How to use Perdix

Using Perdix effectively

This walkthrough uses an analysis of Coach Curtis's gameplay patterns as an example of how to use Perdix. The goal is not to read every number; it is to move from sample quality to repeated patterns, then to the exact games and minutes worth reviewing.

example analysis

Coach Curtis

Win rate 52.2% (94-86)

Role Mid

Win patterns Mode 1: 91.5%, Mode 2: 7.4%, Mode 3: 1.1%

Loss patterns Mode 1: 88.4%, Mode 2: 10.5%, Mode 3: 1.2%

Most played champions Twisted Fate 52.8% WR 36 games Aurora 50.0% WR 26 games Ekko 62.5% WR 24 games

Main findings

Area to improve: Mid game

Based on the analysis, we found that you lose control of the game in your mid game, especially around 6-8 and 12-14 minutes.

Feature with the strongest connection to losses

Resource scaling has the strongest repeated loss-side DMD weight.

Thrown winning states

41/105 early win-like games (39.0%) still become losses.

Ability to come back

30/75 early loss-like games (40.0%) still become wins.

Snowballing

64/105 early win-like games (61.0%) convert into wins.

Focus for the next review block

Review the 6-8 minute window first, then compare it with 12-14 minutes.
Check resource spending, level timing, deaths, and magic damage taken around those windows.
Use the listed games below as the first VOD queue.

Tipping point games

4/8/6 Aurora vs Zed Tipping point: 5 min Main signals: vision rhythm, health management, level timing OC1_700633754

15/10/8 Aurora vs Katarina Tipping point: 3 min Main signals: skill level timing, level timing, kill conversion OC1_700548833

Highest loss-review moments

4/8/6 Aurora vs Zed Loss-review interval: 9-10 min, DMD score 12.51 Main signals: damage taken, experience pace, gold growth OC1_700633754

2/7/6 Aurora vs Xin Zhao Loss-review interval: 16-17 min, DMD score 12.48 Main signals: experience pace, gold growth, damage taken OC1_700462213

How to use

Simplified analysis output

After each analysis, you receive the output shown on the left. These headlines give you the most direct guidance. For example, Coach Curtis's games tend to diverge towards losses during the 6-8 and 12-14 minute windows. Use this to gauge which game phase contains your main issues.

Discover how often a game is thrown when you are ahead, how good you are at coming back from behind, and which gameplay feature has the strongest connection to your losses.

Finally, review the personalized moments identified from patterns in your games.

Coach Curtis trajectory mode clusters — Trajectory mode clusters

Coach Curtis loss-likeness timeline — Loss-likeness timeline

When do you slip?

Games moving to losing states.

The bold graph represents how your average game moves towards either a winning or losing state. When the graph is green, the game is moving further towards a win; when it is red, the game is moving towards a loss.

The lower plot showcases the percentage of games that are closer to losses at each 2-minute interval. Below 50% indicates that the majority of games are looking like wins and vice versa.

Coach Curtis's average game evolves towards a win for almost the entire game, except during the 6-8 and 14-16 minute windows. This does NOT mean he is in a losing state at these times; rather, the game becomes less winning and more losing.

Coach Curtis trajectory control progression — Trajectory control progression