forums

ESPN FPI PROJECTIONS FOR MIAMI HURRICANES 2018 SEASON

avax

avax

Sophmore
Joined
Jan 28, 2016
Messages
3,709
Game 1: LSU
Date: September 2nd
ESPN FPI Projection: Miami win
Win Probability: 65.7%
Projected Record: 1-0
Game 2: Savanah State
Date: September 8th
ESPN FPI Projection: Miami win
Win Probability: 99.9%
Projected Record: 2-0
Game 3: @Toledo
(Photo: Chris Nicoll, USA TODAY Sports)
Date: September 15th
ESPN FPI Projection: Miami win
Win Probability: 86.3%
Projected Record: 3-0
Game 4: FIU
(Photo: Getty)
Date: 98.3%
ESPN FPI Projection: Miami win
Win Probability: 98.3%
Projected Record: 4-0
Game 5: North Carolina
Date: September 27th
ESPN FPI Projection: Miami win
Win Probability: 85.3%
Projected Record: 5-0
Game 6: Florida State
(Photo: Christopher Nee, 247Sports)
Date: October 6th
ESPN FPI Projection: Miami win
Win Probability: 68.3%
Projected Record: 6-0
Game 7: @Virginia
Date: October 13th
ESPN FPI Projection: Miami win
Win Probability: 85.8%
Projected Record: 7-0
Game 8: @Boston College
(Photo: Brett Carlsen, Getty)
Date: October 26th
ESPN FPI Projection: Miami win
Win Probability: 62.5%
Projected Record: 8-0
Game 9: Duke
Date: November 3rd
ESPN FPI Projection: Miami win
Win Probability: 79%
Projected Record: 9-0
Game 10: @Georgia Tech
(Photo: USA TODAY Sports)
Date: November 10th
ESPN FPI Projection: Miami win
Win Probability: 63.5%
Record: 10-0
Game 11: @Virginia Tech
Date: November 17th
ESPN FPI Projection: Miami win
Win Probability: 57.2%
Record: 11-0
Game 12: Pittsburgh
Date: November 24th
ESPN FPI Projection: Miami win
Win Probability: 89.5%
Record: 12-0
 
Advertisement
Whatever. Those butt munches have us ranked like 14 or 16 or something in their FPI. It's all bullsh!t.

Beat LSU.
 
edge said:
Love it. How accurate is this system normally?
Click to expand...

FPI’s accuracy
Over the last 10 seasons, the FPI favorite has won 75 percent of FBS-versus-FBS games, which is comparable to the Vegas closing line. Looking at the last four seasons, that percentage has risen to 77 percent, and in games that FPI and Vegas differed, the FPI favorite won 55 percent of the time. If you want to follow along with how FPI performs throughout the season, feel free to go to the prediction tracker website.

FPI favorites in FBS-only games
By percentage chance to win
WIN PCT 50-60%57%60-70%65%70-80%73%80-90%84%90-100%96%

It’s worth noting that the results of analytics such as FPI are not black-and-white -- they give us likelihoods of outcomes, not certainties. Therefore, when FPI gives a team a 75 percent chance to win and that team loses, FPI is not necessarily “wrong.” A team with a 75 percent chance to win should lose one out of every four times, and if every team with a 75 percent chance to win does in fact win, the system is broken.
Therefore, we track how well a system is calibrated, and as can be seen in the chart above, when FPI gave a team between a 70 percent and an 80 percent chance to win, those teams actually won 73 percent of the time.
Of course, no system will be 100 percent accurate, and every year there are teams that FPI is wrong on. Michigan State is one team that has consistently outperformed FPI’s expectations over the years. As college football fans, we do not agree with every prediction or rating, but in total, FPI has proven to be accurate.
 
Cane[]_[]DigIt said:
FPI’s accuracy
Over the last 10 seasons, the FPI favorite has won 75 percent of FBS-versus-FBS games, which is comparable to the Vegas closing line. Looking at the last four seasons, that percentage has risen to 77 percent, and in games that FPI and Vegas differed, the FPI favorite won 55 percent of the time. If you want to follow along with how FPI performs throughout the season, feel free to go to the prediction tracker website.

FPI favorites in FBS-only games
By percentage chance to win
WIN PCT 50-60%57%60-70%65%70-80%73%80-90%84%90-100%96%

It’s worth noting that the results of analytics such as FPI are not black-and-white -- they give us likelihoods of outcomes, not certainties. Therefore, when FPI gives a team a 75 percent chance to win and that team loses, FPI is not necessarily “wrong.” A team with a 75 percent chance to win should lose one out of every four times, and if every team with a 75 percent chance to win does in fact win, the system is broken.
Therefore, we track how well a system is calibrated, and as can be seen in the chart above, when FPI gave a team between a 70 percent and an 80 percent chance to win, those teams actually won 73 percent of the time.
Of course, no system will be 100 percent accurate, and every year there are teams that FPI is wrong on. Michigan State is one team that has consistently outperformed FPI’s expectations over the years. As college football fans, we do not agree with every prediction or rating, but in total, FPI has proven to be accurate.
Click to expand...

What an informative and well worded response. Are you lost?
 
Advertisement
Cane[]_[]DigIt said:
FPI’s accuracy
Over the last 10 seasons, the FPI favorite has won 75 percent of FBS-versus-FBS games, which is comparable to the Vegas closing line. Looking at the last four seasons, that percentage has risen to 77 percent, and in games that FPI and Vegas differed, the FPI favorite won 55 percent of the time. If you want to follow along with how FPI performs throughout the season, feel free to go to the prediction tracker website.

FPI favorites in FBS-only games
By percentage chance to win
WIN PCT 50-60%57%60-70%65%70-80%73%80-90%84%90-100%96%

It’s worth noting that the results of analytics such as FPI are not black-and-white -- they give us likelihoods of outcomes, not certainties. Therefore, when FPI gives a team a 75 percent chance to win and that team loses, FPI is not necessarily “wrong.” A team with a 75 percent chance to win should lose one out of every four times, and if every team with a 75 percent chance to win does in fact win, the system is broken.
Therefore, we track how well a system is calibrated, and as can be seen in the chart above, when FPI gave a team between a 70 percent and an 80 percent chance to win, those teams actually won 73 percent of the time.
Of course, no system will be 100 percent accurate, and every year there are teams that FPI is wrong on. Michigan State is one team that has consistently outperformed FPI’s expectations over the years. As college football fans, we do not agree with every prediction or rating, but in total, FPI has proven to be accurate.
Click to expand...

Yeah, well said. Still...

FPI gives us a greater chance of beating FSU than it does GT or BC? (I don't necessarily disagree with them on BC) OP also says FPI has us going undefeated. FPI has us favored in all of our games, which is good, and I think accurate, but that's a different thing than projecting we go undefeated. But it also projects us at around #14 in the country and predicts we will lose three games.

If FPI is generally accurate, great. Can you explain the inputs they use? I haven't done much research, but as you said, MSU historically outperforms. It is still my opinion that FPI is weighted by results and biases that aren't necessarily accurate to certain teams.

When FPI tells me that ND is #6, FSU who barely made bowl eligibility last season is #18, and 4-7 UF is 21, while we are #14, I question that. Could just be the orange and green glasses.

ESPN Football Power Index - 2018 - ESPN
 
FPI is also very fluid and will change from week to week. As the poster above pointed out, Duke was 4-0 and was ranked very highly in all the major defensive statistics, and we had played BCC and trailed Toledo at the half, so FPI gave a slight edge to Duke. FPI before the season begins is something I wouldn't rely too heavily on. Let's see how we do against LSU and how the FPI lines adjust for future opponents after the first week is in the books.
 
Advertisement
Well they had ND and VT over us i think. F that, if Rosier dont show up along with the OL, could be like last year or the year before
 
BC is a going to a be a battle . IMO them and GT are the toughest ACC games this season . FSU will not be very good IMO .
 
g8rh8rMD said:
Yeah, well said. Still...

FPI gives us a greater chance of beating FSU than it does GT or BC? (I don't necessarily disagree with them on BC) OP also says FPI has us going undefeated. FPI has us favored in all of our games, which is good, and I think accurate, but that's a different thing than projecting we go undefeated. But it also projects us at around #14 in the country and predicts we will lose three games.

If FPI is generally accurate, great. Can you explain the inputs they use? I haven't done much research, but as you said, MSU historically outperforms. It is still my opinion that FPI is weighted by results and biases that aren't necessarily accurate to certain teams.

When FPI tells me that ND is #6, FSU who barely made bowl eligibility last season is #18, and 4-7 UF is 21, while we are #14, I question that. Could just be the orange and green glasses.

ESPN Football Power Index - 2018 - ESPN
Click to expand...

Fck. I think he made it clear this was about probability. An 80% win chance doesn't mean we'll win just that we should win 8/10...maybe today is a bad day. When you look at the smallish chance that we'll lose each game taken together, it actually becomes likely that we'll lose at some point. I don't know exactly what they're looking at, but a basic understanding of probability clears up most of what you're talking about. If nothing else, think how rarely teams go undefeated. It's not just good teams knocking each other off, even good teams with soft schedules have trouble with it. The numbers will tell you we should have won 8 or 9 games last year and that we shouldn't have lost to Pitt. They aren't really a prediction of what will happen, just what's most likely to happen.

As for O&G glasses, yeah you've got them on. This system doesn't care what you or I think will happen, only what the data suggest is likely to happen. So while you and I think there's no way we lose to Toledo, a rational observer has to admit there's a chance.

And as for those rankings, I would be surprised if the numbers didn't say that nd and fsu should be better than we give them credit for and we should be worse. We, and the voters right now, overvalue the Canes based on what we've got on paper. I think the Canes winning 11+ is a smart bet, but the numbers might not bear that out until we see what we have at OL and QB.
 
Last edited:
That means ESPN expects us to win 9.4 games, so somewhere between 9 and 10 games out of the 12.

(For anyone who isn't great at math and probabilities.)

0.657 + 0.999 + 0.863 + 0.983 + 0.853 + 0.683 + 0.858 + 0.625 + 0.790 + 0.635 + 0.572 + 0.895 = Expected Value = 9.413 wins
 
Advertisement
RightSaidFred said:
That means ESPN expects us to win 9.4 games, so somewhere between 9 and 10 games out of the 12.

(For anyone who isn't great at math and probabilities.)

0.657 + 0.999 + 0.863 + 0.983 + 0.853 + 0.683 + 0.858 + 0.625 + 0.790 + 0.635 + 0.572 + 0.895 = Expected Value = 9.413 wins
Click to expand...
and if you multiply all the numbers you get the odds of going undefeated which in this case happens to be about 4.5%
 
RightSaidFred said:
That means ESPN expects us to win 9.4 games, so somewhere between 9 and 10 games out of the 12.

(For anyone who isn't great at math and probabilities.)

0.657 + 0.999 + 0.863 + 0.983 + 0.853 + 0.683 + 0.858 + 0.625 + 0.790 + 0.635 + 0.572 + 0.895 = Expected Value = 9.413 wins
Click to expand...
Is that not the confidence of win total. Like its confiedent that miami will win at least 9.5 games?
 
kryptonite said:
Fck. I think he made it clear this was about probability. An 80% win chance doesn't mean we'll win just that we should win 8/10...maybe today is a bad day. When you look at the smallish chance that we'll lose each game taken together, it actually becomes likely that we'll lose at some point. I don't know exactly what they're looking at, but a basic understanding of probability clears up most of what you're talking about. If nothing else, think how rarely teams go undefeated. It's not just good teams knocking each other off, even good teams with soft schedules have trouble with it. The numbers will tell you we should have won 8 or 9 games last year and that we shouldn't have lost to Pitt. They aren't really a prediction of what will happen, just what's most likely to happen.

As for O&G glasses, yeah you've got them on. This system doesn't care what you or I think will happen, only what the data suggest is likely to happen. So while you and I think there's no way we lose to Toledo, a rational observer has to admit there's a chance.

And as for those rankings, I would be surprised if the numbers didn't say that nd and fsu should be better than we give them credit for and we should be worse. We, and the voters right now, overvalue the Canes based on what we've got on paper. I think the Canes winning 11+ is a smart bet, but the numbers might not bear that out until we see what we have at OL and QB.
Click to expand...

So... you like FPI? Aside from that, I have little understanding for what your point was. You seem to view FPI solely as a prism through which to view one game between two opponents. OK, I get it, it's not exactly rocket science to look at FPI % chances between team A and team B.

If you think it's that simple, fine, but don't be @'ing me with your Fcks, bro.

I get, and acknowledged in my quoted post, that there is a homer factor when looking at one's own team.

Still doesn't explain FPI putting ND at 21.7 and Miami at 16.5. Two teams with identical records over the last 13 games, with similar offseason attrition and returning players. The one projected to lose by 5 killed the other one just four/five games ago.

Let me know when your next doctorate course on probabilities and statistics opens for registration and I'll be sure to get in on it. Until then, I'll pass on the lecture. I get it. It seems to me you're trying to make some other kind of statement here about how ESPN isn't biased? I'm not sure what that has to do with my post.

As I said, I can't (and don't care to) dispute statistics that show that, over the course of a season, FPI is a good indicator of the likely result of a given game. But what are the inputs they use? What data suggest what is likely to happen?

How is a Notre Dame team that eeked out a win over LSU by 4 in the last game that either team played, 11.3 points better than that same team?

My entire point was that FPI still appears to me to contain flawed data and/or biases that result in outliers not accurately represented by the formula. These outliers may not significantly impact the formula's success in suggesting probabilities of outcomes of individual matchups, but they sure look odd sometimes when presented as a ranking. Until ESPN "opens the source code", I don't see why we should put any faith in it. But, I'll bet you FPI will be used in many many arguments about which teams should and should not make the playoffs.
 
Advertisement
Just for fun.... assuming Miami wins every game week-by-week, this is how the FPI changes.

So, for example, before week 1, our expected win total is 9.4, and our odds of running the table are 4.5%.

If we beat LSU on Sunday, our expected win total jumps to 9.8, and our odds of going undefeated jump to 6.8%. And so on.


EV.jpg
 
Advertisement
Back
Top