Look beyond
the scoreboard.

From a replay
to a review list.

A high kill count does not tell the whole story. NullCS looks at what happened around an encounter: where a player aimed, when they fired, and what they could see.

Models trained on gameplay data combine those measurements and rank players within the match. The report gives you a starting point for watching the replay.

Choose a demo file, wait for the analysis, then compare players and save a report. No account or demo upload is required.

Start here.
Look closer.

Players near the top deserve attention first. A score summarizes what the model found; it is not a verdict about the person.

Some replays show clear cheating. Others need context about skill, visibility, timing, and the information available to the player. Legitimate players can score highly, and cheating can be missed.

The app explains the scores in its built-in guide. A player may receive no score if the demo does not contain enough usable evidence.

Test it.
Keep improving it.

NullCS is a research beta. Tests check which cases it misses and how often legitimate players receive high scores. The dataset labels are imperfect, and more independently reviewed matches are needed.

A newer model has been trained and is still being checked. The current download uses the earlier model; research results on GitHub explain which version was tested.

CS2 updates can also change the data stored in demos. Known issues with reading that data and interpreting scores are documented on GitHub.

The app analyzes one demo at a time. Comparing the same player across several matches is a future feature.

Gerry Jones, Jr.

I started NullCS to explore whether gameplay data could help people review suspicious CS2 matches. It grew from a data science project into a local desktop tool.

I hold a B.S. in Applied Mathematics and am pursuing a master's degree in Data Science. The project covers collecting data, building models, testing their mistakes, and making the results usable.

NullCS reviews saved replays. It does not monitor your computer during play or issue bans, and it is not affiliated with Valve.