MOX guides you through picks: it orders the cards in each pack, highlights a choice, and points out close alternatives. After the Draft, it suggests a deck built from your pool, directly on your PC.
Windows · Public beta · Download from GitHub
Two moments, on your PC, next to Arena. The suggestion is computed locally while you play.
Real screenshot. On the left the pack in Arena, on the right the same pack ranked by MOX.
For every pack MOX ranks the cards, with the win rate from 17Lands data and a plain verdict: strong card, good in your colours, filler. It keeps one main pick and says when equivalent alternatives exist.
Real screenshot. The deck suggested at the end of the Draft.
It suggests the deck to build from your pool: the main colours, three combinations side by side and how many lands to play, from 16 to 18.
The choices are yours. Here we explain how MOX checks its recommendations and what data it may use; no changes happen automatically.
The suggestion is calculated locally while you play.
Only people who enable it share Draft data, without their name or Arena account. A random technical identifier makes deletion possible.
Set, event and period remain separate to avoid mixing different signals.
A rule changes only when tests show a stable improvement.
On your PC: suggestions and personal statistics.
On the site: only aggregated statistics; complete traces stay in private storage.
MOX keeps one main pick but states when statistically equivalent alternatives exist.
Every published percentage includes its sample and a confidence interval.
Set, event and period are not mixed to create more attractive but less useful numbers.
Complete traces stay private; the site shows only aggregates and consent can be withdrawn.
Once there are enough contributions, MOX will publish a view of Limited data here. It is not here today: the sample is too small, and partial numbers would be worse than none.
No date: the view arrives when the data supports it.
The same divisions MOX already keeps apart when it calculates: one set is never mixed with another, and a Quick Draft is never mixed with a Premier.
How often a card is seen in a pack, how often it is picked and how often it ends up played: the three numbers say different things and stay apart.
Percentages appear only when the sample supports them, and the number of observations stays visible next to the value.
What each column measures, and where it stops. A metric without a definition is not data: it is a number.