Fewer than 30 games selected. Differences between wins and losses here are mostly noise.
— wins, — losses, —% (95% CI —–—%). With 20 games the interval is so wide that this section shows what happened, not how good you currently are. Newest game first.
Three habits your data points to, each as a share of games where you hit it. Compare the last 20 games with everything before. A habit moves faster than your win rate, so this is how you see progress.
Each row shows how far apart wins and losses are, in standard deviations (Cohen's d), with a 95% CI. Right of center: higher in wins. Bold: still clear after correcting for testing 23 metrics at once (Benjamini–Hochberg). Faded: not distinguishable from noise. A difference is not a cause; many values are a result of how the game went.
| Metric | Wins | Losses | Difference | 95% CI |
|---|
Win rate by your gold difference @15 against your direct opponent. Ahead: more than +500, behind: less than −500. Whiskers: 95% CI.
One dot per game. Horizontal: your gold difference @15. Vertical: that of the other four.
Losses where you were more than 500 gold ahead at 15 minutes. These are the games where a replay from minute 14 teaches the most: where were you at dragon and herald, when did you die, did you turn your lead into towers? The match ID is in your client under match history.
| Date | Champion | Opponent | Gold @15 | Team @15 | K/D/A | Deaths after 14 | Replay: your deaths after 14 | Towers | Min | Match ID |
|---|
Same opponent, same games: each row is the average difference between you and the player in your position, in standard deviations, with a 95% CI. Right always means better for you, so the direction is flipped for deaths. Bold: still clear after correcting for 12 comparisons. This is a fair comparison with players at your elo, not with averages from the internet.
| Metric | You | Opponent | Difference | 95% CI | Difference in losses |
|---|
Item and rune names from Data Dragon. Faded bars and rows: fewer than 10 games. A higher win rate with an item does not mean the item is better: players who are ahead often build differently and faster. These numbers show what you build and how it went, not what you should build.
An early item is usually the result of a good lane, not its cause.
| First three core items | Games | Win rate % | CI low | CI high | Item 1 done (min) |
|---|
Win rate by your deaths before 14:00. Faded bars: fewer than 10 games.
Win rate by game length. Shows whether you tend to close early or carry long games.
Win rate by how often you died after minute 14. Losing games also produce deaths, so read this as a budget to stay under, not as a cause.
Deaths per game in each 2-minute window. Later windows only count games that lasted that long. At least 5 games per point.
A session is a run of ranked games with at most 90 minutes between them, counted across all champions. Left: which game of the session it was. Right: how the previous game ended. A drop after losses or in later games would hint at taking breaks, not prove it.
Per lane opponent. Top: opponents with 5 or more games, hardest lane first (by gold @15). Below: fewer than 5 games, which shows what happened, not what to expect. Gold difference @15 says more about a matchup than win rate.
| Opponent | Games | W–L | Win rate (95% CI) | Gold @15 | CS diff @10 | Deaths before 14 |
|---|
Rolling average over 20 games. Dashed: 50%.
| Date | Champion | Result | K | D | A | KP | CS/min | Gold @15 | Opponent | Min |
|---|