CROWD EXPECTATIONS + FOOTBALL HISTORY

Ask TablePredict.

What the crowd expects.
What history actually shows.

Explore supporter predictions, club season records and head-to-head results. Source-backed research tools for your MCP-compatible AI assistant.

Explore the research

Inspectable answers. No betting tips. No invented certainty.

120

Preseason table submissions

4 leagues · 2026/27 · 73 Premier League tables

4,180

Historical matches

Premier League · 11 seasons · 2014/15–2024/25

01 · FROZEN CROWD EXPECTATIONS

Where supporters agree. And don’t.

Inspect examples calculated by the research tools. No account needed.

2026/27 · FROZEN SNAPSHOTS
Premier League · 73 preseason tables
CONSENSUS FIRST PLACE

Arsenal

61.6%

45 of 73 tables put this club first. This is a sample share, not a title probability.

Widest disagreement

Newcastle9–14
Brentford9–14
Everton9–14

The range containing the middle half of predictions. Wider means more disagreement.

2026/27 · Frozen at 2026-08-21T19:00:00+00:00 · Inspect the free consensus source

Championship · 20 preseason tables
CONSENSUS FIRST PLACE

West Ham

80%

16 of 20 tables put this club first. This is a sample share, not a title probability.

Widest disagreement

Derby County7–12
Swansea City9–14
Preston North End15–20

The range containing the middle half of predictions. Wider means more disagreement.

2026/27 · Frozen at 2026-08-14T18:00:00+00:00 · Inspect the free consensus source

Serie A · 10 preseason tables
CONSENSUS FIRST PLACE

Inter Milan

90%

9 of 10 tables put this club first. This is a sample share, not a title probability.

Widest disagreement

Como6–10
Fiorentina9–12
Genoa11–14

The range containing the middle half of predictions. Wider means more disagreement.

2026/27 · Frozen at 2026-08-22T16:30:00+00:00 · Inspect the free consensus source

Bundesliga · 17 preseason tables
CONSENSUS FIRST PLACE

Bayern Munich

94.1%

16 of 17 tables put this club first. This is a sample share, not a title probability.

Widest disagreement

Hoffenheim6–10
Werder Bremen10–14
Schalke 0412–16

The range containing the middle half of predictions. Wider means more disagreement.

2026/27 · Frozen at 2026-08-28T18:30:00+00:00 · Inspect the free consensus source

120 submissions is not 120 distinct people. These are small, self-selected samples, not representative polling. A pick share is not a probability of winning. Historical matches do not increase the crowd sample.

02 · HISTORICAL RESULTS

Put the club in context.

Premier League results from OpenFootball’s CC0 archive. No odds or live feed.

Two examples from the 11-season archive. The planned MCP service will support club, season, opponent and venue filters.

Arsenal · Premier League 2024/25

MATCHES

38

W–D–L

20–14–4

MATCH POINTS

74

GOAL DIFFERENCE

+35

OpenFootball · CC0 · Historical example, not current form. Inspect the pinned archive

Liverpool · Premier League 2024/25

MATCHES

38

W–D–L

25–9–4

MATCH POINTS

84

GOAL DIFFERENCE

+45

OpenFootball · CC0 · Historical example, not current form. Inspect the pinned archive

The archive ends in 2024/25: these are not current form or 2026/27 results. Match points are calculated as 3 per win + 1 per draw, before any disciplinary deductions. They are not official league positions or future probabilities.

FOR WRITERS, CREATORS & CURIOUS FANS

From data to a story.

01

Inspect expectations

Median ranks, first-place shares and the clubs with the widest disagreement. Every figure has its sample size.

02

Research a track record

Club season summaries, home and away splits, and head-to-head results. Ask about the exact period you need.

03

Compare and explain

Bring a complete prediction or dated standings. Find gaps against the crowd, with supplied standings clearly marked unverified.

CONNECT TO YOUR OWN ASSISTANT

One season. One pass.

Nine read-only research tools. You bring the AI assistant; we provide the source-backed calculations.

The basic consensus remains free on TablePredict, and the historical results are open data. This pass is for convenient agent access and analysis workflows — not exclusive raw data.

Your AI provider may charge separately. Supports Streamable HTTP MCP clients with Bearer-key authentication; this is not a ChatGPT or Microsoft store listing.

PROPOSED LAUNCH PRICE

€19 / season pass

1,000 successful research tool calls
Access until 30 June 2027 · No renewal

Public preview only. Paid access and the remote MCP service are not available yet.

Final price, tax treatment, seller details and access policies will be confirmed before sales open. No payment can be made here.

PLANNED CONNECTION · NOT ACTIVE YET

Bring your own assistant.

Once access launches, connect a Streamable HTTP MCP client with your private Bearer key.

https://research.tablepredict.com/mcp

This address is reserved for the future service. It cannot accept research calls yet.

Know what you’re buying.

Is this an AI chat subscription?

No. This is a research tool service and an agent instruction bundle. Connect it to a compatible assistant you already use. There is no hosted model, included AI subscription or separate chat account.

What data is actually connected?

Two separate datasets: TablePredict’s frozen 2026/27 crowd aggregates (Premier League 73 tables, Championship 20, Serie A 10, Bundesliga 17), and 4,180 completed Premier League matches across 2014/15–2024/25 from OpenFootball under CC0. The historical archive includes results, goals and match-derived points, not odds. No Polymarket, bookmaker prices, xG, 2025/26 results or current standings feed is connected. The private odds experiment is not part of this service.

Does it update or predict the next match?

Both datasets are pinned snapshots. They do not update automatically. Past results are context, not forecasts; no profitable betting edge is claimed. You can supply a dated table for comparison, but the service cannot verify that it is current or official.

What counts as a request?

One successful research tool call counts as one request. A single assistant answer can use several calls. Invalid inputs, connection handshakes, prompts and methodology reads do not use the allowance. Failed network delivery after a completed calculation may still count. There are no automatic overage charges.