Everything in the report is built from the metadata of your games — results, openings, colours, time controls, clock outcomes, and computer evals where the platform provides them. There is no engine analysis behind it yet, so it deliberately avoids any claim about move quality.
The core idea: relative to your level
Most raw stats just measure how strong you are. A 1000-rated player almost never draws and frequently runs out of time — that’s their level, not their style. So the report measures you as a residual against what’s typical for your rating and time control, or as a comparison of you to yourself. That’s the only honest way to say you’re “decisive” or “time-pressed” — relative to players like you, not to Magnus.
How leaks are found and priced
A leak is a repeatable pattern — an opening, a colour, the clock, or failing to convert winning positions — where your score sits measurably below what the Elo difference in those games predicts. Each candidate is priced in estimated rating points from the size of that gap, then ranked by cost × fixability. The estimates are clearly labeled approximations derived from your own games — no invented cohorts. Patterns with fewer than 10 games are counted but never judged, and every figure in the report carries its sample size.
Healthy zones
Until the scan cohort is large enough for real per-band percentiles, vitals like time-trouble rate are benchmarked two ways: against an absolute healthy zone for your time control, and against the typical value for your rating band. Both baselines are coarse curves from public rating distributions — approximate by design, and replaced with measured baselines as more players are scanned.
The four style axes
Drawish ↔ Decisive
How often your games end in a draw, compared with what's typical at your rating and time control.
Draw rates climb steeply with strength — a 1000 almost never draws, a 2600 often does. So we don't compare you to everyone; we compare you to your level. Drawing far less than your peers means you fight for a result; drawing more means you play it safe.
Specialist ↔ Universal
How concentrated your opening repertoire is — the effective number of distinct openings you play.
Computed as an inverse-Simpson concentration of your opening distribution, so it reflects how you spread your games, not how many you've played. Someone who plays one opening 80% of the time scores Specialist no matter how many total games; someone who spreads evenly scores Universal.
Even ↔ Colour-reliant
The gap between your results with White and with Black.
A purely within-player comparison — it can't be wrong. A large gap means one colour carries your rating (usually a fixable repertoire hole on the weaker side); a small gap means you're two-sided.
Steady ↔ Time-pressed
How often you lose on the clock, compared with what's normal for your time control.
Flagging is mostly about the format — bullet players flag constantly, classical players almost never. We compare you to your format, so 'time-pressed' means the clock costs you more than it costs your peers, not just that you play fast.
The 9 archetypes
Your archetype is the single most pronounced tendency across the four axes (or a balanced read when nothing stands out). It’s a label for a region of the map, not a box — and the card always shows the numbers behind it.
The Berserker
Forces decisive games and lives in time pressure — sharp, chaotic, high-variance.
The Fighter
Plays for a win and draws far less than peers, but keeps the clock under control.
The Grinder
Comfortable in long games and draws — squeezes results patiently.
The Specialist
A narrow, well-drilled opening repertoire — steers games into familiar territory.
The Universalist
A wide opening repertoire — varied and hard to prepare for.
The One-Sided Player
Results lean heavily on one colour — much stronger as White or Black.
The Time Scrambler
Flags far more than peers — the clock, not the position, often decides.
The Metronome
Unusually steady on the clock for the format — rarely flags.
The All-Rounder
No single tendency dominates — balanced across decisiveness, repertoire, colour and clock.
What this is not
Not a skill rating.Two players of very different strength can share an archetype — that’s the point. Style and strength are different things.
Not engine-backed yet.We measure choices and outcomes, not move accuracy. Move-by-move analysis is the paid tier we’re building.
The “typical for your level” baselines are approximate.They’re coarse curves from public rating distributions. As we scan more players, we’ll replace them with measured baselines per rating band — and the residual axes get sharper.
Low-sample claims are gated.If you haven’t played enough of something, that number is marked low-confidence or not shown at all — never overstated.