Straight Value bets - Pre play markets in main leagues around the world

Football, Soccer - whatever you call it. It is the beautiful game.
MrJoeBlack
Posts: 90
Joined: Sun Jun 26, 2016 7:58 pm

So I ran it again with additonal prompts -

=====================================================
🎯 BETTING ANGLES β€” ABSOLUTELY NO HALLUCINATIONS
=====================================================

All analysis below is derived only from values inside your files, not from assumptions about who plays or their roles IRL.

-----------------------------------------------
🟦 ANGLE 1 β€” β€œBrighton Transition Volume > Villa Transition Volume”
-----------------------------------------------
Identity basis (team-level):

Brighton PC3 home: +0.518

Aston Villa PC3 away: –0.384

PC3 = directness / transition threat.

➑ Brighton are more direct, more vertical, and more transition-heavy.
➑ Villa away are less direct and more methodical.

Supporting player data:

Your Brighton dataset includes many players with positive PC3, e.g.:

Baleba: +0.36

(other Brighton rows include +PC3 values β€” all valid)

Your Aston Villa dataset includes many players with negative PC2 and PC1, meaning:

Weak ability to resist transitions

Weak pressing control

Bet Angle:

βœ” Brighton transition-based markets
(e.g., Brighton most transitions, Brighton more shots off transitions, Brighton counters)

βœ” Brighton team shots also increase because directness raises shot volume.

No player or event implied beyond identity values.

------------------------------------------------
🟩 ANGLE 2 β€” β€œBoth Teams to Score: YES (Identity-Driven)”
------------------------------------------------
Identity basis:

Brighton PC1 high β†’ they reach the final third consistently

Villa PC4 high (+0.300) β†’ they generate structured chances away

Brighton PC6 only +0.085 β†’ not strong in deep defensive phases

Villa PC6 +0.119 β†’ slightly better, but Brighton create many entries

Interpretation:

Brighton's identity guarantees plenty of entries.
Villa's identity guarantees few but high-quality chances.

This is purely PC-driven.

Bet Angle:

βœ” BTTS – YES

No player involvement required.

------------------------------------------------
🟨 ANGLE 3 β€” Brighton Set-Piece Advantage
------------------------------------------------
Team PC data:

Brighton PC5 home: +0.270

Villa PC5 away: +0.151

PC5 = aerial / set-piece involvement.

➑ Brighton at home have a clear set-piece superiority.

Player-level confirmation:

Brighton players with PC5 > 0:

Baleba: +1.51

Numerous others in your dataset show +PC5, reinforcing aerial activity.

Villa players with PC5 > 0:

Several (e.g., Kamara +1.68), but team-level Villa PC5 is lower than Brighton's in the away split.

Bet Angle:

βœ” Brighton over set-piece xG / Brighton to score from a set piece
βœ” Brighton corners superiority

All based purely on identity PCs.

------------------------------------------------
🟧 ANGLE 4 β€” Villa High-Quality Chances, Lower Volume
------------------------------------------------
Identity basis:

Villa PC4 away: +0.300 (strongest category for Villa)

Brighton PC6 home: +0.085 (not particularly strong)

PC4 = final third creation quality.

This suggests:
➑ Villa generate fewer chances than Brighton
BUT
➑ The chances they do generate are higher quality.

Bet Angle:

βœ” Villa Over 0.5 xG
βœ” Villa to score (team total over 0.5)
βœ” Villa fewer shots but possibly equal SOT

All of this comes directly from PC4 & PC6 interactions.

No player names needed.

------------------------------------------------
πŸŸ₯ ANGLE 5 β€” Total Goals Lean: Over 2.0 / 2.25
------------------------------------------------
Why?

Brighton PC1 (control) + PC3 (directness) = high volume

Villa PC4 (quality) = high leverage chances

Both PC6 values are below elite
β†’ Not strong at suppressing chances

Neither team presses heavily (PC2 β‰ˆ 0 for both)

This combination leads to:
➑ More chances
➑ Higher xG swings
➑ Higher game-state volatility

Bet Angle:

βœ” Over 2.0 / Over 2.25 (Asian line)

Zero hallucination β€” purely PC-based.

=====================================================
🧾 SUMMARY OF NON-HALLUCINATED BET ANGLES
=====================================================
Bet Category Identity Basis Angle
Transitions Brighton PC3 ≫ Villa PC3 Brighton transition markets
BTTS PC1 high (Brighton), PC4 high (Villa), PC6 modest BTTS Yes
Set Pieces Brighton PC5 > Villa PC5 Brighton set-piece markets
Villa Attack Villa PC4 high, Brighton PC6 average Villa over 0.5 goals
Total Goals High PC1/PC3 (Brighton), High PC4 (Villa) Over 2.0 / 2.25

All grounded in your identity model.
MrJoeBlack
Posts: 90
Joined: Sun Jun 26, 2016 7:58 pm

MrJoeBlack wrote: ↑
Wed Dec 03, 2025 4:21 pm
So I ran it again with additonal prompts -

=====================================================
🎯 BETTING ANGLES β€” ABSOLUTELY NO HALLUCINATIONS
=====================================================

All analysis below is derived only from values inside your files, not from assumptions about who plays or their roles IRL.

-----------------------------------------------
🟦 ANGLE 1 β€” β€œBrighton Transition Volume > Villa Transition Volume”
-----------------------------------------------
Identity basis (team-level):

Brighton PC3 home: +0.518

Aston Villa PC3 away: –0.384

PC3 = directness / transition threat.

➑ Brighton are more direct, more vertical, and more transition-heavy.
➑ Villa away are less direct and more methodical.

Supporting player data:

Your Brighton dataset includes many players with positive PC3, e.g.:

Baleba: +0.36

(other Brighton rows include +PC3 values β€” all valid)

Your Aston Villa dataset includes many players with negative PC2 and PC1, meaning:

Weak ability to resist transitions

Weak pressing control

Bet Angle:

βœ” Brighton transition-based markets
(e.g., Brighton most transitions, Brighton more shots off transitions, Brighton counters)

βœ” Brighton team shots also increase because directness raises shot volume.

No player or event implied beyond identity values.

------------------------------------------------
🟩 ANGLE 2 β€” β€œBoth Teams to Score: YES (Identity-Driven)”
------------------------------------------------
Identity basis:

Brighton PC1 high β†’ they reach the final third consistently

Villa PC4 high (+0.300) β†’ they generate structured chances away

Brighton PC6 only +0.085 β†’ not strong in deep defensive phases

Villa PC6 +0.119 β†’ slightly better, but Brighton create many entries

Interpretation:

Brighton's identity guarantees plenty of entries.
Villa's identity guarantees few but high-quality chances.

This is purely PC-driven.

Bet Angle:

βœ” BTTS – YES

No player involvement required.

------------------------------------------------
🟨 ANGLE 3 β€” Brighton Set-Piece Advantage
------------------------------------------------
Team PC data:

Brighton PC5 home: +0.270

Villa PC5 away: +0.151

PC5 = aerial / set-piece involvement.

➑ Brighton at home have a clear set-piece superiority.

Player-level confirmation:

Brighton players with PC5 > 0:

Baleba: +1.51

Numerous others in your dataset show +PC5, reinforcing aerial activity.

Villa players with PC5 > 0:

Several (e.g., Kamara +1.68), but team-level Villa PC5 is lower than Brighton's in the away split.

Bet Angle:

βœ” Brighton over set-piece xG / Brighton to score from a set piece
βœ” Brighton corners superiority

All based purely on identity PCs.

------------------------------------------------
🟧 ANGLE 4 β€” Villa High-Quality Chances, Lower Volume
------------------------------------------------
Identity basis:

Villa PC4 away: +0.300 (strongest category for Villa)

Brighton PC6 home: +0.085 (not particularly strong)

PC4 = final third creation quality.

This suggests:
➑ Villa generate fewer chances than Brighton
BUT
➑ The chances they do generate are higher quality.

Bet Angle:

βœ” Villa Over 0.5 xG
βœ” Villa to score (team total over 0.5)
βœ” Villa fewer shots but possibly equal SOT

All of this comes directly from PC4 & PC6 interactions.

No player names needed.

------------------------------------------------
πŸŸ₯ ANGLE 5 β€” Total Goals Lean: Over 2.0 / 2.25
------------------------------------------------
Why?

Brighton PC1 (control) + PC3 (directness) = high volume

Villa PC4 (quality) = high leverage chances

Both PC6 values are below elite
β†’ Not strong at suppressing chances

Neither team presses heavily (PC2 β‰ˆ 0 for both)

This combination leads to:
➑ More chances
➑ Higher xG swings
➑ Higher game-state volatility

Bet Angle:

βœ” Over 2.0 / Over 2.25 (Asian line)

Zero hallucination β€” purely PC-based.

=====================================================
🧾 SUMMARY OF NON-HALLUCINATED BET ANGLES
=====================================================
Bet Category Identity Basis Angle
Transitions Brighton PC3 ≫ Villa PC3 Brighton transition markets
BTTS PC1 high (Brighton), PC4 high (Villa), PC6 modest BTTS Yes
Set Pieces Brighton PC5 > Villa PC5 Brighton set-piece markets
Villa Attack Villa PC4 high, Brighton PC6 average Villa over 0.5 goals
Total Goals High PC1/PC3 (Brighton), High PC4 (Villa) Over 2.0 / 2.25

All grounded in your identity model.
Actually not a bad effort from ChatGPT - Brighton even scored from a set piece. Maybe I'll give up bothering and let AI do my insights for me.
MrJoeBlack
Posts: 90
Joined: Sun Jun 26, 2016 7:58 pm

This is the output from my probabilities model. Some potential value trading opportunities here.

--- Aston Villa vs Arsenal ---
Lambdas (expected values):
goals : home=1.098, away=0.967 (Aston Villa Handicap + 1 @2.04)
shots : home=10.055, away=12.051
sot : home=3.522, away=3.760 (BF Sportsbook 3+ Shots On Target Each Team 8/15) (BF Sportsbook Home U3.5 5/4) (BF Sportsbook Away U4.5 7/5)
corners : home=3.265, away=4.175 (BF Sportsbook Both Teams U8.5 7/5) (BF Sportsbook Home U3.5 5/4) (BF Sportsbook Away U4.5 7/5)
fouls : home=10.455, away=11.425 (BF Sportsbook Home 10 or more 2/5) (BF Sportsbook Away 11 or more 2/5)
yellows : home=2.266, away=1.989 (BF Sportsbook Both Teams O3.5 1/2) (BF Sportsbook Home O1.5 8/15) (BF Sportsbook Away U2.5 1/2) (BF Sportsbook Most Cards Tie 16/5)
reds : home=0.040, away=0.042

Match outcome probabilities (from goal model):
P(H) = 38.282% (BF Market 21.739%)
P(D) = 30.174% (BF Market 28.571%
P(A) = 31.544% (BF Market 50.761%)

BTTS probability: 41.305% (BF Market 51.546%)

Over/Under 2.5 goals:
Over 2.5: 34.089% (BF Market 45.871%)
Under 2.5: 65.911% (BF Market 54.644%)
User avatar
jamesedwards
Posts: 5051
Joined: Wed Nov 21, 2018 6:16 pm

MrJoeBlack wrote: ↑
Sat Dec 06, 2025 11:34 am
This is the output from my probabilities model. Some potential value trading opportunities here.

--- Aston Villa vs Arsenal ---
Lambdas (expected values):
goals : home=1.098, away=0.967 (Aston Villa Handicap + 1 @2.04)
shots : home=10.055, away=12.051
sot : home=3.522, away=3.760 (BF Sportsbook 3+ Shots On Target Each Team 8/15) (BF Sportsbook Home U3.5 5/4) (BF Sportsbook Away U4.5 7/5)
corners : home=3.265, away=4.175 (BF Sportsbook Both Teams U8.5 7/5) (BF Sportsbook Home U3.5 5/4) (BF Sportsbook Away U4.5 7/5)
fouls : home=10.455, away=11.425 (BF Sportsbook Home 10 or more 2/5) (BF Sportsbook Away 11 or more 2/5)
yellows : home=2.266, away=1.989 (BF Sportsbook Both Teams O3.5 1/2) (BF Sportsbook Home O1.5 8/15) (BF Sportsbook Away U2.5 1/2) (BF Sportsbook Most Cards Tie 16/5)
reds : home=0.040, away=0.042

Match outcome probabilities (from goal model):
P(H) = 38.282% (BF Market 21.739%)
P(D) = 30.174% (BF Market 28.571%
P(A) = 31.544% (BF Market 50.761%)

BTTS probability: 41.305% (BF Market 51.546%)

Over/Under 2.5 goals:
Over 2.5: 34.089% (BF Market 45.871%)
Under 2.5: 65.911% (BF Market 54.644%)
Interesting stuff. Does this take into account the opposition? eg Aston Villa when playing at home to Arsenal have expected goals of 1.098? Or just Aston Villa playing at home?
MrJoeBlack
Posts: 90
Joined: Sun Jun 26, 2016 7:58 pm

Every team’s playing style is encoded as a vector of principal components (PCs) β€” a mathematical fingerprint of how they play. Therefore the outputs are collateralised - what this means is not only is it Villa at home against Arsenal it calculates all Villa identity types vs Arsenal identity types to create a 'true' probability beyond the metadata. I believe (at the moment) this is what gives the edge.

Form this, I can predict expected event counts, and then converts those into probabilities for outcomes, correct scores, totals, BTTS, and prop markets using Poisson mathematics.
User avatar
jamesedwards
Posts: 5051
Joined: Wed Nov 21, 2018 6:16 pm

MrJoeBlack wrote: ↑
Sat Dec 06, 2025 12:02 pm
Every team’s playing style is encoded as a vector of principal components (PCs) β€” a mathematical fingerprint of how they play. Therefore the outputs are collateralised - what this means is not only is it Villa at home against Arsenal it calculates all Villa identity types vs Arsenal identity types to create a 'true' probability beyond the metadata. I believe (at the moment) this is what gives the edge.

Form this, I can predict expected event counts, and then converts those into probabilities for outcomes, correct scores, totals, BTTS, and prop markets using Poisson mathematics.
Nice! Although the delta to market prices on some would make me question the accuracy of the output. eg 38.3% vs 21.7% on Aston Villa win.
Chroma34
Posts: 26
Joined: Tue Jan 02, 2024 12:36 pm

MrJoeBlack wrote: ↑
Sat Dec 06, 2025 11:34 am
This is the output from my probabilities model. Some potential value trading opportunities here.

--- Aston Villa vs Arsenal ---
Lambdas (expected values):
goals : home=1.098, away=0.967 (Aston Villa Handicap + 1 @2.04)
shots : home=10.055, away=12.051
sot : home=3.522, away=3.760 (BF Sportsbook 3+ Shots On Target Each Team 8/15) (BF Sportsbook Home U3.5 5/4) (BF Sportsbook Away U4.5 7/5)
corners : home=3.265, away=4.175 (BF Sportsbook Both Teams U8.5 7/5) (BF Sportsbook Home U3.5 5/4) (BF Sportsbook Away U4.5 7/5)
fouls : home=10.455, away=11.425 (BF Sportsbook Home 10 or more 2/5) (BF Sportsbook Away 11 or more 2/5)
yellows : home=2.266, away=1.989 (BF Sportsbook Both Teams O3.5 1/2) (BF Sportsbook Home O1.5 8/15) (BF Sportsbook Away U2.5 1/2) (BF Sportsbook Most Cards Tie 16/5)
reds : home=0.040, away=0.042

Match outcome probabilities (from goal model):
P(H) = 38.282% (BF Market 21.739%)
P(D) = 30.174% (BF Market 28.571%
P(A) = 31.544% (BF Market 50.761%)

BTTS probability: 41.305% (BF Market 51.546%)

Over/Under 2.5 goals:
Over 2.5: 34.089% (BF Market 45.871%)
Under 2.5: 65.911% (BF Market 54.644%)
Hey mate, have you had confirmation of the accuracy of you model's outputs? Those margins are massive, aren't they?
MrJoeBlack
Posts: 90
Joined: Sun Jun 26, 2016 7:58 pm

It depends on the market as some are noisier than others e.g. red cards. BTTS is the weakest of the classification markets. O/U is strong. 1X2 is currently showing 15%ROI. Shots on Target are pretty good. Corners, shots, fouls are noisier. However, if I'm honest what I am most interested in is how close the predicted identity of the match is to the reality of it. And naturally my vision of the game is developing with each event. e.g. if I see the identity of the game being dependent on the home team dominating possession and it is evident in the first 20 minutes that they are not then my positions will evolve. If we look at the example of the Villa/Arsenal game - there was nothing really at any point that made me feel as though the 1X2 modelling was incorrect - that does not mean it wasn't, it just means that if you did not know who the two teams were and what price they were and were simply watching the game as a neutral would you think that my prices were a fair reflection of the relative supremacy? I think you'd be hard pressed to find anyone who thinks that they were any less inaccurate than the actual market - and that has nothing to do with the outcome. Hope that makes sense.
MrJoeBlack
Posts: 90
Joined: Sun Jun 26, 2016 7:58 pm

An example of how I approach the week. Liverpool are at home to Brighton on Saturday. Looking at potential team news (injuries, cards, fatigue etc) these are the first set of outputs for HOME and AWAY shots. I like to do this before the market opens. To explain - the primary output (predicted value) for each team in this case is 15.097 vs 14.634 shots. Worth noting that shots are particularly noisy so this is purely for the purpose of illustrating method. The formatting is dodgy as I don't know how to attach a screenshot. But essentially you have line/threshold/percentage chance of exceeding threshold/over odds/percentage chance of not exceeding threshold/under odds. Then, when the market opens, it's simply a case of calculating where the greatest margin exists (dependent upon liquidity). Then update on team news. Then evaluate the position throughout the match relative to match identity expectations.

Home Shots Lambda Odds Odds
Line Threshold 15.097 Over (P(X≀k-1)) Under

7.5 8 98.30% 1.02 1.70% 58.75
8.5 9 96.44% 1.04 3.56% 28.09
9.5 10 93.32% 1.07 6.68% 14.98
10.5 11 88.62% 1.13 11.38% 8.79
11.5 12 82.16% 1.22 17.84% 5.61
12.5 13 74.03% 1.35 25.97% 3.85
13.5 14 64.60% 1.55 35.40% 2.82
14.5 15 54.42% 1.84 45.58% 2.19
15.5 16 44.18% 2.26 55.82% 1.79
16.5 17 34.52% 2.90 65.48% 1.53
17.5 18 25.94% 3.85 74.06% 1.35
18.5 19 18.74% 5.33 81.26% 1.23
19.5 20 13.03% 7.68 86.97% 1.15
20.5 21 8.71% 11.48 91.29% 1.10
21.5 22 5.61% 17.84 94.39% 1.06
22.5 23 3.48% 28.77 96.52% 1.04
23.5 24 2.08% 48.11 97.92% 1.02
24.5 25 1.20% 83.38 98.80% 1.01


Away Shots Lambda Odds Odds
Line Threshold 14.634 Over (P(X≀k-1)) Under

7.5 8 97.78% 1.02 2.22% 45.06
8.5 9 95.48% 1.05 4.52% 22.12
9.5 10 91.74% 1.09 8.26% 12.10
10.5 11 86.26% 1.16 13.74% 7.28
11.5 12 78.98% 1.27 21.02% 4.76
12.5 13 70.10% 1.43 29.90% 3.34
13.5 14 60.10% 1.66 39.90% 2.51
14.5 15 49.64% 2.01 50.36% 1.99
15.5 16 39.45% 2.54 60.55% 1.65
16.5 17 30.12% 3.32 69.88% 1.43
17.5 18 22.09% 4.53 77.91% 1.28
18.5 19 15.57% 6.42 84.43% 1.18
19.5 20 10.54% 9.49 89.46% 1.12
20.5 21 6.86% 14.58 93.14% 1.07
21.5 22 4.30% 23.28 95.70% 1.04
22.5 23 2.59% 38.60 97.41% 1.03
23.5 24 1.51% 66.41 98.49% 1.02
24.5 25 0.84% 118.43 99.16% 1.01
MrJoeBlack
Posts: 90
Joined: Sun Jun 26, 2016 7:58 pm

If someone can explain how to attach a file I'll upload a nice little in-play shot tracker which models whether the target value is on track.
User avatar
jamesedwards
Posts: 5051
Joined: Wed Nov 21, 2018 6:16 pm

MrJoeBlack wrote: ↑
Wed Dec 10, 2025 3:47 pm
If someone can explain how to attach a file I'll upload a nice little in-play shot tracker which models whether the target value is on track.
Same principle as this. https://www.youtube.com/watch?v=puozFcP2xOY

You can only attach certain file types and only up to a certain file size.
MrJoeBlack
Posts: 90
Joined: Sun Jun 26, 2016 7:58 pm

jamesedwards wrote: ↑
Wed Dec 10, 2025 6:34 pm
MrJoeBlack wrote: ↑
Wed Dec 10, 2025 3:47 pm
If someone can explain how to attach a file I'll upload a nice little in-play shot tracker which models whether the target value is on track.
Same principle as this. https://www.youtube.com/watch?v=puozFcP2xOY

You can only attach certain file types and only up to a certain file size.
Thank you for pointing out that I'm an idiot! So essentially what you're saying is click attachments and attach the file! Literally, never seen the tab that is now staring me in the face! Insert monkey face plant emoji... (if only I knew how to do that too!)
Post Reply

Return to β€œFootball trading”