I was really pleased with the way my two draft experiments worked out, and I learned quite a bit about the format from both studying the data and trying to apply the results to actual drafts.
I wanted to try a postmortem draft where I would keep in mind some of the shifting evaluations I had as a result of the data article and the two draft experiments.
Pack 1 pick 1:
Easy first pick - I am confident that Windrider Patrol has the highest percentage, and it's an attractive first pick anyway.
My Pick:
Pack 1 pick 2:
This pack is pretty bad. There are only three cards that interest me: Kor Castigator, Herald of Kozilek, and Tide Drifter. Of those 3, the only one that interests me as a card to start a draft with is the Herald - Kor Castigator is a super replaceable 2 drop, and Tide Drifter, while useful in a lot of decks, isn't really an active strategy type of card, just a nice pick up. So, Herald it is.
Just for fun, the MTGG numbers:
56.90% Herald of Kozilek
52.70% Tide Drifter
52.70% Kor Castigator
None of the other cards break 50%.
Nice job, drafting intuition!
My Pick:
Pack 1 pick 3:
Outnumber is the best card in this pack by far.
Looking it up afterwards, the data agrees - 56.70%. The next highest is Forerunner of Slaughter at a close 56.30%, followed by a sharp drop to 53.00% with Smite the Monstrous.
So far, all my natural choices for "best card in the pack" have been consistent with the data.
My Pick:
Pack 1 pick 4:
Here's my first deviation. The technical best card is Courier Griffin (at an alarming 55.30%), but for some reason it just doesn't strike me as a white signal. Maybe it should? At any rate, I'm looking at the red cards. I like Valakut Predator and think it's generally underrated, but having both the Herald and the Windrider Patrol pushes me towards wanting to draft a U/R devoid deck, so I took Sentinel. Plus, it's a 2 drop, and 5 total p/t for 2 mana is a great deal.
Looking it up afterwards, they are very close - 53.30% for Predator, 53.10% for Sentinel.
My Pick:
Pack 1 pick 5:
Here's a similar story. Predator and Cryptic Cruiser both interest me, but since I just talked about how much I'd rather be in a devoid strategy, I went with Cruiser.
Cruiser's win rate is almost two percentage points below Predator's - 53.30% vs. 51.50%.
My Pick:
Pack 1 pick 6:
Here is a spot where I get to apply one of my biggest takeaways from this exercise. I always overlooked Spell Shrivel since it isn't a card I usually wanted in my decks - I always basically considered it unplayable, or a sideboard card at best. Seeing it with such an absurdly high win rate in the data (58.20%!!) has made me notice it a lot more, and I'm pretty much willing to start one in all of my decks now. I have been very pleased with the results.
My Pick:
Pack 1 pick 7:
Being firmly dedicated to my U/R devoid deck and its potential processor theme, I happily select Mist Intruder over Makindi Sliderunner. I guess I'm passing a pretty nice landfall deck though.
My Pick:
Pack 1 pick 8:
Sure Strike doesn't get enough love - its win rate is reasonably high (52.50%). That said, I'm still trying to mass assemble devoid cards, so I went with Vestige (whose win rate is surprisingly low - 50.70%).
My Pick:
Pack 1 pick 9:
My Pick:
Pack 1 pick 10:
My Pick:
Pack 1 pick 11:
My Pick:
Pack 1 pick 12:
My Pick:
Pack 1 pick 13:
My Pick:
Pack 1 pick 14:
My Pick:
Pack 1 pick 15:
My Pick:
Pack 2 pick 1:
Easy pick - two devoid cards in my colors, and one is far more interesting.
My Pick:
Pack 2 pick 2:
This exact pick came up in my previous draft. One thing to keep in mind, however, is that those experiments were done strictly using "likely win %" as the primary metric, in which case Incubator Drone beats out Deathless Behemoth 53.6% to 51.6%. The "likely win" statistic is derived by combining the actual win percentage with some reduction factor based on the percentage of games in which the card is able to be cast. For example, Ulamog wins something absurd like 95% of the games in which it's cast, but limited games don't usually last until someone has 10 mana, so this is the author's method of unbiasing that data.
In this case, Deathless Behemoth's actual win percentage is 58% and Incubator Drone's is 54%. I think it's reasonable to expect to get to 6 mana in this format, and especially with a cost-reducer like Herald of Kozilek, making Deathless Behemoth the best choice for me.
My Pick:
Pack 2 pick 3:
My Pick:
Pack 2 pick 4:
Super dookie pack. I still basically refuse to ever take Salvage Drone, so it's Elemental vs. Sure Strike for me. I did just talk all that talk about how Sure Strike is underrated, but I think Wave-Wing Elemental is also - I've been on the other side of that card quite a few times recently, and it has been impressing / frustrating me.
My Pick:
Pack 2 pick 5:
If I'm gonna make this card good, now is the time.
Side note, this card's win rate is absurdly high (58.8%), but there's a lot of bias - obviously it's going to be spectacular if you can cast it, but the data collection bots just plain can't see all the times where it can't be cast.
My Pick:
Pack 2 pick 6:
My Pick:
Pack 2 pick 7:
My Pick:
Pack 2 pick 8:
My Pick:
Pack 2 pick 9:
Nice, it wheeled!
My Pick:
Pack 2 pick 10:
My Pick:
Pack 2 pick 11:
My Pick:
Pack 2 pick 12:
My Pick:
Pack 2 pick 13:
My Pick:
Pack 2 pick 14:
My Pick:
Pack 2 pick 15:
My Pick:
Pack 3 pick 1:
Benthic Infiltrator is basically the perfect card for my deck, but Akoum Firebird is just so strong.
My Pick:
Pack 3 pick 2:
I can't pick this Mist Intruder fast enough.
My Pick:
Pack 3 pick 3:
Or this one! It helped that the rest of both of these packs were all dookie butts.
My Pick:
Pack 3 pick 4:
This is another card with a high-but-biased win rate - if it's not good, you just aren't casting it - but I still think it's revealing that the win rates of the counters are so high. They are much better than I thought they were originally.
My Pick:
Pack 3 pick 5:
Here's another lesson I learned from analyzing this data - Dispel's win rate is high, even after taking the bias into account, so now I try to keep more of an eye on drafting one for my sideboard in more spots like this, where originally I would have taken Vestige and probably ended up cutting it from my final deck.
In general, I've had a hard time adjusting to picks like this, and there's a good reason. When I first started to compete (in around 2004 - the release of Mirrodin and Darksteel), I was not good at limited. At all. I would never consider signals AND my card evaluations were just tragically off-base. I would frequently have drafts where I was just scrapping for playables towards the end, being so nervous that I wouldn't have 23 cards to at least make a deck without considering some absurd splash. As a result of that, when I first tried to improve my draft methods, I would overcompensate by locating an open color and constantly snatching up cards in it, basically out of fear that my chosen color would randomly dry up at any point and I'd be stuck making a deck with what I have (which is what it feels like when you can't read signals and end up in the wrong color).
Because of that, I had a hard time thinking to take a card strictly for my sideboard - I'd take a card like Vestige in this spot.
I'm glad that I could share that introspective moment with you.
My Pick:
Pack 3 pick 6:
For some reason this card's win rate is lower than I feel it should be. It's at 52.6%, but I think it should be at 97%. I loved it at the beginning of the format and I still love it now.
Gruesome Slaughter's win rate is high, but there's more bias in play here - you just plain aren't spending 6 mana on this card (or even putting it in your deck probably) unless it's going to be spectacular. I still think it is not very playable.
In any case, this pick is super easy.
My Pick:
Pack 3 pick 7:
With a billion Mist Intruders and Kozilek's Sentinels, I'm not super interested in another 2 drop, especially a non-synergistic one, so on-color special land it is.
My Pick:
Pack 3 pick 8:
I tanked really hard on this pick. Firemantle Mage is awesome, but so is Valakut Invoker. And then there's Murk Strider, who would fit really well with all those Mist Intruders I took.
I looked at my curve and decided I wanted a 3 and not a 4, so Murky was out. After that, I decided that with a ton of Mist Intruders and Kozilek's Sentinels, my aggro draws left a lot to be desired, so Invoker would be more useful.
My Pick:
Pack 3 pick 9:
Crap. If I knew this was coming, I'd have taken Murky. Oh well, I'm still happy to have two of this card, and I'm also happy that my 3 drop slot is filling up nicely.
My Pick:
Pack 3 pick 10:
My Pick:
Pack 3 pick 11:
My Pick:
Pack 3 pick 12:
My Pick:
Pack 3 pick 13:
My Pick:
Pack 3 pick 14:
My Pick:
Pack 3 pick 15:
My Pick:
This draft converter created by Benjamin Peebles-Mundy.
Visit the draft converter today!
I think this deck is very, very good.
I won the draft very easily. I played against three decks with Forest in them, which kind of upset me - it didn't feel like the difficulty level was high enough. Still, this deck fired on all cylinders all the time. I drew and cast every card in it at least once (which I think is unusual) and they all did their jobs and then some.
No comments:
Post a Comment