Methodology
What actually moves your winrate.
OpenMagic reads MTG Arena Standard Bo1 matches captured locally by the Companion engine and recalculates every number live. This page explains where each figure comes from and how far you should trust it.
Where the data comes from
Matches are read from the match log MTG Arena already writes on your machine, captured locally by the Companion engine and synced to your account only after you consent.
Nothing is scraped from the game client, no memory is read and no file is modified. Your hub is recalculated from the recorded matches and refreshes on its own.
Colour legend
Winrate colouring is fixed across the whole site: green from 55% up, amber between 45% and 55%, red below 45%.
- • 55% and above — performing
- • 45% to 55% — neutral band
- • Below 45% — underperforming
How the tier is estimated
Your deck is compared with metagame archetypes that share its colours. When no archetype of the same colour pair or trio exists in the current metagame, the previous set's metagame is used as reference.
The tier is an estimate, not an official ranking. Decks with fewer than 8 matches are flagged as low sample.
Core metrics
Ranked winrate
Standard Bo1 competitive queue, counted separately from the casual queue so ladder pressure never mixes with testing games.
Normal winrate
Standard Bo1 casual queue. Useful to validate a new list before taking it to ranked.
Protour rate
Wins inside the current 16-match window — the short-term read on whether the deck is trending up or down right now.
Rolling winrate (last 16 matches)
Each point looks at that match plus the 15 before it, so a slump or a hot streak is visible over time instead of being flattened by the lifetime average.
On the play vs on the draw
Who starts is a coin flip the game decides for you, not a decision — so it is reported apart from the rest.
Deck version
Every edit creates a new version. Statistics can be read for a single version or for all versions of the deck.
Adjusted winrate (shrinkage)
A 3-0 record is not a 100% winrate. Small samples are pulled toward the average of comparable matchups, so a handful of games never reads like a proven edge.
Alongside the adjusted number we always show the real record and a 95% confidence interval, so you can see how wide the uncertainty still is.
Is this real, or a coincidence?
When hundreds of conditions are tested at once, some will look decisive purely by chance — like flipping a coin long enough to hit five heads in a row.
Findings are corrected with the Benjamini-Hochberg procedure at a 5% false discovery rate. Comparisons that survive the correction are marked; everything else is descriptive and should be treated as a lead to watch, not a conclusion.
Advanced statistics
Opening hand and mulligans
Lands kept, curve shape, creature density and mana-flood signals — cross-referenced with the outcome of the match.
Cards in hand and cards cast
When a card showed up in the opening hand, or hit the table, did the game lean toward a win or a loss? Minimum of 2 matches per card.
Why you won, why you lost
The card recorded as responsible for the outcome, grouped by opponent archetype.
Time of day and weekday
Circular time-of-day analysis in your local timezone, plus a weekday table. Only slots with enough matches are shown.
Fatigue and session length
Matches already played that day, consecutive matches and time spent in the session, tested against results.
Archetype families
Opponents you meet most, and how those families cluster by hour — frequency, not winrate.
These readings describe the matches you have already played — they do not predict the next one. Strategic commentary about matchup roles and card plans is opinion, not a measurement produced by the engine.