I already hinted at the AI Maddeness project that will build out a sports media universe around the NY Jets 2000 season as created by Madden 2001 on the PS1 (emulated). The idea is to try and understand the various AI tools out there by focusing them on an imaginary season of football using 25 year old rosters, schedules, and rules. I’m not sure it makes a lot of sense, but that’s kinda what I like about it. At a more basic level I chose Madden 2001 because it was one of the first video games that really got me wondering about the AI under the hood.
Ai Maddeness Logo Created by Tom Woodward using AI. I love it so!
This post is an attempt to capture the various details that go into determining which teams get a playoff berth. If you’re a fan of the game, you’ll know “playoff picture” is a term pundits start using to talk about various post-season scenarios as the regular season winds down. The NFL prides itself on league parity (and does a fairly decent job), so the final weeks of a season can be dominated by playing out various scenarios for those teams in the playoff hunt. It’s the time of year when the accounting for divisional records, conference records, head-to-head wins, common games win percentage, etc. starts in earnest.
Recently I’ve been playing a test season of Madden 2001 to re-familiarize myself with the game, one thing I realized too late is that Madden 2001 doesn’t share each team’s division or conference records as part of its statistics. This makes determining the playoff picture a lot harder for tight races. The “Your Self-Assurance Annoys Me” post was all about trying to get ChatGPT to read and translate screenshots of game results from the first 16 weeks of the season to give me the division and conference records for every team—that didn’t work so well.
AFC overall, conference, and divisional records going into Week 17
So, I did it the old fashioned way: I WORKED FOR IT! 🙂 And it was mindless and painful and took way too many hours, but now it’s done and I’ll not make that mistake again. Anyway, here are the overall, divisional and conference records for AFC and NFC teams. Breaking it down by conference is useful because non-conference games do not impact the playoff picture and all playoff games are intra-conference until the Super Bowl.
NFC overall, conference, and divisional records going into Week 17
The above conference standings as we move into week 17 tell us who’s in, who’s out, and who’s in the hunt. All teams in green have sealed a playoff spot. Any teams highlighted in yellow are still in the hunt, but are not guaranteed a spot. Teams highlighted in red or gray are out of the playoff picture. The difference between the red and gray teams are the red teams were on the bubble and may have had a shot if they had better conference or divisional records.
NFL 2000 Playoff Seeding Rules as told by ChatGPT
One of the things I had to figure out were the particular rules in place during the 2000-01 NFL season for the playoffs (these are subject to change) to determine tie-breakers, so I asked ChatGPT to give me playoff rules for seeding when it comes to divisional tie-breakers (potentially the case for the AFC Central and the AFC West). These checked out, and it provided a great example from the real-life 2000 season wherein the Oakland Raiders and the Tennessee Titans had the same overall record but were in different divisions, but because the Titans won more conference games they got the 1 seed.
Example from the real-life 2000 season wherein the Titans and Raiders had identical records but the Titans got top seed because they won more conference games
Not surprising given how competitive the NFL can be, by week 16 several possible scenarios emerged after figuring out all the conference and division records. I tried to break things down even more with the following charts for each conference, let’s start with the AFC:
AFC Playoff Picture breakdown heading into week 17 of the 2000-01 NFL season as played on Madden 2001 for the PS1
The Jets are free and clear at seed 1, nobody can touch them and they get a first round bye. The Browns end the season with a 9-7 overall record, which is is enough to clinch the AFC Central divisional championship win or lose along with seed 2. Even if the Titans win and the Browns lose, the Browns still have a better divisional record. The Broncos are only 8-7, but win or lose they will clinch the AFC West division and take seed 3 given the Raiders have a weaker divisional record. So seeds 1, 2, and 3 in thr AFC are already determined before week 17 is even played.
Various scenarios for seeds 4, 5, and 6 and in the AFC Playoff Picture moving into week 17
As for seeds 4, 5, and 6, there ‘s a bit more in flux. While the Colts have the second best record in the conference, they are in the AFC East with the mighty Jets at 12-3, so they’re delegated to seed 4. As for seed 5 and 6, right now that’s the Dolphins and Titans respectively. But if the Titans win and the Dolphins lose the Dolphins will go to seed 6 and the Titans take seed 5—not a huge shift to be honest. But, if the Titans and the Dolphins lose and the Steelers win the Steelers grab seed 6 and the Titans are out of the playoffs—the Dolphins remain seed 5. The reason the Steelers make the playoffs over the Titans with the same overall record is the Steelers have a better record in the division. Whereas if the Titans win and the Dolphins lose they claim a higher seed cause they have a better conference winning percentage. It’s kinda fun to get lost in these weeds.
NFC playoff picture going into week 17
The NFC has even more playoff picture drama. To start, while the Packers have the 2 seed and the Cowboys are 3, if the Packers lose and the Cowboys win then the Cowboys get seed 2 and first round playoff bye. If both the Cowboys and Packers lose then they have the save overall and conference records, so it comes down to common games, and I believe the Packers will win out there, but still need to do the math.
NFC Playoff scenarios going into week 17
As for seeds 4, 5, and 6 in the NFC, Falcons (like the Colts in the AFC) have the 4 seed locked win or lose. But for seeds 5 and 6 in the NFC its a three-way playoff between the Redskins, Eagles and Bears with all kinds of scenarios:
- If Redskins win, Eagles win, and Bears lose then Redskins are seed 5 and Eagles seed 6
- If Redskins lose, Eagles win, and Bears lose then Eagles are seed 5 and Redskins are seed 6Â (Redskins lose seed on conference games record)
- If Redskins lose, Eagles lose, and Bears win then Bears are seed 5 and Redskins are seed 6 (Redskins lose seed on conference games record)
- If Redskins win, Eagles lose and the Bears win then Redskins are seed 5 and Bears are seed 6
- Redskins lose, Eagles win, and Bears win then Eagles are seed 5 and Bears are seed 6 and Redskins are out (Redskins eliminated on conference games record)
It’s a lot! I’m a terrible logical thinker and this shit makes my head spin, and that might be part of my interest. But Madden 2001 just does all these calculations behind the scenes, and many of my scenarios hold-up, but I couldn’t help but think I was missing something, but this is an imaginary season played as a test for the real imaginary season, if you follow me.
Week 17 Games that Matter
Anyway, here are all the week 17 games that matter. The outcome of any of these games will potentially impact the playoff picture, so those are the ones I’ll be playing through over the next week or so as a way to start having AI produce quick game summaries, insights, predictions, etc.
On top of all that, this past weekend I made a 20 minute video that takes you through the playoff picture because I’ve become obsessed. I like the exercise of trying to make the various complexities understandable with some basic charts and visual summaries from Google sheets. Not sure it was all effective, or even correct, but I do know that for it to be any good it needs to be done in 1/10th the time. Thankfully this is a game I’m pretty much playing alone in more ways than one.
Given the lingering feelings I was missing some possible scenarios mentioned earlier, I ran the data I had through ChatGPT to see if it came up with similar results going into week 17. It inevitably got confused. The AFC playoff picture was pretty much a cluster, it could not deal with the AFC Central at all and insisting the Titans would be the AFC Central winners. As it so happens, this scenario is not even possible given the two teams can only have the same schedule if the Titans win given the Browns will still have a better division record (the Browns have a week 17 bye).
Corrected AFC standing in ChatGPT getting confused. Titans cannot win AFC Central (yet they do here) and the Dolphins are guaranteed a spot based on number, so the machine is confused
To be fair, the AFC playoff picture as told by ChatGPT was close at first, but it’s inability to calculate division and conference win percentage correctly started a slide it couldn’t recover from. Once I tried to correct the mistakes it just got more and more confused. On the upside, it provided me a nice chart for calculating scenarios for a potential 3-way tie in the NFC between the Redskins, Eagles and Bears. It lines up each team by wins and losses and then states who the 6th seed could be:
ChatGPT breaks down playoff scenarios for Redskins, Bears, and Eagles based on wins or losses in week 17
I had only figured 5 of these scenarios, I missed the possibilities of all 3 teams winning, all 3 losing, and if the Redskins win and the Bears and Eagles lose. There’s some basic probability statistics I’m missing that ChatGPT’s chart helped me out with. Another thing I realized while digging into these scenarios with ChatGPT is that if the Bears win and the Eagles win they have identical overall win-loss records and conference records, so it comes down to common games played (they did not have any head-to-head encounters during the season). I still need to account for that, so my playoff picture is not entirely complete and the above chart’s 6th seed assumptions is potentially wrong for at least half the scenarios. Finding holes in my predictions thanks to ChatGPT was useful, so there’s that.
What this all reinforces, at least during my early explorations, is that the more you know going into any exchange with the machine the better off you and your assumptions will be. Also, I am appreciating some of the templates for tracking these games, even if its results are nonsense. The containers for the data are more useful then what ChatGPT puts in them, and that has been true more then a couple of times now.
_____________________________________________
*You can see I’m really getting into my lost (rightfully so) career as sports broadcaster.
