The Eh Team #281 – 10 is Enough

[saf feature=”itunes” itunes=”″ cta=”Subscribe in iTunes” align=”none”][saf stitcher=”” rss=”” facebook=”” align=”none”]

Jay Boosh wonders why is it hard to switch playing formats in Magic? KYT picked up GW Tokens for a Standard PPTQ, but was far more impressed with the Esper Tokens brew that won the event. Even Eric Froehlich is playing the RG Land Destruction deck in Modern these days! Matt defends his testing conclusions to Jay. Jeremey is gearing up for his online RPTQ this weekend. Matt is excited for many of the cards in Eternal Masters that have been “downgraded” to Common to make their way into his Pauper Cube.

**WARNING – this podcast may contain rude, crude, & lewd content. Listener discretion is advised.**

You can find us and our guests on Twitter at:
@mtgEhTeam – The official Eh Team Twitter account

@kytmagic – Kar Yung Tom
@JayBoosh – Jay Tuharsky
@J_Schofield – Jeremey Schofield
@mattmendoza – Matt Mendoza


“Eh Team Intro” – @mtgcolorpie
“Throwing Fire” – Ronald Jenkees

Notify of
Newest Most Voted
Inline Feedbacks
View all comments

Jay, you’re freaking out over something very small. RE: win %, calm down brah


i wasnt freaking out 🙂


I’ve listened to all 281 of these things brah, you were definitely freaking out, lol <3


you got me

Tyler Keahey

Jesus. Jay really turned an interesting topic into a shit storm. Good job dude.

Bruce Lee

Percent as in “by the hundred”.
Why list something in percent when the “detailed value” is really just the result of a relatively simple fraction?
Just use fractions. It’s easier.
10 games is a good start for testing. In fighting games, character match ups are typically rated as being 5/5, 6/4 or 7/3. But in those games player actions are the only element of variance. So card games need more reps.
Many people listen to this show, and for various reasons. What some enjoy thoroughly others might find bland.
And that makes me very happy.
Quality dialog instigation from all. Good stories. Great episode.

Jonathan Peter Gower

jay is right, 10 is not an acceptable sample size and I don’t feel the conversation was going to reveal the size of the sample without his enquiery – fractions are much better.
Also I strongly disagree with 4 smaller packs for cube draft. Working out what colours/ strategys are open and wheeling cards are fudermental to draft – aren’t you taking away from the experience. As a cube owner , if some neckbeard says “how can this still be in here” I feel I have done my job and not put shit cards in.

Tim Sargent

I actually agree with Jay on this one. Magic players tend to be very hyperbolic. Hearing someone say that something is a good match-up or I’m 80% against the deck (which is even more that just a good match-up) can be easily dismissed without quantifiable information. Now, I know that Matt did give his numbers after the fact, but without those numbers, the statement can be misinterpreted. What the “number” is is hard to come up with.

Matt had a good point saying that play experience is very important. Some wins can come from just not knowing what deck you are playing against and what cards are important. You can side out the best cards in the match-up for useless cards if you don’t know the decks. Now that RG Ponza is on the map, there may be less of that kind of edge.

Also, in a Modern tournament, you’ll probably play against the same deck at most twice. Listing out all of the specific decks in Modern may water down your results. Using archetypes like zoo, combo, and such may create more significant results. Mind you, this would make the play experience versus those decks even more important. In Standard, playing against specific decks is much more important.

tldr – Quantitative analysis, play experience, archetypes vs decks

I think good points got lost in the arguing but it’s a great topic to cover

Simon Peters

I love the way you guys handle interesting topics via awkward barneys. Good stuff in a world of bland content. Talking of which…

I’m no expert but there are statistical approaches to establishing conclusions from limited samples – e.g. null-hypothesis testing. Other posters raise good points about the importance of non-quantitative aspects with MTG, but stats can still be useful. I won’t go into detail – just google if interested. Simplified…

Pack of green & orange tic-tacs. The null-hypothesis is that green and orange are 50/50 in the pack. You pour 9 into your hand, 8 are green. What can you conclude? A ‘p-test’ (which takes account of sample size) suggests that it is very unlikely, though not impossible, that the null-hypothesis is true. The test reaches its limit at a little over 60%. So based on that sample you can say with some confidence that it is unlikely that the proportion of green over the entire pack is less than 60%, but the sample is too small to say anything more.

So it turns out both Jay and Matt were right to some extent: you can’t conclude anything close to 88% match-up based on 8-1 matches. Doubling the sample and record to 16-2 (so 20ish) IS better: it gets you to about 70%. Reaching 80% with confidence would require the same win record over >50 matches. But 8-1 is statistically sufficient to at least suggest a favourable match-up.

What do I know though, with my half-remembered maths class bullshit? The null-hypothesis – that I am talking out of my ass and am about to get shot down by a superior nerd – seems very likely.

Bruce Lee

Fell asleep face forward with arms underneath so hand were under pillow.
Fore arms had fallen asleep.
Could not extend them to turn off alarm clock.
Stuck in reverse T Rex position
Trying to disable alarm clock with elbows isn’t fun
Jeremy. Your pain is felt.


I love that Jay calls Mendoza out on his bullshit.

Ive got a 100% confirmed enjoyment level for this episode. Sample of 1 listen.


As I recall my Stats classes from more than a dozen years ago…Sample size has to be min of N=32 to draw anything meaningful with a 95% confidence. So sample size 10 is pretty meaningless in terms of statistical analysis. I am surprised Matt does not know this as he sounds pretty confident when he was talking about his approach and all the Math he does.

Nice job on Boosh calling him out on this, and I agree magic players have a bad tendency to generalize results based way too small sample sizes.