CQ
SIGN IN
All Players Player Rankings All Teams Team Rankings Achievements
All Matches Events
Predictions Shadow Bets Models Pick'em Verify
LAB — Team Builder
Whitepaper Transparency Why It Fails
Hub Home Rankings Matches Join Queue Servers
SIGN IN
PREDICTIONS SHADOW BETS MODELS VERIFY
INTEL / PREDICTIONS / MODELS

CQ-GESUS Intelligence Suite

12 specialized models trained on 25M+ parsed CS2 rounds. Every metric is live from the immutable prediction ledger.

MODEL SELECTOR LOGIC

Each upcoming match receives exactly one prediction from the best available model. NEXUS is selected when a liquid Polymarket market exists — it blends sharp market signal with internal output. ORACLE is the fallback for all other matches. Sub-models feed into the TITAN meta-ensemble, currently in development.

ACTIVE & DEPLOYED
TITAN v2
Match Outcome
DEPLOYED AUC 0.742

The meta-ensemble. Strongest overall match predictor.

239
Preds
57.8%
Accuracy
-12.0%
ROI
Ensemble mean probability
PHANTOM confidence
Ensemble View details →
ORACLE v2
Match Outcome
DEPLOYED AUC 0.7531

Market-independent deep stats match predictor.

223
Preds
54.5%
Accuracy
-19.4%
ROI
Map-specific Elo differential
Overall Elo differential
Gradient Boosting View details →
NEXUS v2
Market-calibrated Outcome
LIVE

Where ORACLE signal meets sharp market price.

112
Preds
57.1%
Accuracy
-0.7%
ROI
ORACLE log-odds
Polymarket implied prob
Log-odds blend View details →
PHANTOM v2
Economy & Round Type
DEPLOYED AUC 0.7625

Economy specialist. Knows who wins when it counts.

204
Preds
56.8%
Accuracy
—
ROI
CT full-buy win rate (rolling 20)
T full-buy win rate (rolling 20)
Gradient Boosting View details →
MIRAGE v6
Per-Map Match Outcome
DEPLOYED AUC 0.615

Eight separate models. One for every CS2 map.

No live prediction data yet.
Map-specific Elo differential
Overall Elo differential
Per-map Gradient Boosting View details →
SPECTER v1
Live Round-by-Round
DEPLOYED AUC 0.9035

Round-by-round live win probability. AUC 0.9035.

No live prediction data yet.
CT/T round score
Economy state gap
Gradient Boosting (sequential) View details →
CIPHER v1
Player Performance
DEPLOYED R²=0.307

Predicts individual player output for upcoming matches.

No live prediction data yet.
Player rolling Rating (15-map)
ADR rolling average
Gradient Boosting (regression) View details →
CLUTCH v1
Clutch / 1vX
DEPLOYED AUC 0.877

When it's 1vX — who comes through?

No live prediction data yet.
Enemies remaining
Kills before clutch entry
Gradient Boosting (sequential) View details →
UTILITY v2
Utility Dominance
DEPLOYED AUC 0.649

Flashbangs win rounds. UTILITY measures who controls them.

No live prediction data yet.
Team Elo differential (per-match)
Flash assist differential
Gradient Boosting View details →
PISTOL v1
Pistol & Force-buy Rounds
DEPLOYED AUC 0.9053

Pistol rounds decide maps. AUC 0.9053.

No live prediction data yet.
CT pistol win rate (rolling)
T pistol win rate (rolling)
Gradient Boosting (sequential) View details →
SYNERGY v2
Team Chemistry & Roster
DEPLOYED AUC 0.639

Some lineups are greater than the sum of their parts.

No live prediction data yet.
Team Elo differential
Pairwise co-win rate (top pair)
Gradient Boosting View details →
RETIRED
APEX v3 RETIRED

The original baseline model. Now retired.

AUC 0.67
All accuracy and P&L metrics update in real-time from the immutable prediction ledger. Shadow bets use half-Kelly sizing. P&L units = % of a $1,000 starting bankroll. AUC values are measured on chronologically held-out test sets — no future data leaks.