To be noted as an Aggro deck, the purpose of all cards should be to make winning faster. For example, they use removal to destroy blockers, not to defend against a threat. I separated clear Midrange decks so they don't count as Aggro or Control decks.

The third big archetype is the category of Combo decks. Dredge has always been on the radar, making people pack Leyline of the Void, Tormod's Crypt or Relic of Progenitus in their sideboards. Ad Nauseam Tendrils (ANT) and Reanimator got nerfed a while ago when Mystical Tutor got banned but are still played. Although the unbanning of Time Spiral made Spring Tide, a deck abusing High Tide, more popular, ANT is still the most played Combodeck. Belcher, Dregde, Enchantress and Sneak Attack following close.

Other ANT decks can be found here.

At this point, three quarters of the metagame are covered. The remaining decks could not be put into one of the three big archetypes. They're combinations of Aggro and Control decks. I call them either AggroControl, or ControlAggro decks, depending on the strategical approach of their Control cards. If a decks main aim is to Control the game, then it is a ControlAggro deck. For example, heavy discard decks that also pose aggressive threats fall into that category. The idea of an AggroControl deck is to put threats and control how the opponent answers them. Merfolk and Threshold decks fall under this cape. Every sixth deck on magic-league is a Merfolk deck. Therefore, I subdivided the Merfolk decks into three decktypes to be able to analyze the differences of the versions: UW Merfolk, UB Merfolk and Mono Blue Merfolk. A recent Merfolk list looks like:

Other Merfolk decks can be found _here.

I hope this suffices as overview over what decks one will face in Legacy. I listed typical decks for each archetype. The clearly most played deck is Merfolk. Next to the more quantifying table, you can see the amount of Merfolk decks in the complete meta in the diagram.

2. The best deck of the newly shaped Legacy format is...

...what brings one a far step ahead in the competition. You saw how the metagame looks like. Now I'll go a step deeper to a strategically major decision layer. What deck should I take to the next Legacy tournament I'm playing?

To determine the optimal deck choice, it's not sufficient to simply take the deck that won the last major tournament. That approach will misvalue the effect of playskill and experience of the player, and also ignore the variance the deck was affected by in that specific tournament. Just because a deck won seven 'easy' matchups does not make it the mathematically best option. What one has to look for is to find the deck that performs best on average. To further enlighten you to this idea, let me give you two examples:

DeckA played 7-0 in a major tournament with 100 players. At the same time however, 20 players played the same deck, all going 0-2-Drop. It's obvious the deck is a bad choice. It won less than 1/8 of all matches it played. The 7-0 was clearly a lucky punch.

In the same tournament, 21 players played a DeckB. 10 of them went 5-2, 10 went 4-3 and one player finished 6-1. It won more than 60% of the matches it played. It's a better deck than DeckA.

These two extreme examples showed a simple way to determine the strength of a deck. In several deck tech articles, authors write about match ups using percentages to tell about the decks strong and less strong match ups. Simply extracting the average score out of a big enough data pool reveals the average match up percentage. Your deck might have 8 bad match ups and only 2 positive ones. This however is alright when everyone else plays something that loses to your deck. The average match up is a good way to compare several decks. Technically, if you follow how the average match ups change with time, you can even prove that some match ups are indeed positive or negative. But that's not my matter for now. I'm after the best strategic deck choice, not about how to win a match.

Another way is to look at the average score of a deck. This idea will bring a worse result on decks whose players drop early. A 3-0 is worth more than three 1-0s. In this analysis, even though some numbers become different, the outcome does not change however, so I will not further touch looking at the average score.

The last idea to determine a deck's strength requires the most data. I have the information of how often a deck got played. I also have the information of how many of all matches played it won. If a deck was played by, say 10% of the meta game, but won only 3% of all matches that were won, then it has a negative conversion rate. In my December 2009 Standard(T2) analysis, Jund(15% meta) was played by more people than MonoRed (13% meta), however, MonoRed won more tournaments than Jund (MonoRed: 17%, Jund: 14%). This information made Jund a bad deck in that period. Now I'll apply this quantitative analysis to matches, not tournaments, which will provide even more detailed results. Divide the percentage of how often a deck got played by the percentage of how many of all matches it won. By merging the winning percentages with the meta game, I get a well normed number. Just the amount of won matches is nullified. Instead, I have a value to compare any deck with each other. Of course, this value is only momentary (applies to the past meta game), but it's even possible to draw comparisons between different formats. For example: Survival was never as good as UG Madness was in the 2002 World Championships.

3. What does the scouter say about its power level?

I explained how I come to a conclusion about a decks strength. Now I'll apply these mechanics on the important Legacy decks.

Overall, Aggro decks won more matches than they lost. Applying conversion rates on “the average Aggro deck” provides a conversion rate of 0,99. Aggro decks won an appropriate amount of matches. The percentages of “metagame” and “metagame won matches” are equal.

The best Aggro deck with a reasonable amount of data to work with is Affinity. It won 57% of all matches it played. This means its average score is 1,1 to 0,8. Zoo had a similar result of 54% won matches (average score 1-0,9). The Goblin deck however didn't prove its appearance in the meta game. It's the second most played Aggro deck but lost more than it won. No one wants to play a deck with as few won matches as a deck with 50% match ups..but which was played by only one third of the number of the deck. If you decide you want to play an Aggro deck in Legacy, chose Affinity. Here's an example list of the mathematically best Aggro deck type:

Other Affinity decklists can be found here.

Control decks were played similarly often as Aggro decks. They won similarly often as Aggro decks. Actually, they're not better, not worse than Aggro decks..on average. Unlike in the Aggro decks analysis however, there exists Control decks with outstanding results. The best Aggro deck had a conversion rate of XY. The better Control decks have a conversion rate of over 1,5. Especially effective is the UBG JaceStill deck. It wins 60% of all matches. Of course it's still not the best deck of all decks of the reviewed time frame, yet it's definitely a very good deck choice. Short behind come CB/T UWGr and the Lands deck. Other CB/T decks didn't have the strong results of the currently most popular version. Those that use Natural Order (NO) are as bad as Goblins.

So far, I said Control and Aggro decks are both solid options. But how can there be only good decks around? There can't. At least someone has to grasp the nettle. This time, it's Combo decks. The average score is 0,8 – 1,1. Only one archetype (Dredge) won more matches than it lost. Questions? I conclude none. The message about Combodecks is easy. Banning Mystical Tutor made combo decks unplayable. Time Spiral does not make High Tide a good deck.

The only rational choice is Dredge. It wins 61% matches on average. We all know this can change more quick than any other percentages however, as it's easier to hate Dredge than, for example CB/T UWGr or even Affinity.

Other Dredge decks can be found here.

What's now left are the decks that are mixtures of Control and Aggro, and to a very small proportion, Combo. Dividing the deck types by their strategy emphasis made a lot of sense I show now. Both separations are much better than any of the so far discussed strategy types. The Aggro Control decks consist of Merfolk on one side, and Threshold decks on the other. Only Merfolk wins. Threshold just loses. I know it's popular, but going by the data, I can't see how it will stay popular in future. UB Merfolk is individually seen the second best deck of all archetypes. You have no idea what to play ? Go with the main stream and chose the winning team.

The best deck at the same time is responsible for an entire archetype: ControlAggro.

The more controlling mid-range decks were not played much (less than 5%). However, they put an extraordinary result by winning 9% of all matches played on magic-league. This makes especially StifleNaught better than all decks I ever saw in my numerous analysis. StifleNaught is better than Jund ever was. It's better than any Jace-deck of any format. With a conversion rate of over two, it's better than Survival. It's average result is 5-2 (2,5 – 1,1). On average, this deck made day 2 of any GP. Having this superior record makes it the best deck since Jan 1st banning is applied. It might not be supported much by Legacy authors/theorists of other sites, but its quality can be quantified by the magic-league tournament data. For me, there's no other choice than ControlAggro decks in the the current Legacy.

Other decklists of the same type as the best deck in the Legacy metagame can be found in this overview. These decks all belong to the most promising deck.

4. Lessons learned

I learned a lot as I put my ideas and experience into writing this article. I hope you learned something, too. Even though a deck isn't the best deck, it is often still playable. When a few individual deck types are very good, most other decks have to be bad. It's never possible to have only good decks in a meta game. This makes my way to analyze the meta game applicable to any format. Standard, Extended, Vintage, Legacy, even Sealed ? Sure. I even did that as preparation for GP Firenze, I just didn't write about it.

The conversion rate, which compares the metagame appearance with the relative amount of won matches of all matches played is probably the best indicator for a decks strength. The problem of the simple analysis of a decks win/loss ratio values a deck that went 1-0 thrice as good as a deck that went 3-0 once. My conversion rate eliminates the improper effect of early drops and advantages better decks appropriately. By splitting up rough deck types into many different deck types, the negative effect of mirror matches can be lowered. In this article, no deck type was played by more than 6%. This is sufficient to ignore the effect of mirror matches.

One will face many types of decks in the current Legacy metagame. Control, Aggro, Combo and mixtures all show up as often as the other. To have best chances at winning a tournament, it requires preparation and knowledge of what to expect. StifleNought seems to deal very well with the diverse metagame.

One thing to keep in mind is that data analysis as I do it looks at the past only. I can't tell for sure how the meta game changes. To break the lotto probability, one not only has to foresee the metagame correct. People will always try to play the deck that beats most other decks. Right now, that's StifleNought. Soon, it might be a deck that wins easily against ControlAggro decks.

My analysis will always be momentary. I will never be able to picture the future metagame. I can live with it. At least: knowing about the past is something valuable.