A live research project on deterministic prediction in a noisy system — with decisions frozen first and Reality opened afterwards.
The subject happens to be football. The engineering question is broader: can a compact representation of market structure produce repeatable information on unseen outcomes, without rewriting the rules after the result arrives?
The system does not use team news, player ratings, injuries or football expertise. It works on a structured representation built from observable market data and relations inside the prediction system.
A backtest is not enough. The relevant test is a decision created without the target outcome, frozen, and then confronted with a later result. The project therefore keeps prediction and Reality as separate phases.
The first public experiment contained 16 future matches. The page was frozen on 03 Sep 2026 before kick-off. The preserved public checkpoint below is the state of that page after 8 outcomes had arrived: 8 resolved, 8 still pending in the archived snapshot. The later branch was closed because it did not establish a robust predictive advantage.
8 SINGLE · 5 ANTI / double chance · 3 PASS.
Eight outcomes had been filled into the public page when this snapshot was preserved.
| Date | Match | System | Outside | Stability | Reality in preserved snapshot |
|---|---|---|---|---|---|
| 04 SEP | Ipswich — Liverpool | X2 · ANTI-H | A | UNSTABLE | A · 0–2 |
| 05 SEP | Brentford — Sunderland | PASS | H | PAIR STABLE A | D · 1–1 |
| 05 SEP | Brighton — Leeds | D · SINGLE @3.45 | H | STABLE 3/3 | D · 1–1 |
| 05 SEP | Fulham — Crystal Palace | H · SINGLE @2.25 | H | STABLE 3/3 | A · 2–3 |
| 05 SEP | Newcastle — Bournemouth | H · SINGLE @2.19 | H | STABLE 3/3 | D · 2–2 |
| 05 SEP | Nottingham — Tottenham | 12 · ANTI-D | H / A | UNSTABLE | D · 0–0 |
| 06 SEP | Arsenal — Chelsea | H · SINGLE @1.68 | H | STABLE 3/3 | H · 2–1 |
| 06 SEP | Everton — Man Utd | A · SINGLE @2.20 | A / MIXED | STABLE 3/3 | D · 2–2 |
| 12 SEP | Aston Villa — Nottingham | H · SINGLE @1.86 | H | STABLE 3/3 | PENDING IN SNAPSHOT |
| 12 SEP | Bournemouth — Brentford | 12 · ANTI-D | H | UNSTABLE | PENDING IN SNAPSHOT |
| 12 SEP | Crystal Palace — Ipswich | 1X · ANTI-A | H | UNSTABLE | PENDING IN SNAPSHOT |
| 12 SEP | Liverpool — Fulham | PASS | H | UNSTABLE | PENDING IN SNAPSHOT |
| 12 SEP | Sunderland — Arsenal | A · SINGLE @1.45 | A | STABLE 3/3 | PENDING IN SNAPSHOT |
| 12 SEP | Tottenham — Everton | PASS | MIXED | UNSTABLE | PENDING IN SNAPSHOT |
| 13 SEP | Man Utd — Man City | D · SINGLE @3.50 | A | STABLE 3/3 | PENDING IN SNAPSHOT |
| 14 SEP | Leeds — Newcastle | 12 · ANTI-D | H / D | UNSTABLE | PENDING IN SNAPSHOT |
See the unchanged decisions of Resolver, Universe Filter and Meta-Omega for 9–11 October 2026. The frozen disagreements are part of the test.
ENG1 used a later version of the deterministic pipeline. Two 15-match forward blocks were evaluated. Each block produced 14 SINGLE decisions and one PASS. The outcomes below were opened after the decisions were fixed.
14 SINGLE decisions. No ANTI or EDGE decisions in this block.
| Date | Match | Pred | Odd | Reality | Result |
|---|---|---|---|---|---|
| 01 SEP | Portsmouth — Derby | A | 3.20 | A | ✓ |
| 01 SEP | Preston — Bristol City | A | 2.75 | A | ✓ |
| 01 SEP | Stoke — Norwich | A | 2.25 | H | ✗ |
| 01 SEP | Swansea — Watford | PASS | — | H | PASS |
| 02 SEP | Burnley — Middlesbrough | D | 3.30 | D | ✓ |
| 02 SEP | Millwall — Wrexham | H | 2.08 | A | ✗ |
| 02 SEP | QPR — Cardiff | H | 1.87 | H | ✓ |
| 02 SEP | West Brom — Charlton | D | 3.45 | D | ✓ |
| 05 SEP | Burnley — Bristol City | H | 1.80 | A | ✗ |
| 05 SEP | Portsmouth — Cardiff | H | 2.47 | H | ✓ |
| 05 SEP | Preston — Blackburn | D | 3.10 | H | ✗ |
| 05 SEP | QPR — Middlesbrough | A | 1.99 | A | ✓ |
| 05 SEP | Sheffield Utd — Norwich | A | 2.90 | A | ✓ |
| 05 SEP | Stoke — Charlton | H | 2.06 | H | ✓ |
| 05 SEP | West Brom — Watford | D | 3.25 | H | ✗ |
14 SINGLE decisions. This is the second forward block shown separately rather than merged into the first.
| Date | Match | Pred | Odd | Reality | Result |
|---|---|---|---|---|---|
| 08 SEP | Blackburn — Sheffield Utd | A | 2.17 | A | ✓ |
| 08 SEP | Cardiff — Stoke | D | 3.45 | D | ✓ |
| 08 SEP | Southampton — Swansea | PASS | — | H | PASS |
| 08 SEP | Watford — Preston | H | 2.01 | H | ✓ |
| 08 SEP | Wrexham — Burnley | H | 1.98 | D | ✗ |
| 09 SEP | Charlton — QPR | A | 2.20 | D | ✗ |
| 09 SEP | Derby — West Brom | A | 2.34 | A | ✓ |
| 09 SEP | Norwich — Birmingham | H | 1.92 | H | ✓ |
| 12 SEP | Blackburn — Millwall | A | 2.33 | H | ✗ |
| 12 SEP | Charlton — Portsmouth | D | 3.10 | D | ✓ |
| 12 SEP | Derby — Birmingham | A | 2.15 | A | ✓ |
| 12 SEP | Middlesbrough — Norwich | H | 1.72 | H | ✓ |
| 12 SEP | Southampton — Bristol City | H | 1.58 | H | ✓ |
| 12 SEP | Swansea — Burnley | H | 2.13 | H | ✓ |
| 12 SEP | Watford — Stoke | A | 2.65 | A | ✓ |
The two blocks are shown separately because repeated success on later unseen data matters more than one merged headline number. These results are observations from the recorded experiment, not a claim that the same return will persist.
The visible predictions are only the external test. Internally, the research moved toward the representation of the system itself: what state is being measured, which structures persist between windows, and whether a decision can be justified before Reality is opened.
Instead of asking only which outcome wins, examine relations inside the ensemble and whether they remain stable across neighboring windows.
Internal system states are represented geometrically and audited for repeatability rather than treated as opaque scores.
A later hypothesis tests whether the useful structure belongs not to one static state, but to the law of change between successive states.
Current blind work asks how much field geometry can be reconstructed from upstream representations without using the target outcome.
Research record only. This page documents an engineering experiment and is not betting or financial advice.