Sylas official champion artworkCHAMPION LAB · EARLY DATA
SYLAS LAB

Land the chain.
Spend the passive.
Steal with purpose.

Deterministic E recast, E→Q, Q weaving, and stolen-ultimate follow-through evaluators.

See the mechanicsLearn SylasHow we know
WHAT TO WATCH

The important question is not whether Sylas stole an ultimate, but what the game state made valuable.

Champion-specific action sequences the AI sees in replays now calculate with exact source-event references. Population curves below remain deterministic outcome context while mechanic cohorts are accumulated.

Early dataWe wait for enough players before calling something a benchmark.
Exact actions measured One score per player Compared by current rank Missing evidence stays unavailable
economy

Gold per minute

What it measures
Gold earned divided by game duration. This is a deterministic outcome baseline, not a replay-mechanics score.
Why it matters
More verified conversions usually mean more of the mechanic’s available value is being realized. Review the evidence before assuming every opportunity should be forced.
MASTER401.8n=21 · 26 eligible games · early sample
GRANDMASTER420.8n=31 · 71 eligible games · benchmark ready
CHALLENGER420.5n=39 · 210 eligible games · benchmark ready
Observed medians run from 401.8 in master to 420.8 in grandmaster.

91 players across 307 eligible replays in 3 ranks.

Curves show observed player distributions. Anything the AI could not see in the replay is unavailable, never scored as zero.
SYLAS REPLAY LAB

Current-rank outcomes surrounding the replay analysis.

These curves use official participant totals attached to AI-watched replay games. The champion-specific scores are defined in the mechanics section and calculate per replay.

Your metrics
0/4
Players
91
Player-games
307

How Sylas mechanics are measured

6 champion-specific metrics now calculate or fail closed, alongside deterministic outcome context. Geometry/state-dependent scores remain unavailable.

Games analyzed 9/9/2026 · Player names removed · Rank uses each player’s verified current tier.
LEARN SYLAS

Turn mechanics into better decisions.

See what matters, when it matters, and one drill to take into your next game.

Your mechanicsDrillsMatchupsGame planHow we know
MECHANICS LIBRARY

The Unshackled, made measurable.

01Observed

E recast interval and E→Q delay

A champion-specific deterministic evaluator, run over what the AI sees in the replay, with fixed opportunities and exact evidence references.

Measured as
Returns value, numerator, denominator, exclusions, and source events from the champion action sequence.
Training drill
Review the fastest, slowest, and missing sequence opportunities across five games.
02Observed

Q→attack passive-rhythm proxy

A second champion-specific execution family measured independently from match outcome.

Measured as
Deterministic rate or timing distribution over eligible actions the AI observed in the replay.
Training drill
Classify failures as intentional, denied, or mechanically omitted.
03Coverage gated

verified passive-stack conversion

This deeper score requires state beyond action order and is not silently inferred.

Measured as
Unavailable until the AI can reliably see identity, state, and outcome in the replay.
Training drill
Manually label five examples and counterexamples.
04Coverage gated

stolen-ultimate target value

This mechanic needs synchronized geometry or state evidence before scoring.

Measured as
Unavailable rather than estimated from aggregate outcomes.
Training drill
Review initiating action, target state, result, and exit.
MATCHUP APPLICATION

The same score can mean something different by lane.

Open live Sylas matchup matrix

VS

Syndra

Priority study

Lane ruleInterpret sequence frequency through opponent cooldowns, access, and retaliation.

Best windowReview cooldown and wave state around each eligible sequence.

WatchE recast interval and E→Q delay · verified passive-stack conversion

VS

Ahri

Contextual

Lane ruleInterpret sequence frequency through opponent cooldowns, access, and retaliation.

Best windowReview cooldown and wave state around each eligible sequence.

WatchE recast interval and E→Q delay · verified passive-stack conversion

VS

Akali

Contextual

Lane ruleInterpret sequence frequency through opponent cooldowns, access, and retaliation.

Best windowReview cooldown and wave state around each eligible sequence.

WatchE recast interval and E→Q delay · verified passive-stack conversion

GAME PLAN

Use the mechanic for the right play.

Sequence before outcome

The replay score identifies what happened mechanically; gold and CS describe the resulting game trajectory.

See the matching mechanic

Review the denominator

A champion metric needs eligible opportunities, not merely successful highlight plays.

See the matching mechanic

Fail closed

When target state or geometry is missing, Manalytics reports unavailable instead of inventing a score.

See the matching mechanic
HOW WE KNOW

See exactly what is measured.

We show patterns only after enough real players support them. Early results stay labeled, and missing information stays unavailable.

Measured in replay Coaching guidance Early pattern
Open measurement details
6 mechanics measured

Every result defines what counts, when it applies, and which actions are excluded.

2 things we cannot score yet

verified passive-stack conversion, stolen-ultimate target value stay unavailable until the replay shows them reliably.

KEEP READING

The Sylas lab is one study of many.

Open the research library