The AI Sports Almanac
Home 📊 All Sports Results ⚽ World Cup 🏉 NRL 🏉 AFL 🏉 State of Origin 🏉 Super Rugby 🏀 NBA 25-26 ⚾ MLB Picks FAQ The Algorithm Login Account

⚙ Important context — draw handling

The engine's draw-detection module was not integrated until 20 June (the Australia match), after development issues. Before that date, the engine could not call draws — it picked a team in every fixture, and drawn matches counted as losses. This is the single biggest factor in the early record. From 20 June onward the engine began correctly identifying draws, and accuracy lifted sharply — see the split below. The tournament-wide figure therefore understates the current engine, which carries the draw module.

36–23
Overall W–L
61.0%
Hit Rate
59
Matches Graded
66.7%
Since 20 Jun
3
Pending (R32)
Before draw module (13–19 Jun)
50%
10W–10L · draws uncallable
After draw module (20 Jun–2 Jul)
66.7%
24W–12L · draws live

Full Match Ledger

Provenance. Every prediction on this page — group stage and knockouts, 14 June to date — was generated by the MAX VII reasoning engine alone. The quantitative WC26 pipeline (Dixon-Coles / XGBoost / Monte Carlo) has not been activated at any point this tournament. This record is pure reasoning-layer performance. USA v Bosnia (2 Jul) carried no logged call and is shown for completeness but excluded from all W–L figures. Season record to date: 36–23 (61.0%) · since draw module (20 Jun): 24–12 (66.7%) · three Round-of-32 calls pending today.
DateFixtureMAX VII PickScoreResult
14 Jun Australia v Türkiye Australia 2-0 WIN
13 Jun USA v Paraguay USA 3-1 WIN
13 Jun Canada v Bosnia Canada 1-1 LOSS
14 Jun Haiti v Scotland Scotland 0-1 WIN
14 Jun Brazil v Morocco Brazil 1-1 LOSS
14 Jun Qatar v Switzerland Switzerland 1-1 LOSS
15 Jun Sweden v Tunisia Sweden 5-1 WIN
15 Jun Ivory Coast v Ecuador Ecuador 1-0 LOSS
15 Jun Netherlands v Japan Japan 2-2 LOSS
16 Jun Saudi Arabia v Uruguay Uruguay 1-1 LOSS
16 Jun Belgium v Egypt Belgium 1-1 LOSS
16 Jun Spain v Cape Verde Cape Verde 0-0 LOSS
17 Jun France v Senegal France 3-1 WIN
17 Jun Argentina v Algeria Argentina 3-0 WIN
17 Jun Austria v Jordan Austria 3-1 WIN
18 Jun Uzbekistan v Colombia Colombia 1-3 WIN
18 Jun England v Croatia Draw 4-2 LOSS
19 Jun Czechia v South Africa Czechia 1-1 LOSS
19 Jun Switzerland v Bosnia Switzerland 4-1 WIN
19 Jun Mexico v South Korea Mexico 1-0 WIN
⚙ Draw-handling module integrated — 20 Jun
20 Jun USA v Australia USA 2-0 WIN
20 Jun Scotland v Morocco Morocco 0-1 WIN
20 Jun Türkiye v Paraguay Türkiye 0-1 LOSS
20 Jun Netherlands v Sweden Draw 5-1 LOSS
21 Jun Germany v Ivory Coast Germany 2-1 WIN
22 Jun New Zealand v Egypt Egypt 1-3 WIN
22 Jun Uruguay v Cape Verde Uruguay 2-2 LOSS
22 Jun Belgium v Iran Belgium 0-0 LOSS
23 Jun Jordan v Algeria Algeria 1-2 WIN
23 Jun Norway v Senegal Norway 3-2 WIN
23 Jun Argentina v Austria Argentina 1-0 WIN
24 Jun Colombia v DR Congo Colombia 1-0 WIN
24 Jun Panama v Croatia Croatia 0-1 WIN
24 Jun Portugal v Uzbekistan Portugal 5-0 WIN
24 Jun England v Ghana England 0-0 LOSS
25 Jun South Africa v South Korea South Korea 1-0 LOSS
25 Jun Morocco v Haiti Morocco 4-2 WIN
25 Jun Scotland v Brazil Brazil 0-3 WIN
25 Jun Switzerland v Canada Draw 2-1 LOSS
25 Jun Bosnia v Qatar Bosnia 3-1 WIN
26 Jun Ecuador v Germany Germany 2-1 LOSS
26 Jun Curacao v Ivory Coast Ivory Coast 0-2 WIN
26 Jun Japan v Sweden Japan 1-1 LOSS
26 Jun Paraguay v Australia Australia And Draw 0-0 WIN
27 Jun Norway v France France 1-4 WIN
27 Jun Senegal v Iraq Senegal 5-0 WIN
27 Jun New Zealand v Belgium Belgium 1-5 WIN
27 Jun Egypt v Iran Draw 1-1 WIN
30 Jun Brazil v Japan Brazil 2-1 WIN
30 Jun Germany v Paraguay Germany 1-1 LOSS
30 Jun Netherlands v Morocco Netherlands 1-1 WIN
01 Jul Ivory Coast v Norway Ivory Coast 1-2 LOSS
01 Jul France v Sweden France 3-0 WIN
01 Jul Mexico v Ecuador Draw 2-0 LOSS
01 Jul England v Congo DR England 2-1 WIN
01 Jul Belgium v Senegal Draw 3-2 LOSS
02 Jul Spain v Austria Spain 3-0 WIN
02 Jul Portugal v Croatia Portugal 2-1 WIN
02 Jul Switzerland v Algeria Switzerland 2-0 WIN
02 Jul USA v Bosnia No call logged 2-0 NO CALL
04 Jul Australia v Egypt Australia · EDGE 57% LIVE 1-1 PENDING
04 Jul Argentina v Cape Verde Argentina · PREMIUM 88% PENDING
04 Jul Colombia v Ghana Colombia · EDGE 58% PENDING
Development Note · Beyond the Raw Record

Reading the numbers

The table above is the raw, full-inclusion truth: every match-result call, counted, with the draw-module handicap left in. Below is what the record shows once examined more closely. Each figure is reported with its caveat — this is a development log of a system still being refined, not a marketing sheet.

Decisive matches — pure winner accuracy 86.8%

Remove drawn matches entirely (the engine's known weak point) and look only at fixtures a team won where the engine picked a team: 33 of 38 correct.

Caveat: this is a conditional figure — it excludes draws by design, so it is an upper bound on winner-discrimination skill, not an all-in accuracy claim. Quoted only alongside the draw context. 95% CI [73%, 94%].

Versus the market — is the engine ahead? +4.8% ROI

Backing every engine call flat-stake at the market's own prices returns a positive ROI — the technical definition of beating the market, which a market-following model cannot do. The standout contrarian wins: Australia @ 6.00, USA @ 2.00, Mexico @ 2.10, Norway @ 2.30, and the Egypt–Iran draw @ 2.70.

Caveat: a thin edge on n=48 — directionally the thing that matters commercially (beating the closing line is the real moat), consistent with genuine edge, but not yet proven at statistical significance.

Post-draw-module — the current engine 67.9%

From 20 June, once draw-handling was integrated, the all-in figure (draws included) sits at 67.9% — above the 55–60% band that defines elite pre-game soccer prediction. The tournament-wide 60.4% understates the engine as it now runs.

The same engine, across every code

The World Cup is one proving ground. The same architecture has been ported to every sport ClutchTip covers, reaching competitive accuracy each time:

NRL
#1
Leading a 65-person tipping competition, 8 weeks running
Super Rugby
78%+
Full season finished above 78% accuracy
Eight-Sport Gauntlet
73%
Average across 8 simultaneous sports — including before the engine was taught to read MLB & WNBA data
⚙ The Underlying Question

Is Resonance even possible?

That was always the real thesis — whether a unique AI model composition, running on consumer hardware, could find the natural frequency of a domain and predict it. On the evidence to date: not only is the architecture possible — it is already reaching the accuracy band of the best commercial engines, at a fraction of their resource and infrastructure. The established models (XGBoost, the large ML systems) were refined over years with billions in research behind them. This is one developer, on a phone-accessible engine, with a fortnight-handicapped draw module still in beta. If the architecture can be refined to that same level of maturity, the question may no longer be does it work — but what can't it do.

Methodology & Notice

Match-result calls graded against verified final scores from multiple authoritative sources. Predictions are pre-game only and do not account for in-game variables. ClutchTip is a prediction and analytics service, not a gambling operator. This record reflects group-stage fixtures; the draw rate in group play is higher than in knockout rounds.