FootballEdgeSeason 2026/27Self-updating record
Picks · the model, facing forward · season 2026/27

The machine’s picks.
On paper, in public, losses and all.

Nothing on this page is a bet, and nothing here is advice. Every selection is chosen by a pre-registered rule from the same model read as the record — no human picks them, and no losing one is ever removed. Stakes shown are an imaginary £10. Three seasons of our own testing found no betting profit; the fair expectation for every strand on this page is a paper loss over time. That expectation, printed here, is the point.

The live feed

Every open pick above the conviction bar (a 12-point model-vs-market gap, set before the season and never moved) — the forward half of the honest record. They settle there, wins and losses alike. One pick per match: several legs on one game would be one opinion wearing several hats, not several reasons to believe it.

5 open leg(s), one per fixture. Nothing doubled up on the same match this time.

Quality is our own confidence read on each pick — the size of the edge, tempered down where four seasons of testing show that kind of edge does not turn into closing-line value (home wins and goal totals; draws and aways convert), or where our calibration is weak (BTTS). A high, B solid, C marginal. It rates the model's conviction, not how much to stake, and it is not a guarantee.

Kick-offFixtureThe pickOddsModelMarketGap (pts)Quality
Tue 03 Nov 19:45 GMTQPR v Burnley
provisional		QPR, draw no bet1.5078%62%+16.4A
Tue 03 Nov 19:45 GMTPortsmouth v Swansea
provisional		Portsmouth under 1.5 goals1.6170%58%+12.5A
Wed 04 Nov 19:45 GMTCardiff v West Ham
provisional		Cardiff, draw no bet2.5951%36%+14.6A
Wed 04 Nov 19:45 GMTWrexham v Middlesbrough
provisional		Wrexham under 0.5 goals3.0144%31%+12.8A
Wed 04 Nov 20:00 GMTBristol City v Wolves
provisional		Bristol City, draw no bet2.4154%39%+15.3A

Withdrawn at team news

Since 25 October 2026 every pick goes up provisional and only counts once the confirmed line-ups are in and the pick still clears the pick test on the latest price. Picks that fail that check are withdrawn; picks whose line-up our feed missed before kick-off are unconfirmed. Neither counts on the record or the strands, and both stay listed here, results and all.

DayWhereFixturePickOddsStatusWhyResult
2026-11-01Live feedAthletic Club v Real SociedadDNB home1.533withdrawnprice shortened below 1.50won
2026-10-31Live feedAC Milan v InterDNB away1.528withdrawnprice shortened below 1.50lost
2026-10-31Daily DrawLyon v Angers1X2 draw4.969withdrawnedge fell to +3.8 pts at team newslost
2026-10-30The Daily DoubleLille v LensDNB away2.06withdrawnedge fell to +1.7 pts at team newslost
2026-10-25Daily DrawSC Freiburg v Bayern München1X2 draw5.449withdrawnedge fell to +3.6 pts at team newslost
2026-10-25Live feedMarseille v Le HavreDC draw/away2.491withdrawndistrust guard at team newslost

Every strand, side by side

All the runs at once, dead ones included. With several strands, one is usually on a run — that is arithmetic, not skill, and it is why they are never shown apart.

StrandRunBestW-LPaper P/LCLVBeat the closeStateExit check
Pick of the Day013-9£-51.57+2.2%67% of 12run deadcollecting · 12/30
The Daily Double3422-26£-38.49-0.7%53% of 49run alive · 3collecting · 10/30
Daily BTTS0618-19£-4.42+0.8%62% of 37run deadcollecting · 10/30
Daily Draw026-13£+54.69-2.2%37% of 19run deadcollecting · 2/30
Daily Home225-6£-10.97+0.9%73% of 11run alive · 2collecting · 6/30
Scoreline of the Day023-10£+59.92-0.7%62% of 13run deadcollecting · 13/30
Daily Under (retired)135-11£-75.45-1.7%31% of 16run alive · 1outside the rule

CLV is the average gap between the price we logged and the closing price, per strand — the same credibility check as the record. Exit check is the one rule every live strand lives under, printed before any of them could be judged by it: over its last 30 settled picks logged since 17 Oct 2026 (the day the model changed), beat the closing price under 40% of the time with an average CLV below zero and it retires at the next Monday review. Re-judged every week once the 30 are in. A retired strand keeps its full record here. Win or retire — the run it is on has nothing to do with it. Click a strand for its own page: the rule in full and every pick it has made.

Pick of the Day

The model’s single strongest call of the day — the biggest model-vs-market gap above the 12-point conviction bar, priced 1.50 to 4.00. It picks first; the other strands work around it. Started 17 Oct 2026, with its retirement rule printed from day one.

Current run
0
waiting for the next win
Best run
1
longest so far this season
Record
3-9
won-lost, settled picks
Paper P/L
£-51.57
imaginary £10 a time
Open pick (2026-11-03): QPR v Burnley — QPR, draw no bet @ 1.50. Settles after full-time.
Open pick (2026-11-04): Bristol City v Wolves — Bristol City, draw no bet @ 2.41. Settles after full-time.
DayFixturePickOddsResult
2026-11-01Marseille v ToulouseToulouse, draw no bet2.73LOST
2026-10-31Stoke City v PrestonPreston, draw no bet3.32LOST
2026-10-30SV Elversberg v FSV Mainz 05FSV Mainz 05, draw no bet1.55LOST
2026-10-29Frosinone v LecceLecce, draw no bet3.29LOST
2026-10-28AC Milan v BolognaBologna or the draw2.31WON

The last 5 of 12 settled picks.

Every pick it has made, with its rule in full →

The Daily Double

One pick a day at evens-ish odds. How long can a winning run live before the maths catches it?

Current run
3
consecutive wins, alive now
Best run
4
longest so far this season
Record
22-26
won-lost, settled picks
Paper P/L
£-38.49
imaginary £10 a time
Open pick (2026-11-03): Blackburn v Stoke City — at least one side not scoring @ 2.01. Settles after full-time.
Open pick (2026-11-04): Cardiff v West Ham — Cardiff over 1.5 goals @ 2.07. Settles after full-time.
DayFixturePickOddsResult
2026-11-02Deportivo La Coruna v Osasunaboth teams scoring1.91WON
2026-11-01Lazio v Cagliariboth teams scoring1.99WON
2026-10-31Bournemouth v LeedsBournemouth over 1.5 goals2.00WON
2026-10-28Parma v UdineseParma, draw no bet2.06push
2026-10-27Torino v ComoTorino or the draw2.04LOST

The last 5 of 53 settled picks.

Every pick it has made, with its rule in full →

Daily BTTS

One both-teams-to-score call a day — yes or no, whichever side the model rates.

Current run
0
waiting for the next win
Best run
6
longest so far this season
Record
18-19
won-lost, settled picks
Paper P/L
£-4.42
imaginary £10 a time
Open pick (2026-11-03): Portsmouth v Swansea — at least one side not scoring @ 2.17. Settles after full-time.
Open pick (2026-11-04): Wrexham v Middlesbrough — at least one side not scoring @ 2.23. Settles after full-time.
DayFixturePickOddsResult
2026-11-02Sassuolo v Fiorentinaat least one side not scoring2.54LOST
2026-11-01Getafe v Sevillaat least one side not scoring1.64LOST
2026-10-31Middlesbrough v Charltonat least one side not scoring1.98WON
2026-10-30Villarreal v Espanyolat least one side not scoring2.13WON
2026-10-28Genoa v Juventusat least one side not scoring1.77WON

The last 5 of 37 settled picks.

Every pick it has made, with its rule in full →

Daily Draw

Nobody backs the draw. One a day — the model's likeliest draw that also clears the pick test, priced 3.00 or longer. Value-gated since 23 Aug 2026, so most days it shows nothing rather than a draw with no edge.

Current run
0
waiting for the next win
Best run
2
longest so far this season
Record
6-13
won-lost, settled picks
Paper P/L
£+54.69
imaginary £10 a time
No Daily Draw today — no draw cleared the pick test at 3.00 or longer on a match another strand hadn’t already taken. An empty day stays empty.
DayFixturePickOddsResult
2026-11-01Como v Veneziathe draw5.70LOST
2026-10-24Lazio v Parmathe draw4.39LOST
2026-10-17West Brom v Blackburnthe draw3.51LOST
2026-10-14Lincoln v West Hamthe draw4.05LOST
2026-10-13Burnley v Charltonthe draw3.83WON

The last 5 of 19 settled picks.

But isn’t the draw our best-read market? It is — and that is calibration, not an edge. The model judges draws more honestly than any other market (the proof is here), but the bookmakers price them just as tightly, so a draw with genuine value is rare. That is why this strand is empty most days, and any losses above were logged before the 23 Aug value-gate. Good calibration is our credibility check — it was never a profit signal. The model’s draw reads surface on the Draw Radar (reads, not bets).

Every pick it has made, with its rule in full →

Daily Home

One home win a day — the home side the model rates likeliest to win, priced 1.50 to 3.00, that also clears the pick test. Added 24 Sep 2026, with its retirement rule printed from day one.

Current run
2
consecutive wins, alive now
Best run
2
longest so far this season
Record
5-6
won-lost, settled picks
Paper P/L
£-10.97
imaginary £10 a time
Open pick (2026-11-03): Preston v Sheffield Utd — Preston winning @ 2.88. Settles after full-time.
DayFixturePickOddsResult
2026-11-01Athletic Club v Real SociedadAthletic Club winning2.11WON
2026-10-31Millwall v DerbyMillwall winning1.69WON
2026-10-27Sassuolo v LazioSassuolo winning2.88LOST
2026-10-25Athletic Club v GetafeAthletic Club winning1.63WON
2026-10-24FSV Mainz 05 v Werder BremenFSV Mainz 05 winning1.68LOST

The last 5 of 11 settled picks.

Every pick it has made, with its rule in full →

Scoreline of the Day

The scoreline the model thinks likeliest — which is still usually wrong. One a day, only when the price is no worse than the model’s own number. Expect a hit rate around one in seven and losing runs of twenty or more: that is the correct-score market, not a fault. Added 17 Oct 2026 under the universal exit rule.

Current run
0
waiting for the next win
Best run
2
longest so far this season
Record
3-10
won-lost, settled picks
Paper P/L
£+59.92
imaginary £10 a time
Open pick (2026-11-03): Watford v Lincoln — exactly 1-1 @ 6.46. Settles after full-time.
DayFixturePickOddsResult
2026-11-02Newcastle v Evertonexactly 1-16.63LOST
2026-11-01Juventus v Napoliexactly 1-16.17LOST
2026-10-31Aston Villa v Fulhamexactly 1-17.00WON
2026-10-30Sheffield Utd v Wrexhamexactly 1-16.24LOST
2026-10-29Fiorentina v Atalantaexactly 1-16.70LOST

The last 5 of 13 settled picks.

Every pick it has made, with its rule in full →

Daily Under — retired 24 Sep 2026

Retired after beating the closing price only 31% of the time. No new picks; its full record stays here, losses and all.

Current run
1
consecutive wins, alive now
Best run
3
longest so far this season
Record
5-11
won-lost, settled picks
Paper P/L
£-75.45
imaginary £10 a time
Retired 24 Sep 2026 — no new picks. Nothing hidden: every pick it made is listed below.

Every pick it has made, with its rule in full →

Selection rules, in full: the Pick of the Day takes the single biggest model-vs-market gap of the day among pick-test passers that also clear the 12-point conviction bar, priced 1.50 to 4.00; the Daily Double takes the pick-test passer priced nearest to 2.00 (between 1.80 and 2.30) for each day; Daily BTTS takes the highest-edge both-teams-to-score passer, yes or no; Daily Draw takes the model’s likeliest draw of the day that also clears the pick test (edge ≥ 4 points), priced 3.00 or longer — value-gated since 23 Aug 2026, so a sub-threshold draw no longer reaches the record and a day with no value draw shows nothing; Daily Home takes the home side the model rates likeliest to win, priced 1.50 to 3.00, that also clears the pick test; Scoreline of the Day takes each match’s likeliest scoreline on the model’s grid, keeps it only when the price is no worse than the model’s own number, and headlines the one the model is surest of — it picks last and expects to be wrong most days. One fixture per day across all six: the Pick of the Day picks first, then the Double, then BTTS, then Draw, then Home, then the Scoreline, and a match already taken is skipped by the later strands — several strands on one game would be one opinion wearing several hats. One each per day, chosen by the machine, logged before kick-off, settled in public. A day with no qualifier is left empty rather than filled. Pushes and voids neither break nor extend a run. Draw and Under were added, and the one-fixture rule applied to all four, on 15 Aug 2026 after one BTTS pick had settled — recorded in the decisions log, not hidden. The Daily Under was retired on 24 Sep 2026 after it beat the closing price only 31% of the time; the Daily Home replaced it, and carries a rule from day one: if it beats the close less than 40% of the time with an average CLV below zero after 30 settled picks, it retires too. The Under’s full record stays below. The Pick of the Day was added on 17 Oct 2026 and picks ahead of the Double, so from that date the Double sometimes loses its first-choice match and takes its next-best candidate instead — its record before and after that date is not like-for-like, and we say so here. Its exit rule, printed before its first pick: if it beats the close less than 40% of the time with an average CLV below zero after 30 settled picks, it retires. From 17 Oct 2026 every live strand carries the same exit rule (the scoreboard above shows where each one stands), and the Scoreline of the Day was added the same day, with the strand cap lifted on Karn’s call — six live strands means one is nearly always on a run, which is why they are only ever shown side by side.