's awesome
[15:04] <BR4M> it means we're a tighter format that I would have thought

Discuss.

[Bram]

Add another 100 cards to tha and we're up to 86%. The rest is random, for the most part.

Also, just for kicks, the Top Ten of 2004 (so far):

1. ForceofWill
2. Brainstorm
3. Wasteland
4. PollutedDelta
5. UndergroundSea
6. VolcanicIsland
7. RedElementalBlast
8. ManaDrain
9. Tormod'sCrypt
10. ChaliceoftheVoid

Faerie Conclave closes the gates at the #100 spot (JUST above Exalted).

[Grand Inquisitor]

I'm actually surprised to see tormod's crypt so high.  Can you give more information about what T8's qualify for this analysis?

[Bram]

It's all Pip!

I was using Dr. Sylvan's data (the stuff he posted in his SCG article plus some stuff for white and gold he sent me) to do a regression alalysis, and found out my software package only allowed me to calculate 100 correlations (ergo 10,000 combinations) in one go. So I was forced to divide the >600 cards up into groups of 100, and I noticed this little particularity.

The way I understood it from Phil was that this is data over januari through september of 2004, listing the top 8 appearances for tourneys of over 50 attendees or something. Or maybe all of them. I'm not sure. Tormod's Crypt data (copyright Phil 2004):

12,0 - 9,8 - 6,2 - 6,9 - 5,7 - 6,4 - 10,1 - 7,8 - 7,4 (j-f-m-a-m-j-j-a-s-).

It weirds me out, too.

I was thinking more along the lines of discussionm about the relatively low number of cards that make up the format, rather than specific cards. This means that some 2,4% of our card pool descrives about 75% of our 'good' decks, and 14% describes the entire format. Extrapolating this to Standard would mean 30 cards describe 75% of their T8's and 180 cards make up all their 'good' decks. Would that be about right?

[BWM]

During the beginning of this year, Dragon was big...

Everything that wasn't running black ran Tormod's Crypts...

So the stats are probably a little higher because of that...

---

Are there actually more than 180 legal standard-cards??

I do know that the Top16 of States or something only contained like 60 different cards (just a little more than the Restricted List + 5 Basic Lands)

[Bram]

During the beginning of this year, Dragon was big...
Everything that wasn't running black ran Tormod's Crypts...
So the stats are probably a little higher because of that...

Dude, there were still 10,1 crypts per top 8 (1,3 ON AVERAGE per deck) in july...lmoast as much as in januari. And even in the last few months it's been like 1 per deck.

Are there actually more than 180 legal standard-cards??

Yes, there's some 1200, I think. 8th plus the Mirrodin block plus CoK.

I do know that the Top16 of States or something only contained like 60 different cards

Really? where did you get this info? And was the a pre-CoK tourney or not?

[BWM]

I found it somewhere here on TMD in a random forum...

And I responded to it with something like "Lol, our restricted list is bigger"

but that was just after Braingeyser and stuff got unrestricted, so they beat us...

by a little...

[Dr. Sylvan]

2003, Jan., Feb., Mar., Apr., May., Jun., Jul., Aug., Sep.
_8.0,  12.0, _9.8, _6.2, _6.9, _5.7, _6.4, 10.1, _7.8, _7.4 Tormod's Crypt

is correct. I don't know if Bram was just ignoring 2003 data because it wasn't for an even time period (three months instead of one) but yeah, that's how it works. and the numbers are all "average number of copies per Top 8 for month X".

The nutty thing is that, while January (Dragon ph34r season) was the peak of Crypts, they're still almost as high a year later, so that doesn't explain it. My comment in the article was, "This is often one of the highest-use artifacts in Type One. Unlike other popular zero-mana cards, though, this is seemingly very innocuous. Dragon is always in the back of peoples' minds, and Goblin Welder is approximately the best creature ever, so seeing Crypt at roughly the same high level for a whole year is understandable."

Which roughly translates to "I'm not quite sure why it's awesome, but people do indeed use their graveyards a bunch."

I'm off to class, but this evening I'll dig up the counts I did for the PT Kobe T64 (Block Constructed) and we'll see if they're at all in line with this. (They should be appended to Tony's treatise on Block, if anyone wants to look it up.)

[rvs]

I'm off to class, but this evening I'll dig up the counts I did for the PT Kobe T64 (Block Constructed) and we'll see if they're at all in line with this. (They should be appended to Tony's treatise on Block, if anyone wants to look it up.)

Maybe you should do the same for the Colombus PT, since there were 8 different archetypes in the T8, and it seems like a good comparison to what we (like to?) think our T1 metagame is like.

[Dr. Sylvan]

I would love to do this for Columbus, but I am too busy. Jeek is the man with the script that turns decklists into totals, and Bram is the man with the real stats software. They are the people to do that type of stuff. :)

PT Kobe's T64 is at the end of this: http://www.starcitygames.com/php/news/expandnews.php?Article=7286

[Bram]

Well, I compared Pip's data to the PT Kobe stuff on some fronts.

*DISCLAIMER*

The comparison is obviously somewhat flawed becasue:
1. Pip's data is based on many tourneys and Kobe is just one.
2. Pips data focusses on T8's while the Kobe data is a T64

I think the second one isn't reallt a problem, because PT Kobe is a skill-intensive format with good players attending (unlike the average Vintage tourney, where the good players are usually found in the top eight). Power-level wise (relative to the rest of the format) I think a low Kobe deck can still be on the same level is a T8 Vintage build. In other words: I think that doing it like this, I am in fact comparing the best each format has to offer.

Having said that, the resutls. There were some minor errors in my previous calculations, so here's the new stuff:

VINTAGE
70% of the total number of cards that appeared are made up out of the top 100 most played cards (out of 611 appearing cards in all). With 6464 cards (pre kamigawa) in our format's pool, that means that 1,55% (100/6755) of all avaliable cards in the pool make up for 70% of the slots the good decks.

STANDARD
70% of the total number of cards that appeared are made up out of the top 30 cards (out of 116 appearing cards in all). With 1857 cards in their format's pool, that means that 1,62% (30/1857) of all avaliable cards in the pool make up for 70% of the slots the good decks.

A staggering result: the percentage of about 1,6% is virtually the SAME for both formats.

[rvs]

A staggering result: the percentage of about 1,6% is virtually the SAME for both formats.

This is to be expected from a mostly 'developed' format. That's why You/Jeek should do the same for the 1.x PT in Columbus, and compare. And by should do I mean PLEASE

[Dr. Sylvan]

How were there 1,857 cards available in a Block tournament with only Mirrodin and Darksteel legal? I would think that it was more like 286+165+5 basics.

Oh, and as long as I'm being nitpicky, how did you get the total cardpool of Vintage? When I do expansion set occurrences I use the number of cardnames in a set which are not basic land (this means I counted how many alternate art versions were upping the official set size and discounted them). However, I made no accounting of reprints. The number of cardnames is much lower than the total of the sets. Did you factor that in somehow? This gets harder because the basic sets can't even be ignored---Portal cards have been directly imported to them in some cases, so they aren't completely made up of Expert-level reprints. And when WotC releases a figure on cardnames, I think they include (a) cards banned in Type One, (b) Portal cards not appearing in any T1-legal set, and (c) basic land (which for my purposes is 5 basics, 5 snow-covereds, 10 duals, and 5 fetches). These complications are why I never found the time to look through every expansion's list and cross-check. :)

[Bram]

That shit was BLOCK? I thought it was a gosh darned STANDARD tournament! Well! This changes matters.

Oh, and as long as I'm being nitpicky, how did you get the total cardpool of Vintage?

This initially gave me problems as well. Enter 'The Gatherer'. Sprt on Vintage and you'll have every Vintage-legal card right there. Multiple occurances aren't a problem; they'll list every card one and tell you in which sets it was reprinted.