Average career WR
51.6%
Lifetime wins divided by wins plus losses.
PUBLIC DATA LAB
Each dot is one public player file. The charts compare long-run account results, the newest stored 1v1 sample, model-adjusted skill, and the luck score. This is descriptive analysis: correlation can reveal a pattern, but it cannot prove why that pattern exists.
Average career WR
51.6%
Lifetime wins divided by wins plus losses.
Average sample WR
48.9%
Newest stored all-1v1 report.
Expected win rate
45.6%
Expected from matchup, levels, and trophy gap.
Average skill
56.8
50 means results matched the predicted win chance.
Average luck
43.2
50 is neutral; lower reads more rigged.
STRONGEST LINEAR SIGNAL
Skill score vs luck score
weak negative (r -0.33)
“Strongest” only means the largest absolute Pearson coefficient among the discovery views below — the two diagnostic views (model calibration, evidence funnel) are excluded because their relationships exist by construction. It does not make the relationship causal or necessarily stable.
HOW TO READ THE NUMBER
-1
opposite
0
no line
+1
together
The dashed line is the best linear fit. Dot colors show the luck verdict: red is more rigged, purple is neutral, gold is more blessed. Every dot links to its full case file.
01 - THE CAREER VIEW
The lifetime account win rate reported by the Clash Royale profile against the current all-1v1 Rigged Royale verdict.
r +0.21
198 players
GROUP SYNTHESIS
Comparing the 66 lowest and 66 highest luck scores
Files are sorted by luck score and split into the lowest and highest thirds. Each card averages the comparable metrics inside a group, so the tags read against the field average of every loaded file. This describes what the extremes look like; it does not prove the luck score is caused by these traits.
LOWEST SCORES / MOST RIGGED
66 files / average luck 31.2
HIGHEST SCORES / LEAST RIGGED
PLAYER-LEVEL READOUT
Showing 30 of 198 loaded files
ANALYST NOTES
Career win rate is the latest profile total (wins ÷ wins+losses), spanning the whole account. Luck and skill use at most the newest 200 stored 1v1 battles.
The luck score never takes wins or losses as a direct input. It measures matchup, timing, and level evidence against a fair-draw baseline.
Skill and analyzed win rate are related by construction — skill is the result rate left after subtracting the model's expected wins. That chart is a diagnostic, not a discovery.
The model grading itself: if predicted and actual win rates drift apart across the population, the LOGIT_INTERCEPT knob — not the players — needs adjusting.
An integrity check. The confidence shrink pulls thin files toward 50, so the left edge stays narrow and spread widens with sample size — a funnel, not a flat band.
Small samples, selection bias, changing decks, modes, and model updates all move these coefficients. A live descriptive view, not proof of matchmaking intent.
Weak positive relationship: the two measures tend to rise together.
4% shared linear variation
02 - THE STORED SAMPLE
The result rate inside the newest stored 1v1 sample against the matchup-and-sequence luck score. Wins themselves do not enter that score.
r +0.22
198 players
Weak positive relationship: the two measures tend to rise together.
5% shared linear variation
03 - PILOT VS MATCHMAKER
Whether players who outperform their matchup model also tend to receive easier or harder matchmaking samples.
r -0.33
198 players
Weak negative relationship: the two measures tend to move in opposite directions.
11% shared linear variation
04 - SHORT TERM VS LONG TERM
A long-run account result compared with the current sample's model-adjusted piloting estimate.
r +0.02
198 players
No statistically reliable relationship at this sample size (n=198).
0% shared linear variation
05 - MODEL CALIBRATION
The model's predicted win rate for each file against what the player actually posted. A well-calibrated model puts the cloud on a rising line with slope near 1; a flat cloud would mean the predictions carry no information.
r +0.57
198 players
Moderate positive relationship: the two measures tend to rise together.
32% shared linear variation
06 - THE EVIDENCE FUNNEL
Sample size against the luck verdict. Because the score shrinks toward 50 on thin evidence, small files should hug the middle and only well-evidenced files should reach the extremes. Extreme scores on tiny samples would flag a scoring bug, not rigging.
r -0.44
198 players
Moderate negative relationship: the two measures tend to move in opposite directions.
20% shared linear variation
66 files / average luck 55.1
GAPS THAT CLEAR THE NOISE FLOOR
| mrweasel#J29VYUJP / 30 analyzed |
| 52.9% |
| 37,573-33,452 |
| 73.0% |
| 58.0% |
| 74.0 |
| 57 |
| ke#8GY0V0YG / 29 analyzed | 48.7% | 31,091-32,711 | 52.0% | 51.0% | 50.0 | 56 |
| kierowca340#2Q9PGLRJ / 30 analyzed | 49.3% | 30,106-31,000 | 50.0% | 52.0% | 46.0 | 56 |
| claudiao#89Y2RC0Y / 29 analyzed | 53.5% | 29,941-26,032 | 52.0% | 50.0% | 53.0 | 44 |
| V1s3r10n#RLRVPL0C / 27 analyzed | 49.4% | 21,384-21,903 | 48.0% | 49.0% | 48.0 | 40 |
| Tukie69#LV9J9YJJ / 24 analyzed | 50.7% | 20,993-20,381 | 58.0% | 56.0% | 53.0 | 38 |
| CODE Sweaty#9GRULPJQ / 8 analyzed | 58.1% | 22,822-16,438 | 75.0% | 54.0% | 62.0 | 50 |
| jupel#PRUQVC2U / 30 analyzed | 52.2% | 19,284-17,673 | 63.0% | 50.0% | 71.0 | 36 |
| Dominic alexis#UYUCC9LRG / 30 analyzed | 47.2% | 16,803-18,798 | 50.0% | 50.0% | 50.0 | 44 |
| г๏๓๓є͢͢͢ɭ#R9JJGRVV / 23 analyzed | 49.6% | 15,981-16,270 | 61.0% | 52.0% | 62.0 | 55 |
| earf#P202G2YUU / 22 analyzed | 48.8% | 15,615-16,392 | 55.0% | 58.0% | 45.0 | 56 |
| Ÿøügõw#8CVYUCQ20 / 22 analyzed | 46.8% | 14,749-16,779 | 55.0% | 49.0% | 57.0 | 38 |
| BLIST#8PQPYLGU / 29 analyzed | 55.8% | 17,530-13,875 | 62.0% | 49.0% | 70.0 | 48 |
| porax#8U9PPUQRL / 30 analyzed | 47.2% | 14,801-16,530 | 40.0% | 48.0% | 38.0 | 53 |
| fsk007#QJGG9GCV0 / 30 analyzed | 46.4% | 14,453-16,671 | 57.0% | 55.0% | 52.0 | 48 |
| jtebasta n roca#2LVRG9GPY / 26 analyzed | 50.7% | 15,750-15,338 | 54.0% | 54.0% | 50.0 | 59 |
| tavo92#LQVC0VC / 18 analyzed | 48.7% | 15,099-15,936 | 50.0% | 49.0% | 51.0 | 40 |
| &thhg#PYUJJ2Q2Y / 27 analyzed | 48.2% | 14,924-16,024 | 48.0% | 54.0% | 40.0 | 59 |
| vwuilles#LR0Q98CQ / 30 analyzed | 49.3% | 14,732-15,139 | 43.0% | 49.0% | 41.0 | 30 |
| SirLunze#RGPUVJYU / 23 analyzed | 52.8% | 15,585-13,939 | 43.0% | 41.0% | 53.0 | 39 |
| [chris]#P2RCYJ2Q / 21 analyzed | 58.2% | 17,053-12,252 | 62.0% | 47.0% | 69.0 | 46 |
| 牛逼#VC2CLUL82 / 200 analyzed | 48.9% | 14,171-14,816 | 40.0% | 34.0% | 83.0 | 29 |
| Rubinho14#QQ922RULP / 25 analyzed | 51.3% | 14,863-14,096 | 52.0% | 44.0% | 61.0 | 57 |
| AK SOULZ#8P08GU2GU / 30 analyzed | 57.7% | 16,604-12,190 | 67.0% | 51.0% | 75.0 | 38 |
| BenYii#90VLGL / 30 analyzed | 51.5% | 14,774-13,901 | 57.0% | 51.0% | 60.0 | 51 |
| made6m6#QY2UVUY2 / 30 analyzed | 49.7% | 14,172-14,325 | 50.0% | 49.0% | 51.0 | 53 |
| Budalhoco#9UP2VPC9C / 30 analyzed | 51.3% | 14,505-13,776 | 50.0% | 57.0% | 37.0 | 49 |