As the calendar reaches July and the college football
preseason magazines begin to hit the shelves, I have an annual tradition of
doing a simulation-assisted deep dive into the coming college football
season. Over the years, I have developed
an algorithm and tracking spreadsheet that I use to generate in-season power
rankings and projected point spreads. In the preseason, I have found that I can
use the same algorithm to project the outcome of the full season, if I use
preseason rankings to seed the math. Basically, I ask the question: if the preseason rankings are correct, what
is the most likely outcome of the coming season, considering each team’s
actual schedule and a historically reasonable probability of upsets,
specifically when good teams face slightly weaker teams on the road?
Now, I could (and have in the past) spend several paragraphs
going into the details of my methodology, but the key points are as follows. This
year, I used the average rankings from 5 separate sources (Athlon, Lindys, Phil
Steele, ESPN, and S&P+) to develop a “consensus” ranking of all 130
Division 1 teams. When I list a team, I will always do so with a number in
parenthesis which corresponds to that team’s consensus ranking. I perform
multiple simulations with slightly different parameters that each give slightly
different information. In one case, I
use parameters that generally match those that I would generate in-season. This
simulation does a good job of generating point spreads, which correspond to
game-by-game probabilities of victory.
This calculation also gives the non-integer “expected value” of the
number of total wins each team is expected to have in the season. I run a second simulation that injects a
little more parity (i.e. likely upsets) into the system and kicks out W-L
records and standings.
Finally, this year I also added a third simulation that
modifies that results of the first one.
In a post I made a few weeks ago, I performed an analysis of the
accuracy of preseason ranking. Using
this data, I could “perturb” my initial simulation by injecting the probability
that each team’s preseason ranking is not correct. The result of this simulation is a
probability distribution that any given team will win “x” number of games. The standard deviations of these distributions
are all around 2.0 which, based on historical data, should be about right. I will give a table with these values for
each conference in turn. As a warning,
this post is super long, so I have broken up into sections and tried to bold the key information for those that
just want to skim. So, without further
ado, here are my current answers to several burning questions in this
preseason:
1. What can we expect from MSU this year?
As you might guess, the preseason rankings are not terribly
kind to MSU after the dumpster fire of the season last year, and Athlons,
Lindy’s and S&P+ all rank MSU in the mid to high 40s. ESPN is less kind at
58, but the worst preseason ranking comes from Phil Steele at 70. The bad news is that Phil Steele does tend to
be the most accurate and Steele might be the only source that correctly
considered the coming absences in the roster due to the most recently dismissed
players. As a result, MSU’s consensus
ranking is 52nd, and when I run MSU’s schedule through the simulation,
the win distribution looks like this:
As you can see, the math suggests that MSU is most likely to win 5 or 6 games and MSU’s “expected value”
of wins (m in the table) is only 5.45.
So, if you saw some of the over/under projections from Vegas that had MSU at
between 5 and 6, well, my math sadly agrees.
The slightly better news is that my W/L simulation does have MSU at 6-6
and bowl eligible. In a way, the
schedule is a bit forgiving, despite the projections from some that MSU’s
schedule is the toughest in the nation (which oddly enough, my calculations
actually confirm). The reason that I say
this is that MSU plays 4 home games against teams with lower consensus
rankings: BGSU (104), WMU (76), Indiana (58), and Maryland (69), plus draws a
terrible Rutgers (97) team on the road.
Even if one or more of those teams are better than advertised, the home
field advantage (plus playing Rutgers) should carry MSU to at least 5
wins. So, the trick is to win at least
one “toss-up” game, the most likely of which is to beat Iowa (44) in East
Lansing, but I would certainly not rule out upsetting Notre Dame (20) in East
Lansing, or either Minnesota (49) or Northwestern (29) on the road.
A closer look at the win distribution table gives slightly
different way of looking at the season.
If you trust these numbers (and I think the math is solid based on
historical data), it suggests MSU has a
48% chance to be bowl eligible (win 6 or more games), and a 47% chance of
winning 5-7 games. I think this is
in line with most MSU fan’s expectations.
But, if you want to be more optimistic, the data suggest that there is
actually a 9% chance that MSU finishes with 9 or more wins and an 18% chance MSU finishes with 8 or more
wins. The 8-win level is roughly the
same as winning as a 13-point underdog in a single game setting (which incidentally
is the same spread for MSU’s victory in Columbus in 2015). But, for those Negative Nancys out there (and
you know who you are), there is also a 20% chance that MSU finishes win 3 wins
or less. As for me personally, if I
leave my slide rule at home, I will stay
on the optimistic side of reasonable probability and project MSU at 7-5 this
year.
2. What about the rest of the Big Ten? How will that shake out, most likely?
Simply put, OSU (2)
is likely to win the conference, but look out for Wisconsin (11) as both
teams have favorable schedules. The
projected win distribution for the full conference is shown here:
As for the Big Ten East, I think everyone reading this knows
that OSU (2), PSU (7), and UofM (12) are all projected to be strong in 2017.
The schedule has each of those three teams with one home game and one road game
against the other two, and in this type of situation, my simulation always
projects that the home team will win in each of those match-ups. Thus, the
projection is OSU will beat PSU, PSU will beat Michigan, and Michigan will
actually beat OSU (which I will believe when I see). But, the win distribution clearly shows that
OSU has a much higher expected win total (9.83) than either PSU (8.98) or
Michigan (8.42), in part because OSU is supposed to be better, but also due to
their schedules. Actually, when it comes
down to it, both Penn State and Ohio State have fairly similar schedules. Both
teams play Iowa (44) on the road and both teams face Nebraska (46), but Penn
State does have the slightly tricky road game at Northwestern (29) who might be
good enough this year to cause problems, while OSU draws Illinois (91) with
their other west cross-over. Advantage
Buckeyes. As for the Wolverines, their main problem is that they must travel to
the only team in the West that might be good enough to challenge the “Big 3 of
the East” this year: the Wisconsin Badgers (11). As such, the Wolverines
are by far the most likely of the Big 3 to lose another conference game. So,
even though Michigan might actually beat the Buckeyes this year… but they will
likely still only finish in 3rd place in the East. That said, my
math also suggests that there is a 26%
chance UofM finishes at 7-5 or worse.
Try floating this number at the watercooler at work and see how that is
received. In any event, the fact that the Buckeyes host Penn State
this year suggests that East Division trophy is most likely to land in Columbus.
Out West, the strongest team is projected to be Wisconsin
(11) and they definitely have a schedule that should enable a very, very good
season. In fact, I predict that the Badgers have over a 9% chance to go
12-0, which is the 5th highest of all 130 D1 teams, and the highest
among Power 5 teams. A look at their
schedule shows why. On paper, the three
best B1G teams that they will face this year: Michigan (12), Northwestern (29),
and Iowa (44) all must travel to Madison.
Their toughest conference road games are at Nebraska (46), Minnesota
(49), and Indiana (58), which seems quite manageable. On top of everything, their toughest
non-conference game is a trip out (to the real) west to face BYU (38). While MSU might still be having nightmares
about Mormons, I am doubting the Badger’s schedule is keeping them up at
night. My W/L simulation predicts Wisconsin to be one of only 2 teams to
finish the regular season undefeated.
If this all comes to pass, my simulation would also predict that an 11-1
OSU (2) squad would handle the 12-0 Badgers
(11) in Indy to claim the B1G title.
As a final note on the Big Ten, there are a couple of
notable early September non-conference games that will impact the overall
perception of the conference. For OSU
(2), they will host Oklahoma (4) on September 9th. As OSU is higher ranked and at home, I
project a victory for the Buckeyes. The
trickier game is the opening weekend match-up of Michigan (12) and Florida
(13). Right now, I have the Wolverines winning a very close game and the
current Vegas spread (-3 for UofM at the open, and rising) tends to agree. The only other non-conference game of
projected relevance on the national stage is Penn State’s rematch at home against
Pitt (34) which should be another W for the conference. If the
Big 3 can win all 3 of these marquee match-ups, the Big Ten will be sitting
pretty for multiple (3-4) NY6 Bowl bids and possibly even multiple Playoff
spots.
3. How do I expect the SEC to shake out?
Briefly, Alabama (1) is in good shape to repeat overall and Florida (13) looks to run away with the East by multiple games. The projected win distribution for the full conference is shown here:
As always, the SEC is hard to handicap because over half of
the conference is predicted to have Top 30ish level ability. But, this year
Bama (1) has the dual advantage of both being projected to be very good as well
as having a manageable schedule, which explains why they have the best chance
in the SEC to go 12-0 as well as the highest expected value of total wins
(9.18). Of the Top 8 projected SEC teams other than the Tide, Bama avoids
Florida (13) and Georgia (17) and draws LSU (10), Tennessee (23), Arkansas
(31), and Ole Miss (35) at home. The Tides most challenging conference games,
by far, are at Auburn (9) to close the season and at Texas A&M (27). My W/L
simulation does have Bama losing to Auburn, but still finishing in first place
in the West at 7-1. Auburn (9), in
turn, I project to lose at LSU (10) and also in a close one at Texas A&M
(27). It certainly is interesting that
both Bama and Auburn must travel to A&M.
My simulation suggests that Bama will be good enough to win this game,
while Auburn will not be and this difference will decide the winner of the
West. We will see. Meanwhile, LSU has the misfortune of drawing
Florida (13), Bama (1), and Tennessee (23) all on the road, making them much
less likely to be a West contender.
In the SEC East,
the situation is more clear. Florida is projected
to be the best team, and their schedule is kind. They draw Tennessee (23)
at home and Georgia (17) in Jacksonville (as usual) and their toughest West
cross-over opponent [LSU (10)] also must travel to Gainesville. The Gator’s
toughest conference true road game is at South Carolina (40). In contrast,
Georgia travels to both Tennessee (23) and Auburn (9) [Not mention road trips
to Notre Dame (20) and Georgia Tech (32) in the non-conference], while
Tennessee (23) must travel to Bama (1) and Florida (13). In the end, I project another Alabama-Florida
conference title game where The Tide will once again come out on top.
But, how will this
impact the overall playoff / NY6 scene? Well, that will be determined in
large part by the SEC’s performance in several key non-conference games, the
first of which is the monster match-up on opening weekend between Alabama (1)
and Florida State (3) which could just as likely be a playoff preview. The
Auburn (9) at Clemson (6) match-up the following week is almost as big. I already mentioned Florida’s (12) big
match-up with Michigan (12), but don’t forget the Gators also get to host the
Seminoles in Thanksgiving weekend. The
problem for the SEC is that of the Top 12 non-conference games involving SEC
teams (based on the sum of the consensus rankings), I only have the SEC winning
4 of the 12 (Bama over FSU, Florida over FSU, LSU over BYU, and Arkansas
over TCU) and losing the other 8 (including Clemson over Auburn, Michigan over
Florida, Notre Dame over Georgia, Clemson over South Carolina, GA Tech over
both Georgia and Tennessee, Louisville over Kentucky, and UCLA over Texas
A&M). If this all comes to pass, only
the SEC champion (Alabama) would have a shot at a playoff bid, most likely.
4. Can Clemson and/or the ACC in general repeat their strong performance of last season?
In two words:
probably not. The Tigers (6) sneaky-tricky schedule will basically cause them to fall
short of Florida State (3) and their much softer conference slate while Miami
(18) wins the battle of attrition the Coastal Division. The projected win
distribution for the full conference is shown here:
Interestingly, the win distribution data shows that 5 total
teams [Florida State (3), Clemson (6), Louisville (15), Miami (18), and VA Tech
(25)] all have very similar profiles (i.e. ~2% chance to run the table, 11%
chance to win 11+, ~30% chance to win 10+, etc.) All five teams are projected to be Top 25
quality and if you add NC State (26), GA Tech (32), and Pitt (34) to the mix,
you have a recipe for a very competitive league. As such, my
W/L simulation does predict a lot of parity in the ACC, especially in the
Coastal Division, where I project a 4-way, 6-2 tie between Miami, VA Tech, GA
Tech, and Pitt. In this scenario, Miami (18) would win the tie breaker,
as one of their losses is projected to occur at FSU, while the other three tied
teams are projected to pick up 2 losses within their Division. But, in this type of situation, an upset here
or there could shift the balance of power and results in the Coastal Division
greatly.
In the Atlantic Division, once again the win distribution
data gives a hint as to how the division is likely to shake out. In general, FSU’s (3) expected win total (8.84) and winning percentages are a
slightly, but noticeably, better than Clemson’s (8.30). The reason, once
again, comes down to schedule. At first glance, you may notice that FSU travels
to Clemson this year, a game I would project them to lose. So, you would think
that Clemson would have the edge. The problem is that this is FSU’s only
serious conference road test of the year. Their other 3 road games are at Wake
Forest (63), Duke (64), and Boston College (74). The Seminoles have the benefit of drawing
Louisville (15), Miami (18), and NC State (26) all at home. In contrast,
Clemson has an almost mirror universe schedule as they must travel to
Louisville (15) and NC State (26), and their tough Coastal Division cross over
game is at VA Tech (25). My W/L
simulation projects FSU to go 7-1 and win the division, while Clemson is
projected to finish 5-3 even though the Tigers are predicted to have two
wins over Florida State (3) and Auburn (9).
As for Florida State
and the ACC’s national perception, that could get interesting. In general,
I have the ACC going 10-7 in non-conference games involving other Power 5
conferences, which is second to only the Pac-12’s projected 5-3 record. (For
completeness, I have the Big Ten at 6-6, the Big 12 at 4-6, and the SEC
surprisingly at only 6-9). The problem is that the projected champion, Florida
State (3), has two brutal non-conference games vs. Alabama (1) and Florida (13)
which I project both as losses. As a result, the ACC Champ is projected to
finish at only 9-3, which I think will make the ACC a tough sell for the
playoffs, even though they might in reality have the strongest conference.
5. Was Washington a flash in the pan, or will they repeat as Pac-12 Champs?
No and likely no. Although
Washington (8) is the favorite in the North and still Top 10 quality, my
simulation suggests USC (5) will come away with the crown. The projected win distribution for the full
conference is shown here:
Based on the win distribution, it looks like the Pac-12 will be straightforward. No team seems to
have a major schedule advantage or disadvantage, so the two highest ranked
teams (USC and Washington) are projected to win their Divisions. Washington may have slightly tougher road, as
the 3 next best Pac 12 teams are all located in the North [Stanford (14),
Oregon (22), and Washington State (28)].
The Huskies must travel to Palo
Alto this year, and I project that this will be their sole loss in the regular
season as they host the other two teams. Stanford, though, is unlikely to
be able to leverage this victory to win the Division, as the Cardinal have
tough road games at USC (5), Washington State (28), and Utah (39) and 2 losses
in those 3 games seems likely. In the south, UCS gets the benefit of hosting
UCLA (30), Utah (39), and Stanford (14), while their toughest road game is a
cross-division game at Washington State (28). Even if USC is quite a bit worse than advertised, they could still win
the South Division by several games.
In fact, I have the Trojans running the table in the Pac-12 and picking
up their sole loss of the season in South Bend to the Fighting Irish. If this comes to pass, the Trojans would also
pick up a September non-conference home win over Texas (19) which would add
nicely to a Play-off worthy resume.
6. Does anyone still care about the Big 12?
I suppose someone must, but I am not sure who. Maybe their moms? (HI MOM!) Actually,
it looks like the Big 12 might provide some Oklahoma-sized intrigue this year,
as my simulations suggest the Cowboys of OK State (16) might steal the show. The projected win distribution for the full
conference is shown here:
The preseason publications suggest that the Big 12 will have
five Top 25 caliber teams in Oklahoma (4), Oklahoma State (16), Texas (19), TCU
(21), and Kansas State (24). What the
win distribution and my simulation suggests, however, is that the Oklahoma
schools and K-State have a significant scheduling edge over Texas and TCU. Despite the fact the Oklahoma is on paper the
best team in the league, they must travel to K-State (24), OK State (16), and
have a semi-neutral site game against Texas (19). My W/L simulation suggests losses at K-State
and OK State for the Sooners, along with that early season loss in Columbus to
the Buckeyes to leave them at an underwhelming 9-3. OK State, in contrast, draws TCU (21),
Oklahoma (4), and K-State (24) all in Stillwater and only needs to travel to
play at Texas (19). Due to this advantage, the Sooners essentially need to beat the Cowboys
in their own house to win the league.
The problem with the Big 12 in general, however, is that OK State does
have a sneaky tough September road game at Pitt (34) on September 16th,
which I project them to lose, which would bring their record to 10-2, which is
a number that would make the playoff selection committee pause. My
current projection is that the Big12 will once again get left out of the
play-offs.
7. Will Notre Dame return to glory this year?
Perhaps yes, if 9-10
wins is good enough to be called “glorious.” For whatever reason (a.k.a “recruiting” and
“tradition”), despite the fact that the Irish were pretty much just as bad as
MSU last year and generally have inferior coaching, the pundits placed the
Irish (20) right back in the Top 25 this year. If you believe that, well, then
you should also believe that they have a pretty good shot at having a decent
season. My W/L simulation projects them
to go 10-2, with notable wins vs. USC (5), Georgia (17), and NC State (26) in
South Bend, and with the 2 loses coming at Stanford (14) and at Miami
(18). But, considering the Irish tend to
play several mediocre teams and not a lot of really bad teams, their margin of
error might be a bit narrower. The win
distribution data (which I will include in the next section), suggests an
expected number of wins at only 7.48 and only a 33% chance of winning 9 games
or more and almost a 20% chance of finishing with fewer than 6 wins. Basically,
the Irish seem to have a lot of games where they have a ~60% of winning. They might be favored in 10 games, but
actually winning all 10 might not be very likely.
8. Which Group of Five teams will be in play for a NY6 Bowl? Who will be this year’s Western Michigan?
Most of the preseason
magazines say the Bulls of South Florida, and they are probably correct. The top consensus ranked non-Power Five
teams are: USF (33), Boise State (36), BYU (38), Memphis (48), Houston (50),
Western Kentucky (55), San Diego State (56), Colorado State (57), Navy (59),
Toledo (62), and Appalachian State (68).
As an additional wild card team, my win distribution data suggests that
Ohio (82) also has a surprisingly good chance of going undefeated (11.3%, 3rd
best in the country), so I will add them to the mix as well. The full win distribution data for the teams
above (plus Notre Dame for good measure) are shown below:
As you can see, my simulation is also very high on USF, so much so that the W/L simulation picks them to finish 12-0
and the win distribution calculation says they have a 40% chance to go
undefeated (which is by far the best in the country) and a 67% chance to at
least win 11. A look at their schedule
shows the reasons for this optimism.
Their toughest non-conference game is against Illinois (91) and it is
even in Tampa (just in case playing in Champaign would be too tough). The
highest ranked team that they will face is Houston (50) and that is also a home
game. So, their toughest test might be a road game against UCF (72). To secure
the NY6 bid, it is necessary to be a conference champ, so USF would need to win
the AAC Title game, most likely against Memphis (48). I project Memphis to finish 11-1 with a sole
loss at Houston but an upset win over UCLA (30), which would likely make the
Tigers a good bet to take the NY6 bid in the event of an upset in the AAC title
game.
As for the remaining contenders, BYU (38) is actually not
eligible for the Group of 5 NY6 slot, but I thought it would be nice to include
them in the table above since they were so nice the Spartans last year. The
favorite in C-USA [Western Kentucky (55)] has a 23% chance to run the table,
and are perhaps the next most likely team if the AAC champ for some reason has
2 or more losses. WKU also has a very
weak schedule, as they play the next 2 best teams in their conference at home
[LA Tech (78) and Middle TN (85)] In the non-conference, WKU plays Illinois (91)
as well which I project that they can win, and must travel to Vanderbilt (54),
where I project a loss, but as upset here might just be enough to carry the
Hilltoppers to the NY6. As for the
Mountain West teams, this year Boise
(36) has three tough road games at Washington State (28), BYU (38), and San Diego
State (56), while the Aztecs play at Arizona State (51) and vs. Stanford (14). So, it seems unlikely that either team is
going to finish with less than 2-3 loses. In
MACtion, I do like Ohio (82) to win the East and Toledo (62) to win the West.
Toledo (62) travels to Miami (FL) (18), so 11-1 is likely the best they can do,
but Ohio’s non-conference schedule includes winnable games at Purdue (87), vs.
Kansas (98), and vs. UMass (120), while their toughest conference road game is
at EMU (101). So, maybe a MAC repeat in
the NY6 is possible after all. Finally,
unless App State (68) can somehow beat Georgia (17) in Week 1, I don’t see the
Sunbelt Champ making a run at the NY6.
9. If we add all this up, how would the playoffs and NY6 shake out?
At this point, I need to put the math aside and start to use a little judgement on where teams would be ranked, assuming the W/L results of my simulation are correct. Based on the analysis above, I think that three of the playoff teams would be obvious: SEC Champ Alabama, Big Ten Champ Ohio State, and Pac-12 Champ USC. The tricky part is that fourth team. I project the ACC Champ to be Florida State, but they would enter the Bowl season with a 10-3 record (with loses to Bama, Florida, and Clemson). The Big 12 Champ is projected to be OK State, with a slightly better 10-2 record, but with the 2 losses coming at the hands of less-than-impressive Texas (19) and Pitt (34). I think both teams would get left out in the cold. As I look through the list of projected W-Ls, there are several 1-loss and 2-loss teams that would have an argument:
Wisconsin: 11-1 (only loss to OSU in the BTCG and win over
Michigan, NW, and BYU)
Penn State: 11-1 (only loss to OSU in Columbus and wins over
Michigan, NW, and Pitt)
Washington: 11-2 (losses to USC and Stanford and wins over
Oregon and Washington State)
Florida: 11-2 (losses to Alabama and Michigan and wins over
Florida State, LSU, Georgia, and Tennessee)
Michigan: 10-2 (losses to Penn State and Wisconsin and wins
over OSU and Florida)
Notre Dame: 10-2 (losses to Stanford and Miami and wins over
USC and Georgia)
There are a couple of other ACC and Big 12 teams that would
also have 2 losses (Louisville, VA Tech, GA Tech, and K-State) but it seems
quite unlikely that they would get a nod over any of the teams above when their
own conference champ did not make the cut.
If we consider the resumes of the teams above, the Big Ten looks good and I think it
would be a crap shoot to select either Wisconsin or Penn State for the 4th slot. It would likely come down to the infamous
“eye ball test” and the margin of victory or loss in the individual games. Personally, I would value the Division title
and neutral field loss to OSU and give
the Badgers the nod. As for playoff
seeding, both Alabama and OSU would be coming off late season losses to their
biggest rivals, while USC would be on more of a streak (having only lost to
Notre Dame in late October), but I would still put Bama and OSU at 1-2. Plus, this would allow a proper Big Ten-Pac
12 Rose Bowl match-up which is as the Good Lord clearly intended:
Sugar Bowl: #1
Alabama vs. #4 Wisconsin
Rose Bowl: #2 Ohio
State vs. #3 USC
In this scenario, I would have Bama (1) over OSU (2) in the
National Title game, based simply on their preseason consensus rankings.
As for the other four NY6 Bowls, the 2017 rotation has fewer
contracted bowls, but FSU, OK State, and USF (as the Group of 5 representative)
would all have guaranteed slots. The
other 5 slots would most likely be filled with the teams listed above that were
playoff contenders. I think the most
likely pairings would be as follows:
Orange Bowl: Florida
State vs. Notre Dame
Cotton Bowl: Oklahoma
State vs. Michigan
Fiesta Bowl: USF vs.
Washington
Peach Bowl: Penn
State vs. Florida
10. What are the most impactful games on the overall 2017 season (including some unexpected ones?)
How about these:
09/02: Alabama (1) vs. Florida State (3). The winner makes an early claim to a playoff
spot while the loser’s margin of error gets thinner.
09/02: Florida (13) vs. Michigan (12). The winning team’s
conference looks to have the inside track for a 2nd playoff bid.
09/09: Oklahoma (4) at Ohio State (2). The losers conference
could take a hit which will matter in NY6 placement.
09/09: Auburn (9) at Clemson (6). I don’t have either team
even in the NY6, but a few upsets in other places might make this game much
more important.
09/16: Oklahoma State (16) at Pitt (34). As weird as it may
seem, this game could be the difference between a playoff berth and the just
the Cotton Bowl for the Cowboys.
All three games between PSU, OSU, and Michigan. If any road team can steal a victory, that
teams gets a fast pass to the BTCG and the playoffs
10/21: USC (5) at Notre Dame (20). This might be all that
stands between USC and a 13-0 record heading into the playoffs.
11/04: Auburn (9) at Texas A&M (27). If Auburn gets the win, the balance of power
in the SEC West shifts in their favor.
11/04: Oklahoma (4) at Oklahoma State (16). This looks like
the de facto Big 12 Title Game right now, but can the winner avoid other losses
and make the playoffs?
11/04: Stanford (14) at Washington State (28). If Stanford can win this one on the road, the
balance of power in the Pac-12 North shifts in their favor and sets them up for
the Fiesta Bowl, or better.
11/25: Florida State (3) at Florida (13). If FSU can get the road win, their odds of a
playoff berth go way up.
12/02: America Conference Title Game: The winner of this contest is the most likely
Group of 5 team in the NY6.
11. Yeah, yeah, yeah, that’s all cool, but can we actually use this simulation of yours to do something useful, like go to Vegas?
Uh, maybe? I track the performance of my algorithm in-season
against the spread. Some years it does
well (around 53% ATS) but other years it floats below 50% and I don’t think I
can beat Vegas with it. Those guys are
just too damn good. That said, my expected value of wins for each team is
essentially a calculated over/under and with limited data sets (only about
2 years) my preseason predictions have done pretty well against the over/under
values that I pull during the summer. For entertainment purposes only, here
is a chart of the over-under for all 130 teams as compared to my calculations,
with the biggest outliers (i.e. more than 1.5 games different) identified. Use at
your own risk!
Well, that’s all I got this year, so I hope you enjoyed this
breakdown as much as I enjoyed preparing it.
Go Green.
Comments
Post a Comment