Note at the bottom of the top 25 - if a team has not played 2 in state opponents they are not in the rankings. Russell is on the list, along with Trinity, Pulaski County, Johnson Central, et. al.

 

I looked twice...which means I missed them twice. My bad. But they are still terrible.

No one takes him seriously in central ky either. Except for the 4-5 lex schools he's always pulls for. He's horrible I totally agree with you coyote!!

No one takes him seriously in central ky either. Except for the 4-5 lex schools he's always pulls for. He's horrible I totally agree with you coyote!!

 

Computers don't pull for anyone. The way he chooses to value the data may be questioned but computer programs don't pull for people. Why is it so hard for people to understand that?

I think Litkenhouse does it better. I think Massey and Calpreps does it even better than that. Computers have a rating system and are not the same as an opinion poll.

I think Litkenhouse does it better. I think Massey and Calpreps does it even better than that. Computers have a rating system and are not the same as an opinion poll.

 

Like I said on another thread, CalPreps and MaxPreps are better NOW, because they have more data NOW (out of state games).

 

Lit and Cantrall will catch up near the end of the regular season and I'm fairly confident all four (Cantrall, Lit, Max, Cal) will be fairly similar in their playoff projections.

 

But like you, I laugh when posters talk about leaving out Johnson Central (not enough games against Kentucky teams, which are the ONLY teams Cantrall take into consideration) or having bias/favorites/teams they're partial to (unless teams start off with a certain ranking instead of zero/being equal, that's not very likely).

Computers don't pull for anyone. The way he chooses to value the data may be questioned but computer programs don't pull for people. Why is it so hard for people to understand that?

 

Who programs those computers? When a season starts a coefficient must be placed on every team. Hence why does the computer rate a 47-0 win vs. Phelps lower than a 27-28 loss to a say a Louisville Trinity team. If there was no human input into the algorithm then every team in the top 25 should be undefeated. Human input does matter in computer rankings, especially at the start of a season. The computer generally works out the biaseness toward the end of the season, but it is still there.

Who programs those computers? When a season starts a coefficient must be placed on every team. Hence why does the computer rate a 47-0 win vs. Phelps lower than a 27-28 loss to a say a Louisville Trinity team. If there was no human input into the algorithm then every team in the top 25 should be undefeated. Human input does matter in computer rankings, especially at the start of a season. The computer generally works out the biaseness toward the end of the season, but it is still there.

 

Did you actually read what you quoted? I don't disagree. I said you can disagree with how he chooses to value the data, but the computer isn't rooting for anyone. The data that you are disagreeing with came from last year and it is assigned a value (too high or too low is your choice), he doesn't just make it up. None of the computers just make things up.

Who programs those computers? When a season starts a coefficient must be placed on every team. Hence why does the computer rate a 47-0 win vs. Phelps lower than a 27-28 loss to a say a Louisville Trinity team. If there was no human input into the algorithm then every team in the top 25 should be undefeated. Human input does matter in computer rankings, especially at the start of a season. The computer generally works out the biaseness toward the end of the season, but it is still there.

 

Not necessarily.

 

If you're saying "all undefeated teams should be in the top 25", you're assuming the win quotient/factor trumps everything else, when most computers have built in things like scoring margin, road/home/neutral and strength of schedule.

 

And it is possible to start it at at zero/equal.

 

Take a four-team pool:

A goes 3-0, beats B, beats C, beats D

B goes 2-1, beats C, beats D, but loses to A

C goes 1-2, beats D, but loses to A and loses to B

D goes 0-3, losing to A, losing to B, losing to C

 

Any computer, with a coefficient or not, should tell you a win over B or C is worth more than a win over D (since every team but D beat D and D can't play itself).

 

Where I'm from, there's a 3-0 team (Elizabethtown) that is far from the rankings from other 3-0 teams (Male, Mayfield, South Warren) and behind a 2-1 area rival (Central Hardin). I don't think E'town fans are disappointed/mad/claiming computer bias given their schedule.

Dang I hate math. Math is mean to me.

Dang I hate math. Math is mean to me.

 

I took AP Calculus in high school. Got an A. Got a 4/5 on the AP test.

 

Don't remember a darn thing.

 

Also got As in geometry, physics, chemistry and algebra. You give me a problem to solve from one of those four subjects and I'll look at you like you just asked your dog to speak Latin.

Time for the annual "magic number" discussion in regards to computer ratings.

This ratings system does start off with a starting point, ratings coefficient, magic number, if you will. We have no idea what that number is or how it is arrived at, but speculation is that it is based upon prior success, last season's final ratings, and who knows what other factors.

 

This affects the early season numbers, and so does the lack of data for teams playing out of state.

 

As has been accurately pointed out, the ratings will improve in accuracy as the season wears on. More data = more accuracy.

Sigh....Both of you make good points, but your missing the one I am making. Try to understand how this is done. The computer algorithm does do cross comparisons and point differentials between opponents etc., but must be slated to an opening value based on human input (Hence the human that ranks a team higher in the preseason could be said to be "rooting" for that team to, for example, sell newspapers - just making an example not saying thats the case here). This in turn could be equated to a computer favoring one team over another based on this initial input. I know the computer dose not think, does not have emotions, etc. but how you program it does reflect how it will calculate ratings moving forward. Those rating must have a starting system which is inherently biases by nature because its humans that determine it. The algorithms used by computer rating systems can be more heavily swayed to SOS, common opponent, point differential and some of them do favor one or the other (this is the mathematical basis you were referring to). But, nonetheless, computers must have a starting point to rate and they do not consider all teams and all schedules equal prior to the season starting.

 

Why do rating systems not consider all teams equal prior to the start of the season? The reason there is human input (hence coefficient) is because we know what some teams will have based on history and last seasons results. Common sense tells you one team will be better than another based on size of school, etc. A computer does not know how well a team improved, what their tradition is (although there have been attempts to equate this), or what they have in the pipeline that will now be getting major minutes (again just an example of many variables).

 

Lets take a small Single A school that traditionally does not do very well (insert school here) and move all of Louisville Trinity's players to their program next year. Lets say they open 3-0 against their normal schedule with blowout wins. I bet they would be rated high in the polls. Now lets say the same result happens next year without Trinitys players against the same schedule (which is going to traditional be a weak schedule). They will not be in the top end of their class polls. This is an example of the human input into computers based on preseason.

 

As the year goes on, this biasness is decreased in accordance with the decreasing Chi-Square as more data comes in and thus the rankings get close. But for now that is not the case as (although small), there is human "rooting interest" in these initial ratings and why I dont put much value in them.

 

I suppose I should have never went down this rabbit hole, but hope this illustrates my point.

You and I are saying the same thing 15.

 

"but must be slated to an opening value based on human input " = The Magic Number

You and I are saying the same thing 15.

 

"but must be slated to an opening value based on human input " = The Magic Number

 

Yeah i know. In the midst of typing that response I did not have the chance to see your post. Your point about the magic number is spot on my friend and I agree we are in agreeance (if that makes sense).