MGCOLBY's Picker Thread 2024

mgcolby

Go Eat a satchel of Richards!
Batch Image Batch Image Batch Image
As some of you know I have had a score predicting system for the NFL that I created back in 2012 or 2013. During some seasons I have posted it on patsfans, typically in the old gambling thread. It has always been fun to track how it performed during the several years that I actually managed to keep up with it weekly, it has won both against the spread (ATS) and straight up (SU) in each of those seasons with the exception of last year. I am creating a thread here on our new internet home.

Each week I will post its predictions by Tuesday NLT Wednesday afternoon and the previous weeks results. Sometimes, I may add content in regard to the predictions that I find most interesting or even shocking at times.

For those of you that have followed this in the past, you will see some minor aesthetic changes, such as blanking out the predicted "losing" team point differential and using brighter fonts. But, because, I have spent so much time on the background functions that make this churn with power query to automate the data inputs eliminating any potential errors from a manual process (which were many over the years) and eliminating what used to be an hour or better process into about 30 seconds, I felt the need to tell you this. It helps my sanity (ego)!

For the virgins! Read the shit below.

There are several columns, most I think are self-explanatory but just in case:

Predicted Score: Is the adjusted score to account for HFA based on the teams current four game average head-to-head.
Predicted Spread: Is the difference +/- for each team based on predicted score head-to-head. Hence my statement above, I removed the "losing" teams spread to declutter the visual.
Live Spread: Is based on the line at the time of me completing the weekly spreadsheet.
Spread Dif: Is the difference between the live line and the predicted spread used in this instance to show the line advantage the picker believes is in that team's favor.
O/U Line: Is based on the over/under line at the time of me completing the weekly spreadsheet.
Predicted Points: Is the combined adjusted score points in the predicted score column.
O/U DIF: Is the difference between the live over/under line and the predicted Points used in this instance to show the over/under line advantage the picker believes is in that directions favor.

Now with all that out of the way a few small house keeping things:

1. You probably noticed I said based on the last four games played. Well of course there have only been 2 games played by each team so far. I do this every year I get antsy and want to use it. So, I throw it together to early, but its pre-season for the picker. I never include the weeks prior to week 5 in the season record. So, feel free to disregard predictions prior to week 5.
2. If you do use this (and this goes for any week, but especially weeks prior to week 5) as a tool for any degeneracy, do not come back here whining and crying about the results. By that I mean being angry, upset because you followed this blindly etc... remember it doesn't account for a monsoon in Miami, injuries sustained during the game etc. Don't be a tool and be smart, use it as a tool. With that said constructive criticism, enthusiasm and ass kissing the creator for your triumphant victories (should they so happen) is always welcomed.
3. I always enjoy other people's thoughts and insights on how they use the predictor and general overall conversation about its predictions.

I am too lazy and its too early to give any insights on its predictions, only two games have been played. But fuck it lets roll! Remember don't be a fucking sally about it or I will find a way to make your beeper explode! 2024 Week 3.jpeg
 
Last edited:
Playoff Record:
ATS: 10-3 77%
SU: 10-3 77%
O/U: 5-8 38%

Season Record (Including Playoffs):
ATS: 105-107-8 50% (49.5%)
SU: 123-97 56%
O/U: 89-127-4 41%
 
Thanks for sharing. What software did you use? If you don't mind me asking.
 
I have to note that the formula I had in this version had an error. It was not accounting for HFA.

I'm glad I found it as I was continuing a deep dive through the workbook. How this happened you might ask. Simple, last season I was trying to figure out why it was performing so bad so I started messing with the formula.

At first I didn't believe it was due to the QB injuries across the league. I figured the 4 game average would account for that. My conclusion was the formula was fine, it was the backup QBs not playing consistently from week to week not to mention how many teams played 3rd string QBs and in some cases a dude right off the street.

Anyway I am not going to put a corrected version here as I believe the only game it would flip would be the KC vs Atlanta game. And because this doesn't count for season records. The formula has been corrected and will be accurate moving forward.

Just a public service announcement along with maintaining transparency.
 
Last edited:
I have to note that the formula I had in this version had an error. It was not accounting for HFA. HFA is 2 points, the scores should reflect 1 additional points for the home team and 1 less points for the away team.

I'm glad I found it as I was continuing a deep dive through the workbook. How this happened you might ask. Simple, last season I was trying to figure out why it was performing so bad so I started messing with the formula.

At first I didn't believe it was due to the QB injuries across the league. I figured the 4 game average would account for that. My conclusion was the formula was fine, it was the backup QBs not playing consistently from week to week not to mention how many teams played 3rd string QBs and in some cases a dude right off the street.

Anyway I am not going to put a corrected version here as I believe the only game it would flip would be the KC vs Atlanta game. And because this doesn't count for season records. The formula has been corrected and will be accurate moving forward.

Just a public service announcement along with maintaining transparency.

I'll let it go this week. Don't let it happen again.
 
Terrible. But to be fair two games is not enough data. I will have to tally it up.
 
2-12 ATS
4-10 SU
8-6 O/U

Updated Tally after the Monday night games.
4-12 ATS
5-11 SU
9-7 O/U

Tired Over It GIF by Acorn TV
 
Last edited:
We all got massacred
It was a drive by fucking. Crazy results yesterday. I mean Denver beating down Tampa. The Rams coming back to beat SF, Houston and Tennessee not even getting off the bus. Fuck even the Giants got in on the ass fucking. New Orleans taking a shit on Bourbon St. And Andy Fucking Dalton thinking it's 2013.
 
Week 3 Results (Not that they count)

The reason I settled on four games I think is illustrated in the Bears game. Week 2 Chicago scored some points on Offense (13), but in week 1 the defense and ST's scored a TD skewing the stats. Thats clearly not going to happen every week, but that should even out over the course of four games. Its enough to typically account for outliers in turn overs, points scored off turnovers, STs and adjustment for long term key injuries (in some situations).

2024 Week 3 Final.jpg
 
Week 4 Predictions

This is the week I usually break it out as each team has three full games and its fun to have give it a pre-season go to shake the rust off. This year went with two weeks and obviously it did not go well. Hopefully it gets its mojo back this week. I like the idea of creating an account an giving it dinglebucks. Hopefully I can keep up with making its bets each week. In order for this to work, I will have to bet an even amount on every game every week. I will open the account this week and start betting the games in week 5.

I was looking to brighten it up and maybe make it easier to read. So you will notice a complete different color scheme. Let me know what you think. I think its a bit cleaner and easier to read.

It likes the dogs this week, only laying the chalk in 4 games all on home teams. Also note the spread diff in the Jax/Hou game its only .1, I traditionally have chalked up games less than a half point in spread dif as a push regardless of win or loss given the tightness of the game. But open to feedback on this policy.

Enjoy.

2024 Week 4.jpg
 
So I have been heads down in the shack working on a how I can tweak the picker. I added a new stat to the formula now giving it a fourth dataset. I added a fifth dataset red zone percentage, but it skewered the scores out of whack. So I ditched that for maybe a future wrinkle when I figure out how to weight it correctly.

For now just sticking with the four data sets. However the way I add the points for HFA in the original model had me make two new models to test out. One applies the new dataset directly to the predicted score I post here which has the points HFA baked in.

I am going to run these side by side all season to see which one does best. I'll let you know the results.
 
Last edited:
So I have been heads down in the shack working on a how I can tweak the picker. I added a new stat to the formula now giving it a fourth dataset. I added a fifth dataset red zone percentage, but it skewered the scores out of whack. So I ditched that for maybe a future wrinkle when I figure out how to weight it correctly.

For now just sticking with the four data sets. However the way I add the points for HFA in the original model had me make two new models to test out. One applies the new dataset directly to the predicted score I post here which has the +2 points HFA baked in. And the second model has the new dataset applied directly to the two raw scores the original model produces and then adding the point the home team and taking away a point from the away team creating two new adjusted scores that are then averaged out to get the new predicted score.

I am going to run these side by side all season to see which one does best. I'll let you know the results.
I’ll bet off of your Week 5 predictions. Can’t be much worse than how I’m doing right now.
 
Week 4 Results:
Remember I don't count these in the season record. The official results start this week now that it has four full games to work with.

ATS: 7-9
SU: 8-8
O/U: 8-8

2024 Week 4 Results.jpg
 
Week 5 Predictions

Its finally go time! The picker has shit out all the rust and is ready to regain its place a top the all see'er of the future, prognosticators of all prognogsticators....get the fuck out of here Phil...there's no room for rodents on this mountain!

Well, it improved from a dismal 4-12 in week 3 to a more respectable 7-9 ATS in week 4. Hopefully we'll see the same steady improvement in week 5, cause the shit just got real. I will make a separate post in reference to my new test models and if there is any interest me making those public.

One house keeping issue. I have not done this in previous years, but I removed the HFA math from the international game. I don't believe traveling 5-10 hours across the pond to play in front of a stadium full of drunken soccer fans wearing team gear for all teams and maybe even a New England 2007 Super Bowl champions hat and T-shirt, constitutes an advantage for any team much less home field. Interested in your opinions on this.

Also, not sure why the website I pull the schedule from had Raiders game as a 1pm game. I updated the time to 4pm and left it where it was so I didn't have to rearrange the games.

With out further ado...lets get to it!

Shockers
Er...don't really see any shockers or surprises based on the totality of the results of the first four weeks. Maybe the Saints over the Chiefs, Bengals over the Ravens? Meh

2024 Week 5.jpg
 
Last edited:
The new model(s), I did not go back and run it against previous data because there would be inaccuracies that would take a lot of manual work or some hellacious excel build out to correct the data, that I don't feel like doing. It did show some promise albeit only after one week vs the orginal picker handicapped by only having 3 weeks of data. I will post last week's record at the end of the post.

I would like to explain the new model a little further than I did above. I won't get into the new data set that I am using, but I will say the plan is to use the new data set's season average throughout the year which will obviously matchup with the current datasets this week. But moving forward it will differ from the current data set four game average. Its two-fold, first I'm lazy and the way the data is pulled in it comes as the season average.

Second, while I started to formulate out a way in excel to get the four-game average, I thought why not use the season average to account for things such as a team maybe going through their tough or easy stretch of their schedule, maybe they played a string of sucky defenses or great offenses etc... After thinking about this, I felt using the season average of this data set against the four-game average of the original model could mathematically add some overall season performance against recent performance.

Question is: Do you guys want to see the weekly picks from all three along with the results or just see how the models stack up each week?

Week 4 Results

Combo Record Week 4.jpg
 
Back
Top