Real positions at the top — chosen by a human from the machine's candidates. Below them, the machine's own unselected record: a frozen account that takes every qualifying entry, no discretion. Neither is the registered receipt — the method behind these lives on the track record. Factual records of hypothetical positions; not financial product advice.
Actual holdings, chosen by a human from the machine's candidates. Entries are authored; every price and return below is computed from the data lake's bars at export — never hand-typed. The machine's own unselected record is below.
Gold-antimony at Reynolds Range (Aileron Province, NT). Sabre carried the June program; the whole discovery still fits inside a 91 by 97 metre patch. Held from before the 6 August release, average 4.8c.
Read the research →Aracuai Salinas lithium valley ground ex Rio Tinto, run by the Colina team; the bet is grade under the weathering front, assays mid September to mid October. Position disclosed, average 6.8c.
Read the research →Resource specialists and investment banks both net-buying at least $5k the same session · no price-sensitive announcement in the 3 calendar days ending that session · next-session volume confirmation at 1.5× the 30-day average · entries skipped after an up-gap over 5% · buy the following open (or close, flagged, when the morning list published late) · sell at the close 10 sessions later · no stop Costs on every fill: $5 a side plus a spread allowance by price band (1.5% under 10¢, 0.6% to 50¢, 0.3% above), applied to every fill.
Takes every qualifying entry, no discretion — the control group for the portfolio above. Config paper-book-v1 · frozen (any edit before freeze-end voids this record).
| Ticker | Hit | Entry | Basis | Exit | Net |
|---|---|---|---|---|---|
| DEV | 2026-08-17 | 2026-08-21 @0.215 | open | 2026-09-04 @0.215 | -1.4% |
| GTI | 2026-08-24 | 2026-08-28 @0.03 | open | open (marked) | +6.7% [open] |
| NPM | 2026-08-24 | 2026-08-28 @0.012 | open | open (marked) | -16.7% [open] |
| NVX | 2026-08-24 | 2026-08-28 @0.125 | open | open (marked) | -21.6% [open] |
| OR3 | 2026-08-24 | 2026-08-28 @0.084 | open | open (marked) | -11.9% [open] |
| MEM | 2026-08-27 | 2026-09-02 @0.005 | close_fallback | open (marked) | +20.0% [open] |
| ALM | 2026-08-28 | 2026-09-03 @0.01 | open | open (marked) | +10.0% [open] |
| CUV | 2026-08-28 | 2026-09-03 @8.42 | open | open (marked) | -3.1% [open] |
Declined, on the record: G50 (up-gap, would-have-been -3.8%) · SNX (up-gap, would-have-been 0.0%) · AW1 (up-gap, would-have-been -15.0%) · CBE (up-gap, would-have-been -1.3%) · KLI (up-gap, would-have-been -8.5%) · CBE (up-gap) · ALR (up-gap) · SNM (up-gap) · CTO (up-gap) · ACW (up-gap) · ACW (up-gap)
Open 2 · closed 0 — the comparison publishes once 3 positions close (house rule).
Hand-picked entries, distinct from the machine above and from the registered receipts. Every pick is logged with its thesis and a falsifier BEFORE the outcome exists; scored quarterly against the machine panel. Passes are logged too.
Picks 0 · resolved 0 — counts only until 5 resolve (house rule).
Empty, honestly. No discretionary picks have been logged yet — the journal starts at zero, in public, like everything else here.