Auto-Drafting From Silver To Mythic For Under $5 On Magic Arena

Does the Arena Tutor auto-drafter know something Jake Browne doesn’t? After fifteen drafts took him from Silver to Mythic, he’s starting to wonder…

Ashiok, Nightmare Muse, illustrated by Raymond Swanland

I was having one of those terrible runs where absolutely nothing you draft is working out. 0-1, 1-3 back-to-back in Dominaria. A combined 2-12 over four drafts in Theros Beyond Death. The kind of day that makes you consider if you’d have a more promising future in competitive checkers. I mean, it is a rush when you get to stack two of them on top of each other.

In the midst of this existential crisis, Arena Tutor kept popping up. I will fully admit I was not sure what it actually was when I installed it. As a geriatric millennial, you try to keep up with the kids.

It turns out one of the key features is essentially an auto-drafter. The software recommends picks in real-time while adjusting for prior picks. It is, like a greased watermelon in a pool, pretty slick.

I decided to put it to the test: ten drafts where I would do nothing but what the bot told me. I had an account sitting at Silver, so the goal was to hit Diamond by the end of the day. Smooth brain, full heart, can’t lose.

Wouldn’t you know? The bot was better at drafting than I was.

The Stipulations

No matter what the bot said, I couldn’t defy it more than once per draft. In the case of a tie, I cannot spend more than five seconds thinking about the pick, lest I risk generating a wrinkle or two on the old cranium. I had to use a new player account and couldn’t spend more than the $5 intro bonus to see if this works. I would not have to use their deck generator until the final draft, but I would lend credence to the picks I typically eschewed.

Except for Traveler’s Amulet. The bot is obsessed with Traveler’s Amulet.

Draft One

Draft Log | Match Results

As early as Pick 5, we’re ready to speculate on blue, which brings me to the first thing I admire about the bot: no tunnel vision. Unless it’s Traveler’s Amulet, which we take over the more playable Starlit Mantle Pick 7.

I did my best to honor the picks by playing the two early green cards when, in retrospect, they were easy cuts. You’ll be instructed to take a couple of cards early on in the spirit of staying open, but feel free to cast that spirit out like a priest at an exorcism when it comes time to build.

Result: 5-3

Draft Two

Draft Log | Match Results

Usually, if I’ve taken a Riptide Turtle with my fourth pick, it’s because I’ve run to the restroom and didn’t beat the clock. Again, trying to take the lead from the bot, I’m playing both the Turtle and an aggressive number of Stern Dismissals, both of which exceeded expectations. Callaphe, Beloved of the Sea was consistently panned by the end of the format, yet did a lot of work in this deck. Consider her unfairly maligned.

Result: 7-0

Draft Three

Draft Log | Match Results

One area that impressed me is how well the bot picks up on signals. In Pack 1, we’re not seeing enough green to justify Setessan Champion to the AI, even if it’s a stone-cold bomb. The bot nabs an Iroas’s Blessing and is ready to pivot to Boros. I’m happy to pull the manual override and run it as a splash, but the bot knew before I caught on just how cut off we were. Mainly because I don’t have to pay attention when it’s doing all the work. Props to my unpaid intern.

Result: 7-1

Draft Four

Draft Log | Match Results

These just weren’t our packs. By Pick 3, we’re pretty much out of options in the colors the bot identified early, so it starts on fire. Feel free to give it a reset early by speculating on a playable you like if it wants to take an on-color common too highly. Doubly so if you suspect you’re in the wrong seat, as it rarely bails on your first pick.

Record: 5-3

Draft Five

Draft Log | Match Results

When you see a bomb in Pack 2, the bot will respond to that like a “u up?” text from its hot ex. Sadly, it has no capacity to understand how you splash, leading to me overruling it on an Omen of the Hunt where it wanted yet another Turtle. There’s also the issue of building defensively to be set up when you finally get your bomb online. Feel free to freelance more with your picks anytime you have a card like Ashiok, Nightmare Muse that needs to see the battlefield.

Result: 1-3

Draft Six

Draft Log | Match Results

I could not understand why it kept pulling into black, considering the Staggering Insights we were sitting on for the life of me. The same goes for double Heliod’s Intervention. Overall, the picks worked. This was a good draft to firmly shake me out of my rut and reimagine how I evaluated this set, something we could all use now and then.

Result: 7-1

Draft Seven

Draft Log | Match Results

Not too hard to navigate this one, artificial intelligence or not. After going 1-3 with Ashiok on the splash earlier, this felt like redemption.

Record: 7-1

Draft Eight

Draft Log | Match Results

I’ll take the blame on this one. Splashing for Staggering Insight was 100% my human idea, and we paid dearly for it. Bots aren’t greedy; they take what is coming and make do, like a Dust Bowl-era meemaw. There’s also a Treacherous Blessing that it implored me to play that I feel could have been offset with enough lifegain, but this was a mess.

Record: 2-3

Draft Nine

Draft Log | Match Results

The bot values, over everything, flexibility. This can get you into a bind when it wants to take Inspire Awe, for example, over other potential starters. Cards that are on-color, colorless, or lands will be slightly higher-rated than they should be. Like a novel by Sapphire, don’t be afraid to give it a push in the right direction.

Record: 7-1 [Overall record: 48-16 (75%), Diamond 4]

Draft Ten

Draft Log | Match Results

Here, for my final draft, I go full “Jeebus, take the wheel” and turn over the draft and deck to Draft Sim.

Please, do not make this mistake.

For all the praise I have for the draft client, the deck builder is nightmare fuel. Nine creatures and three Iroas’s Blessings (plus an Impending Doom for some reason?) is wild to consider. While I did run into Archon of Sun’s Grace and a Dream Trawler in two of these losses, I would have clearly built differently. I’m hoping future versions of Arena Tutor can take into account the need for creatures.

Record: 1-3

While hitting Diamond 4 was cute, the egregious tenth draft left me scorned. I’ll spare you more draft logs (the equivalent of someone telling you about their dreams or fantasy football team) but it took five more drafts (three trophies) to crack Mythic. What seemed like a harebrained idea amounted to a 73.4% win rate and the top of the 17Lands leaderboard.

At the end of the day, Arena Tutor is impressive software responsible for earning a few gems while helping rank aggressively. My three biggest takeaways for you:

  1. The Bot Doesn’t Care About Creatures: Make sure you’re stepping in when needed to ensure you can actually attack for damage.
  2. The Bot Doesn’t Splash: If you have a bomb, make sure you’re prioritizing fixing and actually getting it on the battlefield.
  3. The Bot Overvalues Artifacts/Lands: Don’t always ignore the suggestions, but recognize where you each have blind spots.

I’ve also only had enough time to test this theory in Theros Beyond Death, so it’s plausible that the algorithm knows this set particularly well. I’ll certainly be looking to try it out more over the coming weeks.

Until next time, I’ll see you asleep at the wheel of a Tesla.