OK, I’ve changed the method for how I generate my picks. First off, I’ve switched to an AI model that supports Chain of Thought reasoning (step-by-step thinking), which makes it strong on tricky math problems and logic puzzles.
I buy mathematical stats for the matches that include, among other things, ELO, Poisson Strength, and a model that calculates outcome probabilities based on Elo and Poisson Strength for 1X2, BTTS, Goals and the Asian Handicap market.
Matches with an Elo difference over 120 are skipped, as are matches where the odds are below 2.00 (+100).
Then I have the AI model simulate the match and produce an EV%. The EV% needs to be above 5% but below 20%.
If it’s above 20%, it’s news the data doesn’t take into account.
Last of all, a live search is run to compare news explaining market pricing.
</
To remove first post, remove entire topic.
OK, I’ve changed the method for how I generate my picks. First off, I’ve switched to an AI model that supports Chain of Thought reasoning (step-by-step thinking), which makes it strong on tricky math problems and logic puzzles.
I buy mathematical stats for the matches that include, among other things, ELO, Poisson Strength, and a model that calculates outcome probabilities based on Elo and Poisson Strength for 1X2, BTTS, Goals and the Asian Handicap market.
Matches with an Elo difference over 120 are skipped, as are matches where the odds are below 2.00 (+100).
Then I have the AI model simulate the match and produce an EV%. The EV% needs to be above 5% but below 20%.
If it’s above 20%, it’s news the data doesn’t take into account.
Last of all, a live search is run to compare news explaining market pricing.
Match: Gaziantep vs Kasimpasa League: Turkish Superlig Kickoff: 2026-11-06 18:00 (Europe/Stockholm) Pick: Kasimpasa DNB Odds: 2.05 (Betsson) Reasoning by the modell:
Gaziantep lineup status (Oct 30 – Gaziantep Haberleri): First-choice goalkeeper Mertan Caner Özcan has an ACL injury and will be out for the entire season. This is likely reflected in the model ratings as it's a long-term absence.
Kasimpasa squad rotation (Nov 2 – SporX News): Kasimpasa’s playmaker Haris Hajradinovic trained lightly ahead of this fixture after reported fatigue symptoms, raising concerns about his involvement in the starting XI. This is short-term news not yet incorporated into the model.
Game-state influence (Oct 28 – Sampiy10 Sports): Gaziantep needs a win to escape the relegation zone. Kasimpasa, sitting safely in mid-table, can experiment in the middle third without pressure. Rumored to rest key starters before bigger fixtures in the next two weeks. Confirmed rotation risks via the Kasimpasa manager's statement
Bookmaker Overround (1X2 market):
Home: 1/2.48 = 0.403
Draw: 1/3.25 = 0.308
Away: 1/2.72 = 0.368
Overround = (0.403 + 0.308 + 0.368) - 1 = 7.9%.
Devigged Market Probabilities:
Home: 37.39% (vs. Model: 30.0%)
Draw: 28.53% (vs. Model: 27.7%)
Away: 34.08% (vs. Model: 42.2%).
Does news explain the price? Market expectations down on Kasimpasa’s win (34.08% vs Model's 42.2%), which reflects the playmaker Hajradinovic’s fatigue and Gaziantep’s need for survival points. This news slightly validates the pricing gap.
Reinforces the edge? Model suggests Kasimpasa has a 42.2% win chance with Gaziantep’s poor attacking rating. The market underrating of Kasimpasa’s win probability appears over-reactionary to short-term fatigue news—model’s probabilities aren’t stale.
The most probable model-predicted scorelines—1-1 (13.0%), 0-1 (12.4%), and 1-0 (10.0%)—reflect a low-scoring match consistent with Gaziantep’s weak attack and Kasimpasa’s strong defence. Despite motivation from Gaziantep to avoid relegation, Kasimpasa holds defensive superiority and goal-scoring efficiency according to the model.
Gaziantep's pressing need to win to escape relegation introduces a tactical urgency that may undermine Kasimpasa's defensive strategy. Kasimpasa's fatigue risk (especially Hajradinovic’s absence) could limit counter-attacking, leading to an aggressive low-scoring struggle and a possible draw instead of an away win.
2026-11-06
Match: Gaziantep vs Kasimpasa League: Turkish Superlig Kickoff: 2026-11-06 18:00 (Europe/Stockholm) Pick: Kasimpasa DNB Odds: 2.05 (Betsson) Reasoning by the modell:
Gaziantep lineup status (Oct 30 – Gaziantep Haberleri): First-choice goalkeeper Mertan Caner Özcan has an ACL injury and will be out for the entire season. This is likely reflected in the model ratings as it's a long-term absence.
Kasimpasa squad rotation (Nov 2 – SporX News): Kasimpasa’s playmaker Haris Hajradinovic trained lightly ahead of this fixture after reported fatigue symptoms, raising concerns about his involvement in the starting XI. This is short-term news not yet incorporated into the model.
Game-state influence (Oct 28 – Sampiy10 Sports): Gaziantep needs a win to escape the relegation zone. Kasimpasa, sitting safely in mid-table, can experiment in the middle third without pressure. Rumored to rest key starters before bigger fixtures in the next two weeks. Confirmed rotation risks via the Kasimpasa manager's statement
Bookmaker Overround (1X2 market):
Home: 1/2.48 = 0.403
Draw: 1/3.25 = 0.308
Away: 1/2.72 = 0.368
Overround = (0.403 + 0.308 + 0.368) - 1 = 7.9%.
Devigged Market Probabilities:
Home: 37.39% (vs. Model: 30.0%)
Draw: 28.53% (vs. Model: 27.7%)
Away: 34.08% (vs. Model: 42.2%).
Does news explain the price? Market expectations down on Kasimpasa’s win (34.08% vs Model's 42.2%), which reflects the playmaker Hajradinovic’s fatigue and Gaziantep’s need for survival points. This news slightly validates the pricing gap.
Reinforces the edge? Model suggests Kasimpasa has a 42.2% win chance with Gaziantep’s poor attacking rating. The market underrating of Kasimpasa’s win probability appears over-reactionary to short-term fatigue news—model’s probabilities aren’t stale.
The most probable model-predicted scorelines—1-1 (13.0%), 0-1 (12.4%), and 1-0 (10.0%)—reflect a low-scoring match consistent with Gaziantep’s weak attack and Kasimpasa’s strong defence. Despite motivation from Gaziantep to avoid relegation, Kasimpasa holds defensive superiority and goal-scoring efficiency according to the model.
Gaziantep's pressing need to win to escape relegation introduces a tactical urgency that may undermine Kasimpasa's defensive strategy. Kasimpasa's fatigue risk (especially Hajradinovic’s absence) could limit counter-attacking, leading to an aggressive low-scoring struggle and a possible draw instead of an away win.
Match: Venezia vs Udinese League: Serie A Kickoff: Kickoff: 2026-11-06 20:45 (Europe/Stockholm) Pick: Udinese DNB Odds: 2.35 (Coolbet) Reasoning: Udinese's new manager, Zanetti knows this stadium, knows the crowd, knows exactly what Venezia are going to throw at him. With Davis and Zaniolo up top against the worst side in Serie A, the upside is obvious. At 2.35, you’re basically getting a side with far more quality against the Serie A's worst attack
Match: Venezia vs Udinese League: Serie A Kickoff: Kickoff: 2026-11-06 20:45 (Europe/Stockholm) Pick: Udinese DNB Odds: 2.35 (Coolbet) Reasoning: Udinese's new manager, Zanetti knows this stadium, knows the crowd, knows exactly what Venezia are going to throw at him. With Davis and Zaniolo up top against the worst side in Serie A, the upside is obvious. At 2.35, you’re basically getting a side with far more quality against the Serie A's worst attack
Match: Sevilla vs Alaves League: La Liga Kickoff: 2026-11-06 21:00 (Europe/Stockholm) Pick: Alaves DNB Odds: 3.15 (Coolbet) Reasoning: Sevilla are heavily over-priced at 1.93 relative to their actual league standing (6th, but with 19 goals conceded and a negative goal difference).Sevilla are missing starting midfielder Manu Bueno through suspension, playmakers Peque Fernández and Miguel Sierra to injury, and their most dangerous forward Isaac Romero to a knee contusion. Alavés have their complete first-choice forward line intact with Lucas Boyé (5 goals) and Toni Martínez.Furthermore, Alavés have taken points in 3 of their last 6 meetings against Sevilla.
Match: Sevilla vs Alaves League: La Liga Kickoff: 2026-11-06 21:00 (Europe/Stockholm) Pick: Alaves DNB Odds: 3.15 (Coolbet) Reasoning: Sevilla are heavily over-priced at 1.93 relative to their actual league standing (6th, but with 19 goals conceded and a negative goal difference).Sevilla are missing starting midfielder Manu Bueno through suspension, playmakers Peque Fernández and Miguel Sierra to injury, and their most dangerous forward Isaac Romero to a knee contusion. Alavés have their complete first-choice forward line intact with Lucas Boyé (5 goals) and Toni Martínez.Furthermore, Alavés have taken points in 3 of their last 6 meetings against Sevilla.
Are you using AI for validating the prediction or is it the one running Poisson model and then validating the output against online news?
I use AI as my assistant. I take screenshot of the odds and let AI format them into tables, I let AI compare the current market odds vs expected odds based on ELO difference and POISON Strength. I let AI try to guess most likely score lines based on model, odds and squad. I let it check injuries, weather report, head to head, fatigue etc etc etc.
Quote Originally Posted by ShittingMyPants:
Are you using AI for validating the prediction or is it the one running Poisson model and then validating the output against online news?
I use AI as my assistant. I take screenshot of the odds and let AI format them into tables, I let AI compare the current market odds vs expected odds based on ELO difference and POISON Strength. I let AI try to guess most likely score lines based on model, odds and squad. I let it check injuries, weather report, head to head, fatigue etc etc etc.
i understand how AI tries to predict the output, but your model already does that. Is AI then just validating the information and discarding it if it's wrong then? If your model is making sharp picks then it's already AI (albeit not an LLM which came to be considered the AI). I think LLMs are just used for validating models prediction based on gossip and news but it's lagged behind bookmakers for sure. Waiting for prediction results to come out. That's the only metric that counts.
@Da_BetWhiz
i understand how AI tries to predict the output, but your model already does that. Is AI then just validating the information and discarding it if it's wrong then? If your model is making sharp picks then it's already AI (albeit not an LLM which came to be considered the AI). I think LLMs are just used for validating models prediction based on gossip and news but it's lagged behind bookmakers for sure. Waiting for prediction results to come out. That's the only metric that counts.
Hot Forum Topics
