Baseball Around the SEC - Week 6 (3/21 - 3/27/24)

ELO is 100% algorithm. If it seems to produce superior ratings to the RPI, and I agree that it does, this is because ELO uses better input data.

The RPI basically just multiplies a team’s winning percentage by the combined winning percentage of its opponents, which isn’t great at capturing opposing team power and completely ignores margin of victory.

Meanwhile, ELO focuses on margin of victory controlled for opposing team power, so it is concerned not just with wins and losses, but also the quality of the performances. E.g. If Florida beats a bad team by one run, it’s RPI might go up, but it’s ELO will almost certainly go down, as a team of Florida’s power is expected to beat a bad team by more than one run. ELO is certainly flawed, but I agree with you that overall it is close to as good as we can get. Probably the only consistently better rating would be the result of prediction market based on the activity of a large community of real sports bettors.

 

Ive never seen it broke down like i have RPI. good info.

But if margin of victory included I may change my tune. IMo 1-0 is the perfect baseball game.. but I also think 3/7-0 is the perfect football game. Nothing more dominant than scoring once, saying thats all we need and sit on it. Requires elite D/pitching, which Im fond of. Again JMO lol

In the DSR break down. the Win expectancy category??? Expected by who?? Once season in swing I guess expected by computer, ok fine. But LSU vs lets say Famu game 1 of year. Neutral site. . Whos expected to win and why? if say lsu just bc its lsu,, thats human bias im talking about and dont like.

 

Media influence is out of hand at least for me. I liked it much better when all you basically heard from the media was an occasional “Whoa Nelly” as they viewed, called, and enjoyed the outcome of the game.

 

I know it's early and it doesn't really matter, but how did TAMU jump up in the rankings while UF stayed in place? Is it just their record - even though 2 of their 3 losses are to the Gators?

 

I think I get what you are saying about a 1-0 baseball. Indeed, that is a beautiful game. Still I think we need to consider why a 1-0 game ends up with so few runs. It is undoubtably because of great defense, rather than a lack of intention on offense. If the Gators are up 1-0 on MSU in the 5th, I think there is no way that Sully tells Shelton, “Hey we already have a beautiful 1-0 lead, so no need to try to get any more.” Instead, Sully is going to try very hard to manufacture some insurance runs. Therefore, as beautiful as a 1-0 win is, this result also indicates a failure on the part of the Gators to achieve their offensive goals in most innings. That’s information, and any algorithm that ignores that information will suffer a decrease in predicative power.

How do algorithms work in game 1 of the season is a good question. Traditionally, they didn’t. Programmers usually waited until several games had been completed before releasing their algorithms. Today, some programmers actually do use some external data to start their algorithms, and these external inputs are eventually dropped as the game data accrues. This external data might include things like last year’s rating, percentage of returning offense, rating of new players etc. As a result, the early ELO are indeed less robust than those that have trained on the outcomes of 20+ games’ worth of data.

 

Agree with that mostly, but like I said above you are penalizing, or rewarding, styles of play, especially so in baseball. Obviously no one will try to NOT score.

Would you rather be a team w/ great pitching where you know you only need a few runs to win(Ark).. Or a team that needs to score 10+ to win(UGa)? The later will probably end up w/ a higher margin of victory per win,, but a no brainer IMO Id rather be the pitching team.

Obviously your elite teams score a lot w/ great pitching.

 

Ah I think I understand your point better. We can all probably agree that beating a team 11-0 is a better outcome than beating that team 1-0. But I think you are wondering if there shouldn’t be a difference between beating a team 4-1 and beating that team 14-11. The first outcome seems better, even though they share the same margin of victory.

At this point, I should confess that I have been using the term margin of victory loosely. Indeed, a good algorithm will identify a team that wins by an average of 4-1 as a stronger squad than one that wins by an average of 14-11 over the same competition. Scoring ratio is actually a better predictor of outcomes than pure scoring margin. Winning 4-1 means you scored 4x the number of runs as your opponent. Therefore, a team that wins 4-1 would be equal to its opponent’s score, even their opponent’s score were doubled and their own average scoring was halved. Not so for a team that wins 14-11. Instead, they would lose such a contest 7-22.

And to be clear, scoring ratio isn’t exactly what a good ELO is training on either, as baseball teams often hold each other runless and we aren’t permitted to divide by zero. Every algorithm is going to differ slightly, and I understand all the details of none of them, but I do still want to make clear that most programmers today do appreciate your point regarding the value of low scoring victories. Probably it’s more impressive to win 10-4 than 1-0, but 1-0 is certainly more dominant than 11-10.

 

@GatorRade—

I took two things from your post to be both true and pure—

“I understand all of the details of none of them”, and

“1-0 is certainly more dominant than 11-10”…….

Regarding baseball—the other sports have to fend for themselves.

 

Human I put applies to more than rankings. Humans run pro and amateur ball and teams no longer are built wit 1-0 games in mind. I'm a Bob Gibson fan- also a Walter Johnson fan, he holds the record for being the loser in 1-0 games- but players are bigger, stronger and more powerful. No players are nicknamed Peewee anymore. Offensive games sell more tix, which is why people own teams.

 

Was strictly speaking on the input on rankings/selections, which we dont use in pro ball. But agree its a different game now. Still think a win is a win, whether 1-0 or 15-0, shouldnt make a bit of difference.

My philosophy of it should be difficult to score(baseball/fball) will never change though. No matter how much all the rule changes are targeted to more scoring.

 

How sad—we are paired with Mizzou as the only losers in the conference on Tuesday.

 

I have a 6-0 mid week record Im looking to trade for a 3-3 conf record... bargain I tell ya..

 

I get it that weekends are more important and should be—but our bunch just seems to have a disconnect when it comes to midweek games.

Everybody’s going to drop some games they feel like they should’ve won—but you want your level of play to be consistent and competitive.

 

Yep. if struggle or average on weekends, midweeks become ultra important.

 

UK might have been for real after all. Now 15-1 in conference.

 

yep i bought in after the au sries lol.. bats much better than i thought, even vs bad pitching. but bad pitching is the norm most everywhere now I guess.