MGCOLBY's Picker Thread 2024

[mgcolby]

Here is the current record going into tonight's games. I found a descrepency since I last posted it thanks to running the previous game data history against the new confidence score methadology. I had one of the days with a decent amount of games w-l cell formula starting at row 52 instead of row 2. So it ignored a bunch of w & l's. I have corrected that and here is the current (verified) record....Full disclosure and all.

​

 

[mgcolby]

Working on homing in the confidence rating system a bit further. But wanted to get the early games in.

Edit: Replaced with the full slate.

Before I get into the day's picks I want to explain what I have been doing. The prediction model I came up with is pretty basic in its formula. Which I think will tend to work ok during the regular season. However, I do not believe it will hold up well in David vs Goliath games. I felt I needed a way to incorporate SoS/Current Opponent. I have a more detailed model that I will most likely use next season. For now I have Grok running the current model against the new model and giving a confidence score based on how well they align. I also realized last night it wasn't really giving enough negative ratings for games with large gaps between talent. I wanted to find a way to accurately reflect that SIU Edward is highly unlikely to upset Houston. They call it march madness for a reason but the confidence in a low chance result should be reflectively low. So I worked on that last night and this is the result. I do want to stress that unlike the NFL picker that has over a decade of tweaking and results this is an infant still developing.

My goal is to create an NBA version next. Which given 32 or whatever teams vs 364 should be a walk in the park. Figure I'll run out the string on this one and get it to where its fully automated. I'm close but not there yet. After the NBA, I am hoping to create and College football picker based off the NFL picker. I just once again have to figure out the strength of opponent. That is the most challenging part of College athletics and what I am doing versus professional sports.

Alright enough rambling:
AI's Five Best Plays

• 1. Wofford Over 133.5 (Game 9, 18:50:00)
  Confidence: 8.1
  Reasoning: Both models predict a high-scoring game (Picker Total 153.27 vs. O/U 133.5), making the Over a strong play.
• 2. SIU Edward +28.5 (Game 4, 14:00:00)
  Confidence: 8.0
  Reasoning: Both models agree SIU Edward covers +28.5 (Picker Line +2.27, my spread +19.24), a high-confidence play despite being an underdog.
• 3. McNeese +7.5 (Game 6, 15:15:00)
  Confidence: 8.0
  Reasoning: Both models agree McNeese covers +7.5 (Picker Line -5.7, my spread +0.83), a strong play for the underdog.
• 4. Omaha +18.5 (Game 14, 21:45:00)
  Confidence: 7.5
  Reasoning: Both models agree Omaha covers +18.5 (Picker Line +0.33, my spread +11.54), a solid play despite the large spread.
• 5. NC Wilmington +14.5 (Game 16, 22:10:00)
  Confidence: 8.5
  Reasoning: Both models agree NC Wilmington covers +14.5 (Picker Line -1.53, my spread +7.28), a high-confidence underdog play.

AI's Best 3-Leg Parlay (New Model)

• Leg 1: SIU Edward +28.5 (Game 4, 14:00:00)
  Confidence: 8.0
  Odds: -110 (assumed)
• Leg 2: McNeese +7.5 (Game 6, 15:15:00)
  Confidence: 8.0
  Odds: -110 (assumed)
• Leg 3: NC Wilmington +14.5 (Game 16, 22:10:00)
  Confidence: 8.5
  Odds: -110 (assumed)
• Combined Odds: Approximately +595
  Reasoning: These are three high-confidence spread plays where both models agree the underdog covers, offering a good balance of risk and reward across different game times.

Results

 

[mgcolby]

Here is the current record. I will at some point have the win breakdown of the confidence level plays. Although I just changed them up so not sure how useful the previous ones are going forward after today. But I will work on getting that out hopefully today. AI is a blessing and a curse!

​

 

[mgcolby]

03-22-25
AI Picks of the Day:
Top 5 Plays

• North Carolina MoneyLine -125 (Game 7, 16:05:00, Confidence 8.8)
  Reasoning: Both models predict North Carolina to win (your margin 8.00, my margin 8.96), a high-confidence play within the -300 cap.
• Grand Canyon +10.5 (Game 8, 16:35:00, Confidence 8.1)
  Reasoning: Both models agree Grand Canyon covers +10.5 (your spread -0.60, my spread +5.88), a strong underdog play.
• Robert Morris +21.5 (Game 2, 12:40:00, Confidence 8.0)
  Reasoning: Both models agree Robert Morris covers +21.5 (your spread -5.73, my spread +6.33), a high-confidence underdog play (more on this below).
• Lipscomb +14.5 (Game 3, 13:30:00, Confidence 8.0)
  Reasoning: Both models agree Lipscomb covers +14.5 (your spread -2.23, my spread +6.71), a strong underdog play.
• Akron +14.5 (Game 12, 19:35:00, Confidence 8.0)
  Reasoning: Both models agree Akron covers +14.5 (your spread -5.53, my spread +3.47), another high-confidence underdog play.

3 Leg Parlay

• Robert Morris +21.5 (Game 2, 12:40:00, Confidence 8.0, Odds -110)
• Lipscomb +14.5 (Game 3, 13:30:00, Confidence 8.0, Odds -110)
• North Carolina -1.5 (Game 7, 16:05:00, Confidence 8.7, Odds -110)
• Combined Odds: Approximately +595
  Reasoning: High-confidence spread plays where both models agree, offering good value across different game times.

 

[mgcolby]

03-22-25

Posting this tonight because I am drinking. I don't want to be late tomorrow. So remember these things can change with the lines. But just remember the lines in this model are weak in terms of competition. As we go through the tournament it should right itself.

 

[mgcolby]

@Deus Irae Thanks man for supporting this!

 

[mgcolby]

@everyone else! eat cock and watch!

 

[Lord Guts]

@everyone else! eat cock and watch!

I've been busy updating the sportsbook on my free time nigga

 

[Deus Irae]

@Deus Irae Thanks man for supporting this!

It's my pleasure. You're doing a modern version of what I was doing back in the '80s, and I'm enjoying watching the evolution.

 

[CM Punk]

@everyone else! eat cock and watch!

I'm not a Eagles fan

 

[Fly Eagles Fly]

I'm not a Eagles fan

You are right you'd rather get sodomized.

 

[mgcolby]

03-23-25

Been working on this all day from an Azure automation perspective. I got to get this out of excel. Here is today's/tomorrows whatever.

Corrected Sheet fuck ups in red:


Old sheet leaving for reference:

 

[mgcolby]

There is an issue with those lines. I don't know what happened with the formatting rule. But any -differential should not be a play. Example Elon at Army The play whould Elon -6.5 not Army plus 7. Hopefully folks caught that before betting. I will correct it shortly. Sorry, trying to do to many things at once and was hoping I would get the azure automation finished yesterday, which I didn't. Which in turn had me doing the regular process at 3 in morning. Which I was clearly tired and I fucked up.

 

[mgcolby]

I had no idea there were games on today. Probably not worth running now with one game left on the schedule.. But I am getting far along with the automated process. Hoping to have it flowing for Thursday's game. Even with the scripts and macros it is a 15 step process to run every day and most of them require some level of manual intervention. I don't have the time for that. @Lord Guts I am not sure where or how you're pulling the lines for the sports book but get-odds API is a excellent and mostly free. I think the free tier gets you 500 pulls a month. But they offer odds on everything, I use it in my excel workbooks.

 

[Lord Guts]

I had no idea there were games on today. Probably not worth running now with one game left on the schedule.. But I am getting far along with the automated process. Hoping to have it flowing for Thursday's game. Even with the scripts and macros it is a 15 step process to run every day and most of them require some level of manual intervention. I don't have the time for that. @Lord Guts I am not sure where or how you're pulling the lines for the sports book but get-odds API is a excellent and mostly free. I think the free tier gets you 500 pulls a month. But they offer odds on everything, I use it in my excel workbooks.

I'm using a python script got sports odds api and having it set where we can upload it using a csv file

 

[mgcolby]

03-27-25

Missed my own deadline and this is the new model which the confidence model was based on. So Its not quite where I want it. But I am just now tweaking the SQL outputs for the reports. Its almost fully automated in Azure now. I still have to copy it into paint and make the picture to post etc...Should be finalized shortly.

Tonights picks:

 

[Deus Irae]

03-27-25

Missed my own deadline and this is the new model which the confidence model was based on. So Its not quite where I want it. But I am just now tweaking the SQL outputs for the reports. Its almost fully automated in Azure now. I still have to copy it into paint and make the picture to post etc...Should be finalized shortly.

Tonights picks:

View attachment 17457

I hope you hit at a 75% clip (I took Arkansas)

 

[mgcolby]

I hope you hit at a 75% clip (I took Arkansas)

Well it took Arkansas on the spread....

 

[Deus Irae]

Well it took Arkansas on the spread....

I was just being flip, since I'm just playing straight up in the pick'em contest (when I've actually played instead of just letting it go on its own).


And, unfortunately, "we" got "our" asses kicked by Alabama, so "I" ended up at just 2-2..

 

[mgcolby]

03/28/2025

Went old school today. Tried using the not ready for prime time system yesterday and in a rush I put the wrong totals in. Prediction scores were what they should of been etc...Hopefully I'll have this beast conquered shortly. Until then the old way: