Last weekend, I did two Coldsnap drafts. In the first, I had five Skreds, four Elves, six Snow Lands and four Auroch Herds into double Rimehorn… And I lost in the first round.
I entered another 8-4 queue, and realized I didn’t really want to play about two seconds after the draft started. I rare-drafted like crazy, ended up with a very marginal pile, and split in the finals.
That got me to wondering — how much influence does luck play in drafting?
I’m an economist. Economics is all about finding ways to test assumptions, and to develop statistical measures of events. So let’s look at the statistics.
There’s no question that the first deck was far better. It was tighter. It had a decent curve, lots of removal and a great win condition. The second deck had some similar cards: it was U/G/r/b, with Frost Raptors, a small Auroch chain, a task mage and Garza Zol as a finisher. It also ran triple Frostweb Spiders and just two snow lands, one of which was Arctic Flats.
Watching the game summaries, I could clearly see how I lost the first match. In two games, I drew seventeen lands and one Bull Auroch. I was running seventeen lands and had four in my opening hand both games.
In the two matches I won with the other deck, I drew reasonably well and played very tightly (which is a real surprise — and a rare event). My reasonable draws meant I drew the Druids and Aurochs when needed, and never drew the chaff. My opponents weren’t seriously mana-screwed or flooded. It is possible they had bombs and could not draw them — but, of course, I couldn’t see that.
Of course, one example means nothing. Flukes happen. I have beaten powered Vintage decks with standard or block decks. Mono Green aggro has beaten Solar Flare. People have fallen eight hundred feet down a mountain and lived. It just doesn’t happen often, and certainly one single example proves nothing.
I have also looked at some of the replays of my games, and some of my notes from past drafts. A large number of the drafts mention mana flood or mana screw…. Still, that doesn’t prove anything, either. I don’t make notes about the mana in games where it played out as expected. I remember the mana problems, and I remember my mistakes, but I remember those because they are unusual. People remember the unusual.
Here’s an example: I was at Pro Tour: Atlanta last year. I walked past a certain shopping center on my way to and from the venue many times. Lots of cars pulled out of the parking lot, but the only one I remember was a dark gold Camaro. The Camaro tried, unsuccessfully, to pull out in front of a truck. The crash sent the Camaro spinning across three lanes of traffic.
Dramatic. Unusual. Completely atypical, so I remember that particular Atlanta car, but none of the others.
In short, anecdotes and memories prove nothing. Proof means numbers, statistically valid results, and defensible methodologies.
Getting hard numbers on the effect of luck in drafts is going to require better data than just notes on mana screw or random results. I think I have a method. I will take a series of draft decks that lost in the first round, and decks that won (or at least drew) in the finals. I will then play the winning decks against the losing decks, playing ten game sets, swapping players after five games, and playing six games of each set sideboarded. I’ll also try to avoid decks that relied on one card (like Umezawa’s Jitte — not that I ever opened one, or Loxodon Warhammer).
One drawback, of course, is that the decks will be removed from their draft pools. I can’t help that – but by swapping decks, I can balance out player skill. I can also avoid choosing too many decks with the same basic archetypes or color patterns.
I considered doing this test with Coldsnap drafts, but I don’t have all that many CSP draft decks recorded. I have won a fair number of Coldsnap drafts, but they are too similar — a large number have an Aurochs Herd chain, several have three or more Skreds, etc. That makes the matchups somewhat unrealistic — not to mention that a few of the wins are in 4-3-2-2 drafts, and I would like to use only 8-4s if possible.
I can’t use Ravnica/Guildpact/Dissension drafts. I somehow corrupted the spreadsheet where I store the decklists for RGD drafts — besides, I always draw in RGD 8-4s, on the (rare) occasions I make T8. And I started this project way before Time Spiral was even released. (It takes a long time to squeeze in lots of games between random draft decks, when you can have a lot more fun playing practically anything else.)
That leaves Ninth Edition. I have plenty of draft winning 9E decks logged. I also have a number of decks that lost in round one, but I think should have done better. Besides, 9E online drafting will continue for a while. That is not necessarily true of Coldsnap — it may soon be as hard to find a Coldsnap draft as to fill a Mirage draft.
Here’s a list of the decks. I’ll start with the losers…. And, yes, the first loser is a bad deck.
(EDITOR’S NOTE: I could have made all of these in the cool deck format, but I chose not to because of Peter’s notes. If y’all think that it’s really vital, I could go back and do it — I just think you lose more than you gain — T.F.)
Loser U/G
(Notes on draft: Lost in first round to opponent with Master Decoy, Icy Manipulator, etc.)
10 Forest
7 Island
1 Blanchwood Armor
1 Vulshok Morningstar
1 Treetop Bracers
1 Counsel of the Soratami
1 Mana Leak
2 Dehydration
1 Rampant Growth
1 Reclaim
1 Treasure Trove
1 Llanowar Behemoth
1 Phyrexian Hulk
1 Silklash Spider
1 Aven Windreader
1 Zodiac Monkey
1 Giant Spider
1 Llanowar Elves
1 Order of the Sacred Bell
1 Trained Armodon
1 Wind Drake
2 Rootwalla
Sideboard
1 Creeping Mold
1 Lumengrid Warden
1 Ley Druid
1 Craw Wurm (premium)
1 Reclaim
Loser R/B
(Notes: Opponent plays turn 2 Fellwar Stone, turns 3 & 4 Bog Wraiths.
Ditto game 3.)
7 Swamp
8 Mountain
2 Island
2 Dark Banishing
1 Flame Wave
1 Volcanic Hammer
1 Slate of Ancestry
1 Icy Manipulator
1 Blackmail
1 Enfeeblement
1 Razortooth Rats
1 Deathgazer
1 Bottle Gnomes
1 Aven Fisher
1 Giant Cockroach
1 Gravedigger
1 Hollow Dogs
1 Looming Shade
1 Ravenous Rats
1 Anaba Shaman
1 Flowstone Shambler
1 Goblin Brigand
1 Goblin Piker
1 Hill Giant
1 Hell’s Caretaker
Sideboard
1 Execute
1 Shatter
1 Regeneration
1 Elvish Berserker
1 Naturalize
1 Zodiac Monkey
2 Sift
1 Boiling Seas
Loser U/W
(Notes: Played seventeen lands over the course of two games.)
11 Island
6 Plains
1 Mana Leak
2 Remove Soul
1 Rewind
1 Counsel of the Soratami
1 Sleight of Hand
1 Time Ebb
1 Aladdin’s Ring
1 Dream Prowler
1 Thieving Magpie
1 Aven Fisher
1 Aven Windreader
1 Horned Turtle
1 Lumengrid Warden
1 Sea Monster
1 Wanderguard Sentry
3 Wind Drake
1 Angel of Mercy
1 Infantry Veteran
1 Master Decoy
1 Warrior’s Honor
Sideboard (yes — all green)
1 Viridian Shaman
1 Giant Spider
1 Rootwalla
1 Zodiac Monkey
1 Scaled Wurm
1 Volcanic Hammer
1 Guerrilla Tactics
1 Icy Manipulator
1 Rod of Ruin
1 Soul Feast
1 Coercion
1 Dark Banishing
1 Fellwar Stone
1 Unholy Strength
1 Bog Wraith
1 Drudge Skeletons
1 Nekrataal
1 Orcish Artillery
1 Ravenous Rats
1 Anaba Shaman
1 Scathe Zombies
2 Flowstone Shambler
1 Giant Cockroach
2 Lightning Elemental
1 Goblin Piker
1 Hill Giant
Sideboard
2 Mind Rot
1 Demolish
1 Shatter
1 Disrupting Scepter
Winner U/W/b
7 Island
8 Plains
2 Swamp
1 Dark Banishing
1 Pacifism
1 Remove Soul
1 Rod of Ruin
2 Mana Leak
2 Time Ebb
1 Sleight of Hand
1 Aven Cloudchaser
1 Aven Flock
1 Azure Drake
1 Counsel of the Soratami
1 Crossbow Infantry
1 Foot Soldiers
2 Infantry Veteran
1 Leonin Skyhunter
1 Master Decoy
1 Pegasus Charger
1 Sanctum Guardian
1 Skyhunter Prowler
1 Wind Drake
Sideboard
1 Annex
1 Coercion
1 Deathgazer
1 Evacuation
1 Tempest of Light
2 Warrior’s Honor
1 Peace of Mind
1 Blanchwood Armor
1 Chastise
1 Rod of Ruin
1 Giant Growth
1 Pacifism
1 Rampant Growth
1 Reclaim
1 Treetop Bracers
1 Blinding Angel
1 Craw Wurm
1 Elvish Berserker
1 Grizzly Bears
1 Kami of Old Stone
1 Master Decoy
1 Order of the Sacred Bell
1 Pegasus Charger
1 River Bear
1 Rootwalla
1 Trained Armodon
1 Zodiac Monkey
3 Wood Elves
Sideboard
Blessed Orator
Biorhythm
Circle of Protection: Red
Cruel Edict
Enrage
Elvish Warrior
Shock
Spirit Link
Angelic Blessing
Let’s look at the matchups, make some predictions, then talk about how the matches played out.
Loser U/G vs. Winner R/B:
Loser U/G has a few fat guys, and two fliers (three counting Treetop Bracers). It also has Blanchwood Armor and Vulshok Morningstar, so even its little guys can get big. Its removal is limited to a couple Dehydrations and the Mana Leak, plus the sideboard Creeping Mold. U/G’s best game plan seems to be to get a fattie or flier boosted with Armor and/or Morningstar, then beat. Actually, just beating across the board should be good — red burn has trouble with fat butts, like the one on Silklash Spider.
Winner R/B has a lot of removal, plus the Icy Manipulator. It can probably kill anything U/G plays, or drop a regenerator in front of it. Over time, it should be able to get some damage through here and there, then win with Anaba Shaman and burn, if necessary. It could be in trouble if U/G gets enough fatties in play that removal plus regenerators plus Icy cannot deal with them all.
In practice, that’s what happened. Sideboarding consisted of bringing in artifact destruction on both sides, replacing Coercion with Demolish and Zodiac Monkey with Creeping Mold. The matchup proved dead even, with neither side having an obvious advantage.
Loser U/G vs. Winner U/W/b
Loser U/G has two Spiders — including Silklash Spider. Silklash is pretty damn good against a deck with many fliers. Winner U/W/b has one Dark Banishing and a couple of Counterspells to stop Silklash… Although it also wants to apply early pressure, since all of its groundpounders are smaller that the green guys. Going very long, U/G has better card drawing in the slow and inefficient (but reusable!) Treasure Trove.
In practice, the match came down to tempo. If either deck had mana issues, it lost. Likewise, if either deck lost tempo, it never caught up. Time Ebb was powerful, as was the Morningstar. Silklash was huge — neither Icy nor Pacifism can get fliers past Silky. The result was slightly in Loser U/G’s favor: 6-4.
Loser U/G vs. Winner G/W:
At first glance, the U/G deck looks to have all the answers. Blinding Angel seems trumped by Silklash Spider. Both sides have Blanchwood Armor, but U/G also has Mana Leak, Dehydrations and the Morningstar. Looking more closely, however, G/W has a better mana curve, more landwalkers, and better tempo.
In practice, G/W kicked the U/G deck’s butt game after game. The Wood Elves were huge: G/W always had plenty of mana, and it could usually get the Blinding Angel to stick — at least the second time. Silklash was often relevant, but by the time it came down, UG was usually so far behind that it could not catch up. 7-3 in favor of G/W.
Loser R/B/u vs. Winner R/B
Both decks have burn, Dark Banishing and Icy Manipulator. R/B/u also has Flame Wave, which is deadly if — and it is a big if — it can get the RRRR to cast it. Loser R/B/u also has Hell’s Caretaker (yes, which dies to Winner R/B’s Rod of Ruin, Anaba Shaman, etc.) which can generate lots of card advantage when it starts swapping Ravenous Rats, Gravedigger, and Aven Fishers around. Winner R/B’s trump card is probably Bog Wraith.
Loser R/B/u also has Slate of Ancestry.
In practice, Slate was the biggest factor in any long game. If the deck stalled, then Slate would start drawing a lot of cards and eventually find Flame Wave for the win. The games were all battles, and many came down to a deck winning the turn before it would have lost.
Loser R/B/u vs. Winner U/W/b
Winner U/W/b has lots of 2/1 fliers, and a few bigger creatures. Loser R/B/u has Anaba Shaman, Icy Manipulator, and lots of removal and Slate of Ancestry. Ahead of time, I liked the removal suite and Icy over U/W/b’s evasion plus some tempo cards.
In practice, the matchup was even more one-sided. U/W/b won one game when R/B hit its first nine land drops, and never saw anything of note. U/W/b won one other game — and that was really close. The rest were all blow-outs, with R/B/u so far ahead that the matches went really quickly.
Loser R/B/u vs. Winner G/W
This matchup looked really good for R/B/u, which has tons of removal and decent creatures. G/W has a few tricks — starting with Blinding Angel. However, G/W has a lot of mana acceleration, and a good curve — meaning it has tempo. Slate is fine, but not if the opponent keeps throwing Craw Wurms at you. On the flip side, G/W’s only answer to the Hell’s Caretaker engine is Rod of Ruin.
Reclaim proved very good — generally getting back either the Angel, Chastise, or Giant Growth. The tempo was also huge. R/B/u generally had to power into Flame Wave to win. Post-sideboarding, Execute should have been great, taking out the Aven Fisher and Islands for Execute and more Mountains…. But it never got drawn. The results ended up dead even, but had R/B/u actually drawn the Executes, it might have won several more — the games were that tight.
Loser U/W vs. Winner R/B
A horde of fliers backed by counters versus a handful of removal and an Icy. This looked one-sided, and it was. Winner R/B could occasionally pull out a match on the back of Orcish Artillery and the like, but those wins were rare.
Loser U/W vs. Winner U/W/b
This matchup looked more even, but the number of and power of Loser’s fliers was stronger. This is very close to a mirror match, but Winner U/W/b tended to rely more on groundpounders and tempo, and Loser U/W’s Horned Turtles and Lumengrid Warden trumped that. Sea Monster was nice, here.
Loser U/W stomped all over Winner U/W/b.
Loser U/W vs. Winner G/W
And, finally, Loser UW matches up against the only green deck I could find that I won a draft with — and the deck I figured was the worst of the bunch. G/W has Blinding Angel and tempo — and Craw Wurms. U/W has a ton of fliers and counters.
In the event, the games were all long, and close. Aladdin’s Ring actually became relevant in a number of games, both as a means of killing Blinding Angel and as a win condition. Several games were very swingy: G/W sent Blinding Angel into multiple Wind Drakes (but had the Giant Growth to help), or G/W ripped and dropped the Blanchwood Armor on the Blinding Angel or River Bear the turn before Aladdin’s Ring could kill it. G/W squeaked out a 6/4 record here — but not by much.
So here are the results.
| Loser UG | Loser RBu | Loser UW | |
|---|---|---|---|
| Winner RB | 5/5 | 4/6 | 3/7 |
| Winner UWb | 5/5 | 2/8 | 3/7 |
| Winner GW | 7/3 | 5/5 | 6/4 |
After a ton of games, which took a couple of months to play out (yes, I had lots of better things to do), the decks which lost in the first round ended up with a 50- 40 record against decks that won their drafts. Without the G/W deck’s strong record, the results would be even more lopsided against the winning decks.
My goal in this article was to look at the impact of luck in drafting. If drafting is primarily skill based, then well-drafted decks should beat poorly drafted decks. If results of drafts are primarily driven by luck — in terms of match-ups, opponents, mana screw, etc. — then the decks which were highly successful in some matches may not be successful in others.
I took three decks which won their drafts, and three decks that were defeated in the first round. I then played them against one another. With ninety games played, 60 percent of them post-sideboard, the loser decks actually beat the winner decks the majority of the time.
In short, there was no correlation between the raw power of decks which won drafts, and those eliminated in the first round.
That makes you think, doesn’t it?
Obviously, skill plays a big role in drafting — and more skilled players generally do beat unskilled players. However, you can eliminate the skill factor by switching players, alternating which deck goes first every round, and so forth. This test made sure player skill was equalized, as was going first, etc. Games with massive bad luck, like drawing ten straight lands, were replayed. In short, I did everything I could think of to eliminate outside factors.
The result: the decks which were beaten in round one of the draft proved better than those that won their drafts.
People may argue that none of these decks are bad. Obviously, it’s possible to draft a terrible deck. However, I don’t think any of these qualify. However, some of these decks lost round one, and some did not lose a game all draft.
Why?
It seems to me that these results prove that a serious element of luck exists in drafts. It means good decks lose matches. This, to me, raises questions about single elimination drafts. At Friday Night Magic, and similar events, people play all rounds of the draft — it is 3 rounds of Swiss, instead of single elimination. At the local stores, those drafts seem more popular than single elimination. I, certainly, prefer them.
I would like to see a similar option appear online. You could run drafts in which everyone plays all three rounds, and the prizes are one pack per match won. The total payout would be twelve packs — the same as an 8-4. The difference would be that more people could win at lest one pack — and that no one would win a lot of packs. That might drive away some of the people playing for profit — but those people could still play in the 8-4s. It would make people like me play more — because I hate paying 15 TIX for two games of magic.
I would love to discuss this further, in the forums, but I will be heading to Paris a week early, and I’m not going to spend much time in Paris in front of a computer.
PRJ
pete {dot} jahn {at} Verizon {dot} net