Beat the textbook rules
Five fixed benchmark tracks (SCFO-BENCH v1.7, engine v1.15.0). Configuration and seed are fixed, so everyone graded on this engine sees byte-identical data. The house baselines are the three detectors from the proof, run here exactly as your model would be: against a database copy with no answer key in the directory. Beat the textbook rules and publish it.
The entry track: a compact SAP ECC manufacturer with the full four-cycle fraud surface. Fast to generate, fast to evaluate - start here.
| # | Detector | Precision | Recall | F1 | Grade | When |
|---|---|---|---|---|---|---|
| 1 | Targeted forensic rules (ceiling) house baseline | 0.05 | 0.98 | 0.10 | F | |
| 2 | Textbook CAAT rules house baseline | 0.04 | 0.63 | 0.07 | F | |
| 3 | Generic statistical anomaly detection house baseline | 0.02 | 0.24 | 0.04 | F |
The standard reference track: SAP ECC manufacturing at medium-intensity fraud across procurement, revenue, treasury and the close. The headline comparison number.
| # | Detector | Precision | Recall | F1 | Grade | When |
|---|---|---|---|---|---|---|
| 1 | Targeted forensic rules (ceiling) house baseline | 0.05 | 0.98 | 0.10 | F | |
| 2 | Textbook CAAT rules house baseline | 0.04 | 0.65 | 0.07 | F | |
| 3 | Generic statistical anomaly detection house baseline | 0.01 | 0.10 | 0.02 | F |
Five modules including HR/payroll at high intensity: ghost employees, payroll diversion, collusion between the vendor master and the payroll file.
| # | Detector | Precision | Recall | F1 | Grade | When |
|---|---|---|---|---|---|---|
| 1 | Targeted forensic rules (ceiling) house baseline | 0.12 | 0.95 | 0.22 | F | |
| 2 | Textbook CAAT rules house baseline | 0.11 | 0.72 | 0.19 | F | |
| 3 | Generic statistical anomaly detection house baseline | 0.07 | 0.19 | 0.10 | F |
The Oracle Fusion track: retail seasonality, module schemas (AP/AR/CE/GL), medium intensity. Tests whether a model's skills transfer across ERP architectures.
| # | Detector | Precision | Recall | F1 | Grade | When |
|---|---|---|---|---|---|---|
| 1 | Targeted forensic rules (ceiling) house baseline | 0.12 | 1.00 | 0.21 | F | |
| 2 | Textbook CAAT rules house baseline | 0.09 | 0.62 | 0.16 | F | |
| 3 | Generic statistical anomaly detection house baseline | 0.02 | 0.34 | 0.03 | F |
S/4HANA with ACDOCA: the post-migration world. Pharma procurement discipline, medium intensity. Models trained on ECC structure meet the single-table journal here.
| # | Detector | Precision | Recall | F1 | Grade | When |
|---|---|---|---|---|---|---|
| 1 | Targeted forensic rules (ceiling) house baseline | 0.05 | 0.98 | 0.10 | F | |
| 2 | Textbook CAAT rules house baseline | 0.04 | 0.63 | 0.07 | F | |
| 3 | Generic statistical anomaly detection house baseline | 0.01 | 0.09 | 0.02 | F |
Create a free account, open the Benchmark tab and pick an official track. Press Generate this track to build its data on your account: every plan can, the free one included, and each track costs one generation (on the free plan, one of the month's 5). Run your detector on that copy, then paste the record ids it flags. You get the full scorecard either way; ticking publish with a model name puts it here.
Track definitions are fixed; when the engine moves the tracks are regenerated on it, and a score is ranked beside the baselines only when both were graded on the same engine, never silently. Baselines regenerate byte-identically from proof/bench_baselines.py on the engine version they name.