That first deck you shared, Azorius Oculus, is so good. With Rakdos Aggro that is supposedly only a tiny bit worse win rate-wise, I was stuck in Plat 1 for 2 weeks. But with the Oculus deck I was able to climb to Mythic #900 in just 2 days and that is while missing 6 rare dual lands, having replaced them with commons. Thanks for making this video!
Great video, but @14:08 Kutzil's Flanker doesn't have any mode that can flashback spells from the graveyard. Scry 2/gain 2 life, exile opponent graveyard, or the +1/+1 counter mode are the 3 options.
Thanks for the great content, Joe. I have one question. Why do you never focus on the agregate winrates for archetypes as you do for the popularity ? I think this is very important data to configure a metagame tier list.
On the main page it’s just win rate of the day not the week+. Without knowing the underlying game count to those win rates it could just be a small amount of games resulting in broader noise (10-20 games where 1-2 wins can be 5-10% swings for example). The popularity will be the deck archetype relative to the total population of games played as opposed to just total population of the specific decks games played for win rate. Aggregate for the week is what rolls into the individual deck profiles that aren’t community members specific decks. I like looking at population for meta share of played as it helps when I’m making comments on card choices. For example, when red was 30+% you need to pack cheap removal. If it was say heavily control then maybe you want duress/hand hate etc.
The Arabella archetype feels very underrated. I'm using a poor man's version with Virtue of Loyalty at the top end and a lot more removal than the standard lists and it's been doing well.
How statisticslly significant is the winrate of a deck with 120 games? When there are many deck versions with minor card variations, won’t some of them have higher WR and some lower just by pure chance?
I use the % more to aggregate in terms of bands /tiers of decks. Say 65-68% is likely just a handful of games going either way but when you have 10+% deviations usually starts indicating some strength of deck differences. There is also player skill which is hard to factor out. With same deck archetype different cards I like to look before recording to see if different configurations yield more significant variances in win rates. Say land differences, creature or removal spell difference could vary the win rate
I think once you get over a sample size of 15 to 20 matches, you could start using probability rules and hypotheses, but undergraduate school said 30+ sample size is golden. But I think you'd need an intake survey to be sure the Arena players were sober and actually playing competitively.😂. Otherwise the data is still pretty skewed if the players were just playing casually to finish side quests, etc..
@@matthewludivico1714 at 30 games, the margin of error at 95% confidence is +/-18% or at 80% confidence it is +/-10%. That’s fine for some types of basic survey, but for win rates or election prediction where percentages matter, you’ll want at least 500. And if you have a lot of variations of the same deck, it gets a lot worse as it becomes very likely there will be variations that are statistical outliers. It’s like trying to figure out the average height in a village by looking for the tallest man.
It's not pure chance on Magic the Gathering Arena because it's like these guys become Yu-Gi-Oh and have the heart of the cards and draw every card they need. This something weird going on with the matchmaking and with the card drawing in this game
Joe: I think I forgot something... Me: If you forgot then it probably wasn't important Joe: Yeah, you're right... *Timeless Bo1 Meta Video standing in the rain*
Only the leyline decks can output turn 2 lethal where the other aggro decks are usually turn 4. Lockdown, binding and sunfall can deal with them. Domain also goes over the top of all the midrange/non aggro decks so while they likely lose more to leyline they win against the other decks
That first deck you shared, Azorius Oculus, is so good. With Rakdos Aggro that is supposedly only a tiny bit worse win rate-wise, I was stuck in Plat 1 for 2 weeks. But with the Oculus deck I was able to climb to Mythic #900 in just 2 days and that is while missing 6 rare dual lands, having replaced them with commons. Thanks for making this video!
Nice, I've climbed the ladder pretty fast with this demon deck. Thanks for showing it in your video.
nice run with the deck, looks like a tight list
Great video, but @14:08 Kutzil's Flanker doesn't have any mode that can flashback spells from the graveyard. Scry 2/gain 2 life, exile opponent graveyard, or the +1/+1 counter mode are the 3 options.
Thanks meant exile graveyard but brain might have gone elsewhere haha
Thanks for the great content, Joe. I have one question. Why do you never focus on the agregate winrates for archetypes as you do for the popularity ? I think this is very important data to configure a metagame tier list.
On the main page it’s just win rate of the day not the week+. Without knowing the underlying game count to those win rates it could just be a small amount of games resulting in broader noise (10-20 games where 1-2 wins can be 5-10% swings for example). The popularity will be the deck archetype relative to the total population of games played as opposed to just total population of the specific decks games played for win rate. Aggregate for the week is what rolls into the individual deck profiles that aren’t community members specific decks.
I like looking at population for meta share of played as it helps when I’m making comments on card choices. For example, when red was 30+% you need to pack cheap removal. If it was say heavily control then maybe you want duress/hand hate etc.
Ok Joe, but what I miss in your metagame analysis are the winrates for deck archetypes (and not for specific versions of decks).
Yeah! 🎉The best video in the week 😊
The Arabella archetype feels very underrated. I'm using a poor man's version with Virtue of Loyalty at the top end and a lot more removal than the standard lists and it's been doing well.
I thought she’d slot in well into convoke but the lists are pretty stock and seem to have fallen off with all the lockdowns
How statisticslly significant is the winrate of a deck with 120 games? When there are many deck versions with minor card variations, won’t some of them have higher WR and some lower just by pure chance?
I use the % more to aggregate in terms of bands /tiers of decks. Say 65-68% is likely just a handful of games going either way but when you have 10+% deviations usually starts indicating some strength of deck differences. There is also player skill which is hard to factor out. With same deck archetype different cards I like to look before recording to see if different configurations yield more significant variances in win rates. Say land differences, creature or removal spell difference could vary the win rate
I think once you get over a sample size of 15 to 20 matches, you could start using probability rules and hypotheses, but undergraduate school said 30+ sample size is golden. But I think you'd need an intake survey to be sure the Arena players were sober and actually playing competitively.😂. Otherwise the data is still pretty skewed if the players were just playing casually to finish side quests, etc..
@@matthewludivico1714 at 30 games, the margin of error at 95% confidence is +/-18% or at 80% confidence it is +/-10%. That’s fine for some types of basic survey, but for win rates or election prediction where percentages matter, you’ll want at least 500. And if you have a lot of variations of the same deck, it gets a lot worse as it becomes very likely there will be variations that are statistical outliers. It’s like trying to figure out the average height in a village by looking for the tallest man.
It's not pure chance on Magic the Gathering Arena because it's like these guys become Yu-Gi-Oh and have the heart of the cards and draw every card they need. This something weird going on with the matchmaking and with the card drawing in this game
Joe: I think I forgot something...
Me: If you forgot then it probably wasn't important
Joe: Yeah, you're right...
*Timeless Bo1 Meta Video standing in the rain*
I skipped all the non-standard Bo1 last week for the most part, data wasn’t there most. Aiming for Tuesday on timeless stuff
@@MTGJoe Hahaha all good mate, just teasing! Hope you are feeling better!
I'm surprised no one is playing that invasion/bramble familiar combo deck with the overlords. It seems like it could be good.
could just be a volume of games thing that it is not showing up in aggregate stats. I try every week to over every unique/non-meta decks
@@MTGJoe for sure. The meta seems pretty open still. I just like that deck lol
How does Domain survive in a weeny aggro environment?? T2 Gain 3 life lose 15 life
Only the leyline decks can output turn 2 lethal where the other aggro decks are usually turn 4. Lockdown, binding and sunfall can deal with them. Domain also goes over the top of all the midrange/non aggro decks so while they likely lose more to leyline they win against the other decks
I wonder if there is any 5 color Domain Discard control😂
It plays leyline to pump your lizards lol
First