Equity execution study.
Across two eligible, overlapping starts on 24 and 25 June 2026, the model completed at least 97.8% of a CHF 5m-equivalent fresh entry and 92.9% of a CHF 10m-equivalent fresh entry over three sessions. Every difficult residual remained visible.
- Published scenarios
- CHF-equivalent 1m–10m
- Principal schedule
- Three sessions
- Portfolio treatment
- Whole 300-name book
- Model mechanics
- Raw-to-result reproduced
Completion and modelled cost by scenario.
| CHF-equivalent fresh-entry scenario | Total target completed | Weaker side completed | Lower-volume weaker side | Base minimum names completed | Worst modelled normal cost |
|---|---|---|---|---|---|
| CHF 1m | 99.15% | 98.30% | 97.53% | 297 / 300 | 24.39 bp |
| CHF 2m | 99.02% | 98.03% | 95.71% | 295 / 300 | 24.93 bp |
| CHF 5m | 97.77% | 96.23% | 86.87% | 281 / 300 | 25.34 bp |
| CHF 10m | 92.86% | 91.46% | 76.13% | 248 / 300 | 25.51 bp |
What was tested
Fresh entry into the frozen 18 June 2026 target: two overlapping starts on 24 and 25 June, three sessions and five one-hour slices per session. The base schedule uses 0.50% participation and a 75% haircut to causal forecast volume; the lower-volume view uses 0.25% and 50%. Fills are capped again by observed current-interval volume.
Completion is absolute frozen-weighted share completion. The weaker-side measure keeps long and short progress together, and unfilled names remain residuals. Cost is the worst modelled one-way basis-point charge on executed notional. CHF equivalents use the dated 30 June 2026 exchange-rate receipt. A separate raw-input recomputation checked reference prices, volume limits, completion, side balance and cost.
Limitations
Two overlapping starts are a small, dependent sample. These modelled scenarios do not set a maximum capacity or live-capital limit. Borrow and locates, spreads, depth, queue position, realised impact, financing and operating continuity remain unmeasured live inputs.
Methods appendix
Receipts and downloadable aggregate manifest
The public manifest contains the published table, units and assumptions. It identifies the original table receipt and internal recomputation by hash; the internal inputs and security-level records remain private.
Table receipt: ff21a0bf6cbe8fe0bafa875eed5d433f6fc53f578890602403697424275d1b9d
Internal recomputation: 0a8c5ad2c79c60d5e4283ca69e69040e4ef9f871d7e9e650e00c999b656349bb
Cost-accounting method
A synthetic worked example explains the distinction between repricing an order schedule and changing its policy.