re: Week 8 Power Ratings and SEC slate

What is the purpose Milk ? Because if your model doesn't pick winners ?

It just doesn't fricking matter. Not in the slightest.

quote:

---

And how is this model going to make you money this weekend ?

It won't. But you cranked up a spreadsheet. Mixed up some variables. And came up with some shite on a page.

I almost feel bad about pissing on you for this. Because at least you, you know, mathed.

Jut do this. Go Recursive. That means apply your model backward. See how well your model works then.

Except you can't find a historical data set that shows college football results, performance data, and lines. Because only Vegas has that shite.

Whomever working for them Mob Guys that had the foresight to start writing that shite down ? And convince them ? Genius. Let alone "Hey Vito, we need a database".

---

Are you on drugs?

Who said it's supposed to make me money?

This was not done on a spreadsheet. It is a lot of complicated Math in a program.

Recursive does not mean backwards (wtf?). And I can't "run the model backwards", that's not how it works.

I do have historical Vegas spreads and the model does not use them. Why do you think historical spreads are available? Do you think they're under lock and key or something?

quote:

---

Whomever working for them Mob Guys that had the foresight to start writing that shite down ? And convince them ? Genius. Let alone "Hey Vito, we need a database".

---

..... what?

quote:

---

Sorry, but your model is not as good as you think it is.

It’s been dumped on for weeks on end, and you’re still on here pushing it and getting offended about negative feedback. Your model, and the collective metrics you think matter in totality, do not mean what you think they mean.

Bottom line, your model sucks and it has always sucked. Either rework it, improve it, replace it, or whatever. But at this point, you need to embrace failure, utilize negative feedback, and try again, or just quit.

But no one is buying what you’re selling.

---

I am not offended in the slightest.

My model continues to produce similar results to Vegas (which is the standard). If it had severe flaws, it would show itself against Vegas by being way off compared to Vegas spreads - and that's just not the case. I'm well aware that you people (most of you) have no idea how this stuff works and you're just butthurt that your team is "ranked" low - and that tickles me every week.

quote:

---

Anybody that could distill college football ... or basketball ... or horse racing ... or anything else to a model ? They wouldn't show you the model.

That's literally the Holy Grail of Gambling. Just crank the model, get the winners, then go out and make money.

This is the absolutely most annoying thing about the Information Era. The idea of "well, the number says". Its the new version of my parents listening to Walter Cronkite and then saying "Well, Walter said".

Its hard enough to make real models, using two or three variables, that predict real life stuff enough to be used to build real life stuff.

This "model" ? LIke all like it. Is fricking dumb.

---

Models are right on games straight up about 75% of the time. More than that is pretty good, 70-75% is about average. I've used one on Yahoo and finished in the top 99%.

In a straight pickem, they will regularly out perform most humans and finish in the top.

The reason that doesn't work to just crank the model and get the winners is because of the spread. Here's a common scenario and one for this weekend.

My model has Alabama as 16 point favorites over Texas A&M. The current spread is 14, but it opened at 15.5.

Now which way does the model say to bet on that? Do you bet on Alabama to cover? It's more attractive at 14, but it's basically just predicting the same thing as Vegas was too start.

And that's common. Especially since Vegas uses models heavily.

Kentucky I have as 15.5 point favorites against Vandy. The spread is now 10.5 for Kentucky. That's about the best difference I'm going to see. But I'm not betting money on that shite given there is a +- 11-13 margin on the spread.

As for Kentucky, I have you ranked #4. But it also predicts you will finish the season 9-3 and ranked #20. 13 point dog @ Tenn and 8 point dog @ Missouri. Some of that is home field advantage.

But the cool thing about models is they update each week. Kentucky has outperformed most of the year, so their power ratings and so forth have increased. So if you continue to exceed expectations, then those predictions will adjust.

Anyway, when it comes to the whole of college football, I'd take the model over you all day.

quote:

---

ScoggDog

---

I truly feel bad for this person and will stop poking fun.

quote:

---

Models are right on games straight up about 75% of the time. More than that is pretty good, 70-75% is about average. I've used one on Yahoo and finished in the top 99%.

---

Making any money at it ?

Because here's my model. I take Georgia and Alabama at home, unless they play each other.

Good enough 75% of the time. Who cares.

Of all the people in this thread, I think this guy is the last one you should try explaining this stuff to.

quote:

---

LSU played without Trey'Dez Green, Whit Weeks, and 2 other starters. Princewill was out over half the game as well iirc.

---

MS State was without 2 starting LB, 2 rotational DL (played significant snaps prior to injury), and Isaac Smith went out during the game on top of no Kamario Taylor and Fluff Bothwell. There's no replacing Taylor just like there'd be no replacing Leavitt, and non-qb injuries are much easier to compensate for as a team.

quote:

---

Making any money at it ?

Because here's my model. I take Georgia and Alabama at home, unless they play each other.

Good enough 75% of the time. Who cares.

---

I literally explained the rest and you purposely choose to ignore it.

So go frick yourself.

quote:

---

I literally explained the rest and you purposely choose to ignore it.

---

How much money have you made ? Because we can ALL use "a model".

Even a "just picking winners" model. Here's a new one of mine. Georgia wins. Good for around 5 of 6 outcomes.

I used to know a couple guys that though Sagarin cracked the code. They stopped a few grand under.