Week 4 Betting Results

                                                                                                                                             

The Aggregate's week four results are in. We were unfortunately not able to publish our predictions for week 4 of the college football season, but we did make them. We should be able to publish them for this coming week even if we aren't able to mass advertise them yet. Without further ado here are the results.

Amazon Prime Free Trial

Overall:

The Aggregate: 45-18 (71.4%)

The Aggregate's main predictor. If a team has a 50.1% chance to win according to the Aggregate then they fit into this category. The only game the Aggregate didn't predict this week was the Texas Tech vs. California game as they both had a 50.0% chance to win.

ESPN's FPI: 49-14 (77.8%)

Taken off of ESPN's website. It should be noted ESPN had the worst cumulative rate last year, but early this year is leading.

W-L: 37-27 (57.8%)

We track this because it's often suggested that the most important information when predicting games is a teams win loss ratio. This is often not actually the case, but it's good to have the data to back that up. When teams have an equal record than the second decider is home field advantage.

Jeff Sagarin: 42-22 (65.6%)

By far the best predictor last year as it was at an over 80% hit rate going into the college bowl season. It is definitely a surprise to see it as the low man after one week. It will be interesting to see if this changes throughout the season. 

Vegas's Betting Line: 39-14 (73.6%)

Vegas had the almost the worst percentage last year, but has a good start this season. It should be noted that there are less games here because when we pulled the numbers there were still certain games the Vegas sites we follow had not posted a line for. 

LSU T-Shirts on Amazon

Summary:

ESPN's FPI has been the best indicator through this week. It will be interesting to see if that continues as it finished both behind the Aggregate and Jeff Sagarin's ratings last year. It is at least possible to think that ESPN's metric is better with limited data than the Aggregate or Jeff Sagarin's as last years data was more from the back half of the season then the front. 

Also Vegas's betting line did very well. Better than the Aggregate which is exactly opposite of what our goal is here. The fact that the Aggregate out performed the Vegas betting line was one of the reasons we decided to actually invest in this betting scheme this college football season. However, because of the way we are betting we still made out more details on that below.

Prime Video Free Trial

Betting Results:

The Aggregate made 11 bets for the fourth week of college football. We put money on 10 games in which the Aggregate had predicted at least an 80% chance or better for a team to win. We then more then doubled the money in three games in which the Aggregate saw a 90% chance or better. We also made one parlay bet that included four money line bets on teams in which the Aggregate saw one team as a 90% chance or better. 

We won 8 out of the 10 single game bets and won the parlay bet. As a result we came out ahead and made about a 6.5% gain on our money for the week. Unfortunately because of a dismal week 0 we are still down about 10% on our total money. 

Hurricane T Shirts on Amazon

Single Game Bets:

LSU over UCLA

Chance to win: 96.7%

Money line: -2100

Gain %: 2.9%

Duke over MTSU

Chance to win: 91.7%

Money line: -600

Gain %: 15.2%

Miami (FL) over USF

Chance to win: 100%

Money line: -850

Gain %: 11.8%

Mizzou over Vandy

Chance to win: 88.9%

Money line: -1200

Gain %: 7.1%

Sam Houston over New Mexico

Chance to win: 88.9%

Money line: -850

% Change: 7.1%

Kansas State over Brigham Young

Chance to win: 88.9%

Money line: -225

% Change: 4.8% (total)

Clemson over NC State

Chance to win: 88%

Money line: -1200

% Change: 7.1%

Memphis over Navy

Chance to win: 88%

Money line: -355

% Change: 4.8% (total)

Kentucky over Ohio

Chance to win: 86.1% 

Money line: -1350

% Change: 7.1%

Fresno State over New Mexico

Chance to win: 86.1%

Money line: -600

% Change: 14.3%

Parlay

Miami over USF

Duke over MTSU

LSU over UCLA

Indiana over Charlotte (100%)

Money line: -263

Return on Investment: 38.3%

Total

90% Single game picks: 3/3 

ROI: 8.9%

Remaining Single game picks: 5/7

ROI: 10.2%

Parlay: 1/1

ROI: 38.3%

Total: 9/11

ROI: 6.6%

Week 0 Aggregate CFB Picks

 

College football is finally back and the Aggregate has predicted the first four games. Some special notes out here before you begin using the Aggregate predictor as a betting guide. 

One, the aggregate was effective last year and correctly selected almost 79% of winners during the regular season college football games. That sample size was about 170 games. This is likely too small a sample size to prove the aggregates ability. Personally I do believe in the data and am wagering this year based on that research, but you should know the wagers are very small. With the development of more automation in my process I should have a much larger sample size to work with after this season. However, whatever advice we may give is all at your own discretion. Your decisions are you own. I do promise to be honest about the Aggregates winning and losing and all of the data so that you can see what is working and what is not. 

Amazon Prime Free Trial

Two, the Aggregate was unable to accurately predict against the spread or over/under last year. Officially the Aggregate was 12 games under .500 against the spread and 3 games under against the over/under. Both of these percentages would only have lost money last year and I do not plan on betting any money on these categories although we will still list that information. I have made some small changes where those formulas are concerned including now factoring standard deviation to those calculations and am hopeful to get better results this season. I will continue to track how the formula does.

Three, the Aggregate betting system will work this way. Because of my extremely small bank roll I will be betting on approximately 14 games a week with the exception of Week 0 and 1 which will be a combined 14. For me these 14 games will be chosen by whichever teams have the best chance to win. The goal is to win as many as possible and if we are winning 79 percent than we will be making money. The actual percentage should be much higher as the 79 percent was based on all games even games where the favorite my only be favored by a percent or two. 

Four, other than the 14 moneyline bets I will place one bet per week on a parlay. Parlay's are notoriously tricky, but are very lucrative. This parlay will always just include moneyline bets as that's the category where the Aggregate has done so well in. The method behind picking those games will be simple. Any game where the Aggregate see's a 90% chance for a team to win or better will be included in the parlay for that week. The reason behind this is because last spring when the Aggregate broke down the NCAA basketball tournament it never failed to correctly predict a winner when the chance to win was 90% or higher. This is also a very small sample size and not one specifically from college football. I do not have the data for this for college football last year as I wasn't tracking it yet, but we'll have more data on this after this year. You should know that in the NCAA tournament only 9 games out of 65 met this criteria so the likelihood of many games with this are small. If for some reason there are not multiple games that meet this criteria than we will not place a parlay bet. 

Five, although we will be betting the same amount each week it would be smart to only bet small amounts early on. The reason for this is simple. At the beginning of the season we don't have very much data. The Aggregates preseason projections are based off of last year's teams rating, players lost and gained, and coaches lost and gained. This is our first year coming up with a preseason rating so it is very new and has no proven data. Because of this I weighed also Phil Steele's best power rating into the calculation so as to mitigate any outliers. As we gain more data about this I will move away from this. As we move further into the season though teams will show us who is best and who we can trust based on their results. This is the surest way to feel confident in betting with the Aggregate.

So now that you have been forewarned here is the Aggregate's Week 0 predictions:

 

Week Zero CFB Predictions
Teams	Record	Score Prediction	Chance to Win	Vegas Line	O/U	Moneyline
Saturday, August 24th
#10 Florida State	0-0	38	100.0%	-10.5	55.5	-$450.00
Georgia Tech	0-0	24	0.0%			$360.00
Montana State	0-0	46	33.0%	-13.5	54.5	-$550.00
New Mexico	0-0	44	66.7%			$425.00
SMU	0-0	42	100.0%	-27.0	55.5	-$3,500.00
Nevada	0-0	13	0.0%			$1,500.00
Sunday, August 25th
Deleware State	0-0	21	0.00%			N/A
Hawaii	0-0	35	100.00%	-40.5	55.5	N/A

The aggregate really likes three games. Florida State over Georgia Tech, SMU over Nevada, and Hawaii over Delaware State. In each of these cases the Aggregate sees the favorite as having a 100% chance to win. It is very early so that data can be unreliable, but the Aggregate is using three different preseason evaluation tools to get those results. The Aggregate is betting on one of those games. It is just the Florida State game. That games odds are the best this week and we are anticipating about a 19% ROI. All of our data points to their victory.  The Aggregate would have bet the SMU game, but the ROI basically came out to 0% for the level of investment that we are inputting. We would also have bet on the Hawaii game, but there is no moneyline available for that game. The Aggregate did invest in a 2 pick parlay with Florida State and SMU moneyline picks. We are anticipating a 23%ROI for this bet. 

FSU T-Shirts on Amazon

The one game that the Aggregate did not touch is the Montana State, New Mexico game. There are many reasons for this. First, our point calculator, which is a mean based predictor, predicted that Montana State would win the game while our Chance to win calculator, which is a median based predictor, holds that New Mexico will win the game. Second, as we said before this data is pretty small and we are unsure of it's ability to accurately predict anything yet. A couple of our models really like New Mexico's coaching hires and that has given them an edge that they may not actually have especially considering it's their first game. Third, we do not have a good system yet for evaluating FCS schools. Although we have made strides to correct this we still are not able to track all FCS schools and their results and as a result their ratings are based off of an  amalgamation of indicators, but not their actual results or off season moves. 

Mountain West Coaches Adjusted Winning Percentage

   

I've often felt that winning percentage is not a great look at how good coaches actually perform as it does not factor in a multitude of items including environment, player talent and competition. As a result I devised a formula to try and tackle some of that problem and give a more accurate portrayal of what part of the wins and losses the head coach was responsible for. 

This metric, the adjusted win percentage for FBS coaches, tries to take into account at least the players that each coach is responsible for. For example when a new head coach wins in their very first year the credit only lies somewhat with them. Many of the players they are winning with were recruited by the previous staff. Often these coaches go on to not be successful in the long term. The best examples I can think of are Ed Orgeron at LSU or Gene Chizik at Auburn. Both should be noted actually had their best year a few years in, but neither lasted after a prolonged time. On the flip side new coaches often lose heavily in their first year because it takes time to recruit and install new systems and to get a full buy-in from players. Think Sabin's first year at Alabama or Bobby Petrino's first at Arkansas.

The basic thrust of the formula is this, the longer a coach has coached a team the more accountable he is for their wins and losses. The transfer portal has changed some of this in that the formula now holds coaches more accountable sooner. Gone are the days when you had four years to show some results.

Special notes, these statistics are based on fbs wins and losses only. The time period for the coaches is endless so all the coaches years are included, but it is only an evaluation of current coaches not coaches from the past. 

With all that said here is the fourth in the series, the Mountain West conference. You will notice that this includes only teams eligible for the Mountain West Championship game. Despite playing six games against Mountain West teams this year neither Oregon State nor Washington State are eligible in the Mountain West standings at the time of this posting. 

Biggest thing that stands out about the Mountain West is their lack of under .500 coaches. They have only two. Compared to the conferences I've evaluated so far that is extremely low. The next closest conference to that was Conference USA with six. 

What does that mean for the Mountain West? Mostly I think it means they have a lot of retread coaches. There are three coaches in the Mountain West with 100 plus adjusted wins. None of the conferences I have evaluated so far have had any. Seven Mountain West coaches had previous coaching stops as head coaches and then there is Troy Calhoun who has been at Air Force since 2007. 

So who is the Coach with the best adjusted win percentage in the Mountain West? Well it's Spencer Danielson although the adjusted winning percentage only gives him credit for .75 wins and .25 losses. In other words they went 3-1 when he took over last year, but since he took over at the end of the season the adjusted winning percentage only gives him credit for 25% of the wins and losses. We'll get a better idea of how good he is after this season. 

Behind him is Bronco Mendenhall with a career adjusted winning percentage of 64.7%. You've got to feel good if you're a Lobos fan right now. You just hired a coach with a very good winning pedigree. 

Note: Jay Sawvel and Jeff Choate have no wins or losses as an FBS coach which is why they're an incomplete here. 

MW Coaches	Team	W	L	W Pct
Jay Sawvel	Wyoming	0.0	0.0	#DIV/0!
Jeff Choate	Nevada	0.0	0.0	#DIV/0!
Spencer Danielson	Boise State	0.8	0.3	75.0%
Bronco Mendenhall	New Mexico	115.3	62.8	64.7%
Troy Calhoun	Air Force	117.3	75.3	60.9%
Jeff Tedford	Fresno State	101.8	67.8	60.0%
Ken Niumatalolo	San Jose State	95.8	75.4	56.0%
Blake Anderson	Utah State	56.8	44.8	55.9%
Barry Odom	UNLV	21.0	17.3	54.9%
Jay Norvell	Colorado State	33.3	28.3	54.1%
Sean Lewis	San Diego State	20.8	22.0	48.5%
Timmy Chang	Hawai'i	6.5	13.0	33.3%

Mountain West T-Shirts on Amazon

End of Year Rankings for Mountain West

 The end of the year rankings are here. The Mountain West finished as the Aggregates 8th best conference this season. The biggest change to the rankings since championship week was the drop of Utah State following their loss to 45-22 loss to Georgia State. Probably the biggest surprise in the end of the year Mountain West rankings is that UNLV who played in the MWC championship game is only ranked 5th in the Aggregate.