root@kaus.com:~/projects$ cat r6-replay-lab.md
R6 Replay Lab
researchPersonal · 2026–present
Kbit · page guide
1 / 4
Use the section map on long studies.
Jump between methodology, results, experiments, limitations, and related writing without losing your place.
2 / 4
The narrative carries the claim.
Read each section as problem, method, evidence, and boundary. Negative results remain part of the story.
3 / 4
Inspect figures alongside the narrative.
Read figures in place, scroll wide charts or tables horizontally, and follow any source links for the underlying evidence.
4 / 4
Keep exploring the corpus.
A local replay-analysis project for testing Rainbow Six Siege performance hypotheses against round-level evidence.
I am starting a replay-analysis lab to understand how performance changes across roles, teams and eras. The project begins with preserving recordings and validating parser output, then connects round events to specific review questions. The repository currently holds the specification and intake checkpoint; native replay analysis remains unvalidated.
Technologies
- Python (planned)
- Replay event data
Skills
- Evaluation design
- Data provenance
- Performance analysis
The question
A seasonal K/D ratio mixes individual decisions with opponent strength, team role, account history and sample size. I want to test explanations for changes in performance against complete matches and specific rounds, with room for the evidence to disprove an initial interpretation.
The planned method
Preserve stable recordings with hash checks, inventory their builds, and validate a reviewed decoder against independent match evidence before using exported events. Round-level review will examine opening involvement, trades, manpower changes and objective context where the required fields are reliable.
The retrospective will separate outcomes in the conditions actually played from comparisons under similar conditions. Historical ratings and account ownership need their own evidence; missing context stays visible.
Current evidence
The project currently has a repository, specification and intake checkpoint. The original starter package still needs to be imported and inspected, and no native recordings have been decoded or analyzed for this project. There are no validated tactical findings or player-improvement claims yet.
notes
- The original starter package has not been imported or independently tested. Personal recordings and account mappings remain private.