The thing about magic which sets it apart in terms of rationality training is that you can't go at it with theory alone. It takes hours of experimentation to get your deck built right, testing a variety of maindeck* and sideboard* permutations until you find one which is optimal against the current metagame**. As soon as the metagame changes, it's back to testing. Since magic is never "solved," at least not permanently, it forces you to become good at coming up with guesses and testing them, at least within this specific domain.

terms for the bewildered:

*maindeck and sideboard: your deck consists of at least 60 cards in the maindeck and exactly 15 in the sideboard. During a match (at least 3 games, sometimes 5) after the first and second games, you can trade cards from your maindeck for cards in your sideboard.

**metagame: the game outside the game, in other words, the different decks you are likely to face. In any given format, there are usually a handful of strong decks which dominate the field in some variation or another. As such, you end up playing against them more often then other decks.

the advantages of rationality may be drowned out by the effect of raw brains and the ability to systematize

If this is true, I'd be skeptical of it faring much better elsewhere.