Tag Archives: Monte Carlo simulations

Vols’ bowl chances drop to 29 percent, but 2-0 finish still possible (with video)

KNOXVILLE, Tennessee — Tennessee Vols keep losing and its opponent keep looking tougher.

Those have been the two big reasons the Vols’ likelihood of being bowl eligible has continued to slip in each week of our projections.

After Tennessee’s third consecutive loss and Vanderbilt’s impressive win at Florida, I’m now projecting the Commodores as a 2-point favorite at Neyland Stadium on Nov. 23 based on the current Sagarin ratings. That drops the Vols’ estimated likelihood of winning to 47 percent (down from 53 percent the last time we ran the simulation.

If Tennessee loses to Vanderbilt, the final game at Kentucky on Nov. 30 won’t matter as much. I’m giving the Vols a 4-point edge based on the Sagarin ratings, although I predict the actual point spread (and the final score) will be larger. In any case, that gives UT a 62 percent chance of winning (down from 64 percent a week ago).

With only two games left, running a simulation really isn’t necessary. We can quickly calculate the odds of going 2-0 (29.1 percent), 0-2 (20.1 percent) and 1-1 (50.8 percent).

We ran the simulation about 1,000 times and it was, in fact, very close to those numbers. (The longer I run the simulation, the more it should approach the numbers above).

Simulated results of UT's season, updated Nov. 10, 2013

The Vols’ likelihood of making a bowl has continued to drop since we started these projections.

Infographic  UT's bowl chances over time

Later in the week, we’ll take a look at where Tennessee might end up if the Vols reach .500. I’ll update last week’s bowl projections to reflect recent games and change some bowl estimates that people complained about (like Mizzou to the Sugar).

Vols’ chances of postseason bowl sink below 50 percent, according to simulation

Simulated results of UT's 2013 seasonKNOXVILLE, Tennessee — The last two weeks have not been kind to Tennessee’s bowl aspirations.

The odds of a trip to the postseason have slipped from about 80 percent after Tennessee gained a big win against South Carolina to just under 50 percent today after blowout losses to Alabama and Missouri.

Here are the estimated odds of Tennessee winning each of its last three games:

Nov. 9: vs. Auburn: 28 percent (down from 41 percent)

Nov. 23: vs. Vanderbilt: 53 percent (down from 62 percent)

Nov. 30: at Kentucky: 64 percent (down from 75 percent)

The odds listed above are educated guesses, but I tried to use a consistent method. I calculated an expected margin of victory based on Sagarin predictor ratings, which are sometimes similar to the point spreads set by Vegas. Then I converted the spread to a rough money line and converted the money line to an implied percentage. In the case of the Auburn game, I used the actual Vegas spread.

Then I entered all the numbers in my make-shift simulator and ran the rest of the season about 1,000 times.

Here’s what it spat out:

8-4, 5-3 overall: Eliminated (from 4.46 percent last week)

7-5, 4-4 overall: 11.32 percent (down from 25.73 percent)

6-6, 3-5 overall: 35.83 percent (down from 41.13 percent)

5-7, 2-6 overall: 40.16 percent (up from 24.92 percent)

4-8, 1-7 overall: 12.70 percent (up from 3.76 percent)

The records in bold would qualify the Vols for bowl eligibility. The simulation gives that a 47 percent chance of happening.

Through about 1,000 simulations, the mean number of UT wins was 5.46.

So, as has been the case for most of the season, the best guess for the Vols’ final regular season record is 5-7 or 6-6.

Vols’ odds to make a bowl game drop to about 70 percent after new projections

Butch Jones greets Nick Saban before Saturday’s game (AP photo by Dave Martin).

KNOXVILLE, Tennessee — Four games in November will decide Tennessee’s season.

A 2-2 record will send the Vols to their first bowl game since 2010, Derek Dooley’s first season. The Vols started that year 2-6, but won their last four games to reach bowl eligibility.

In 2013, the schedule is harder for first-year coach Butch Jones, but Tennessee must only win two of the final four games.

How likely are they to do so? We updated our Monte Carlo simulation after the Vols’ loss to No. 1 Alabama and tweaked some of the inputs to make them more consistent.

Here are the odds of Tennessee winning each of its last four games:

Nov. 2: @ Missouri: 23 percent

Nov. 9: vs. Auburn: 41 percent

Nov. 23: vs. Vanderbilt: 62 percent

Nov. 30: at Kentucky: 75 percent.

The odds listed above are educated guesses, but I tried to use a consistent method. I calculated an expected margin of victory based on Sagarin predictor ratings, which are sometimes similar to the point spreads set by Vegas. Then I converted the spread to a rough money line and converted the money line to an implied percentage.

Then I entered all the numbers in my make-shift simulator and ran the rest of the season about 1,000 times.

Here’s what it spat out:

4-0 finish, 8-4, 5-3 overall: 4.46 percent

3-1 finish, 7-5, 4-4 overall: 25.73 percent

2-2 finish, 6-6, 3-5 overall: 41.13 percent

1-3 finish, 5-7, 2-6 overall: 24.92 percent

0-4 finish, 4-8, 1-7 overall: 3.76 percent

The bold numbers give Tennessee bowl eligibility.

I ran a few smaller simulations just for the heck of it.

Missouri has a higher-than-I-expected 17 percent chance of winning out this season and a 45 percent chance of finishing 10-2, 6-2.

Will No. 1 Alabama finish the regular season with an undefeated record? My quick simulation put the odds at about 70 percent.

Updated predictions after Vols’ big win: UT’s bowl chances now nearly 80 percent

Infographic for Vols' final record

KNOXVILLE, Tennessee — Tennessee took a huge step toward reaching bowl eligibility with an upset win of then-No. 11 South Carolina on Saturday at Neyland Stadium.

The Vols (4-3, 1-2 SEC) only have to win two more games to guarantee a .500 record and a trip to the postseason.

How likely are the Vols to reach that goal? We recalculated our Monte Carlo simulation to find out.

Here are the somewhat subjective odds we are using for the final five games:

Opponent — Chance of winning

at Alabama — 8 percent

at Missouri — 38 percent

Auburn — 42 percent

Vanderbilt — 63 percent

at Kentucky — 81 percent

To make these percentages less subjective, I’ve tried to base them on the anticipated point spread for the game. Given that Alabama is favored by 28 over Tennessee, eight percent is extremely generous. (In other words, teams who are 28-point underdogs win much less often than that). Missouri could be undefeated and ranked in the top-5 when the Vols travel to Columbia, so that spread could grow, too. Auburn is likely to be 8-1 and in the top-10 when the Tigers travel to Knoxville, but the spread will be mitigated by the Vols’ home-field advantage. Vanderbilt is a wild card, too, as the Commodores just beat Georgia.

In any case, when we plug those numbers into the simulation and run it 1,000 times, here’s what we get.

Record in last five games — likelihood — overall record, SEC record

5-0 — 0.91 percent — 9-3, 6-2

4-1 — 9.75 percent — 8-4, 5-3

3-2 — 32.69 percent — 7-5, 4-4

2-3 — 36.45 percent — 6-6, 3-5

1-4 — 17.56 percent — 5-7, 2-6

0-5 — 2.64 percent — 4-8, 1-7

Bowl eligible? 79.8 percent

Miss a bowl? 20.2 percent

Because a Monte Carlo simulation uses random numbers, this projection may change every time the spreadsheet is edited. In other words, every time a change is made, the random simulation is re-calculated another 1,000 times, providing new projections.

[gdoc link=”https://docs.google.com/spreadsheet/pub?key=0AgBQnqCTaaxzdGZtNzJjVlRBb2lLWjllaF9QcFNrT1E&output=html&widget=true” height=”500″]

Odds say Tennessee has about a 57 percent chance of going to a bowl game

The Music City Bowl in 2010 was the last time Tennessee was in the postseason (Photo by Michael Patrick, News Sentinel)

KNOXVILLE, Tennessee — If all goes according to conventional wisdom, Tennessee’s chances of going to a bowl game after this season are about 57 percent.

How do we know that? Short answer: Math.

The long answer follows.

Remember those annoying but frighteningly accurate blog posts I did in an attempt to predict the Vols’ season? (You can review the preseason and mid-season versions here and here.)

The one flaw in those predictions was my inability to run a true simulation that would “play” the season many times over and over.

Well, I developed a makeshift Monte Carlo simulation using this spreadsheet. I only ran the simulation 1,000 times — ideally, I’d do it about 10,000 times, but I might break Google.

I used the following odds of winning for Tennessee’s final six games. (The process of obtaining these odds wasn’t arbitrary, but the percentages aren’t meant to be infallible. They’re just good estimates for the purposes of this exercise).

South Carolina — 33 percent

Alabama — 8 percent

Missouri — 40 percent

Auburn — 45 percent

Vanderbilt — 66 percent

Kentucky — 79 percent

Then I fed these odds into my Monte Carlo gizmo, and here are the results it spat out….

Likelihood of Tennessee finishing the season…

6-0: 0.20 percent

5-1: 4.92 percent

4-2: 17.83 percent

3-3: 34.84 percent

2-4: 30.33 percent

1-5: 10.04 percent

0-6: 1.84 percent

Bowl eligible? 57.79 percent

Not bowl eligible? 42.21 percent

When you open the spreadsheet, those numbers may change. Why? The simulation generates 1,000 new random numbers each time, essentially “playing” these six games another 1,000 times and generating new results.

[gdoc link=”https://docs.google.com/spreadsheet/pub?key=0AgBQnqCTaaxzdGZtNzJjVlRBb2lLWjllaF9QcFNrT1E&single=true&gid=0&output=html&widget=true” height=”450″]

If any math majors out there have any ideas on how to improve this model or make it easier, drop me a line.