Lucky teams win the close games - Page 2 - cougarguard.com — unofficial BYU Cougars / LDS sports, football, basketball forum and message board

Indy Coug · 01-26-2007, 03:57 PM

Quote:

Originally Posted by jay santos

Certainly randomness comes into play on any game determined in the final minute, but still the good teams are winning.

That isn't necessarily the case.

http://www.cougarboard.com/noframes/...tml?id=1999977

jay santos · 01-26-2007, 03:59 PM

Quote:

Originally Posted by Indy Coug

Take the basketball teams that were the best in games (eg. won at least 2/3 of those games) decided by 3 or 5 points or less and compare their records in similar games the very next year.

Teams change so much year to year, I don't know why you would think that's a good test.

Detroitdad · 01-26-2007, 04:03 PM

Quote:

Originally Posted by Indy Coug

Let's take the histograms of the winning percentages of teams in close games versus overall winning percentages:

Games decided by 3 points or less

Code:

Bin	Frequency
0	0
0.025	0
0.075	0
0.125	0
0.175	0
0.225	0
0.275	0
0.325	1
0.375	2
0.425	8
0.475	29
0.525	34
0.575	25
0.625	8
0.675	8
0.725	2
0.775	0
0.825	0
0.875	0
0.925	0
0.975	0
1	0

Overall winning percentage

Code:

Bin	Frequency
0	0
0.025	0
0.075	0
0.125	0
0.175	0
0.225	0
0.275	2
0.325	7
0.375	4
0.425	15
0.475	14
0.525	22
0.575	18
0.625	12
0.675	8
0.725	9
0.775	5
0.825	1
0.875	0
0.925	0
0.975	0
1	0

If you plot these histograms in Excel, note how much more tightly and more normally distributed around 50% the close games histogram is.

So luck is one possible explanation in a universe of many. Would you not need to have much more data, such as what percentage of games were played at home, what percentage of a team's games were close, the relative experience of the players and the effect that it has on the overall outcomes in close games. How about officials? How about homefield?

Anyone who has seen the bounces of the ball must believe in the effect of luck, but I have a hard time seeing it as THE explanatory variable.

Indy Coug · 01-26-2007, 04:03 PM

Well, there are many teams that consistently win year after year after year. You typically don't see 20 to 25 win teams drop off the face of the earth next year.

The reason they win year after year after year isn't because they consistently win a bunch of close games, it's because they beat the crap out of most of their competition, so the close games are less frequent.

Detroitdad · 01-26-2007, 04:07 PM

Quote:

Originally Posted by Indy Coug

That isn't necessarily the case.

http://www.cougarboard.com/noframes/...tml?id=1999977

I see where you are coming from with this, but does quality of opponent have something to do with it? Clearly teams like my dear old Alma Mater, Alabama, get most of their gaudy record by playing inferior opponents and that may be why their overall winning percentage is out of whack with the rest of their resume.

jay santos · 01-26-2007, 04:16 PM

Quote:

Originally Posted by Indy Coug

Well, there are many teams that consistently win year after year after year. You typically don't see 20 to 25 win teams drop off the face of the earth next year.

The reason they win year after year after year isn't because they consistently win a bunch of close games, it's because they beat the crap out of most of their competition, so the close games are less frequent.

Agreed. My intuitive sense is that good teams win a lot of blow out games and they also win more than pure luck would suggest of their close games. My basketball game data supports that. I don't think your tests with football data over multiple years is conclusive, and what I saw of your histogram, it supports my basketball data. Good teams don't win all their close games, but they win more than 50%.

Indy Coug · 01-26-2007, 04:37 PM

Here are the 25 winningest teams (1975-2005)

Code:

Team	              W	L	T	Win%
Nebraska          	309	73	2	80.7%
Florida St.       	289	83	4	77.4%
Michigan          	282	88	7	75.7%
Miami (FL)        	274	92	0	74.9%
Ohio St.          	278	92	7	74.7%
Oklahoma          	272	95	6	73.7%
Penn St.          	269	102	2	72.4%
Georgia           	260	105	6	70.9%
Alabama           	262	112	2	69.9%
BYU               	270	116	2	69.8%
Tennessee         	257	109	8	69.8%
Florida           	257	111	6	69.5%
Texas             	256	111	5	69.5%
USC               	255	112	8	69.1%
Notre Dame        	248	116	4	67.9%
Auburn            	241	117	7	67.0%
Texas A&M         	244	123	3	66.4%
Boise St.         	242	123	2	66.2%
Washington        	239	123	3	65.9%
Troy              	227	119	4	65.4%
Clemson           	234	122	8	65.4%
UCLA              	225	130	9	63.0%
Virginia Tech     	226	132	4	63.0%
Arkansas          	225	136	6	62.1%
Nevada            	226	138	1	62.1%

Here are their records in games decided by 3 points or less during that span:

Code:

Team	             W	  L	T	Win%
Florida St.       	14	23	4	39.0%
Virginia Tech     	19	26	4	42.9%
Alabama           	24	32	2	43.1%
Florida           	17	23	6	43.5%
Nevada            	21	25	1	45.7%
Michigan          	31	35	7	47.3%
Troy              	28	29	4	49.2%
Nebraska          	16	16	2	50.0%
Notre Dame        	29	29	4	50.0%
USC               	27	26	8	50.8%
Boise St.         	29	28	2	50.8%
Texas A&M         	27	26	3	50.9%
----------------------------------------------------------------------
Washington        	29	25	3	53.5%
Ohio St.          	25	21	7	53.8%
Clemson           	34	28	8	54.3%
UCLA              	39	30	9	55.8%
Arkansas          	31	21	6	58.6%
Auburn            	35	23	7	59.2%
Texas             	34	19	5	62.9%
Tennessee         	39	21	8	63.2%
BYU               	40	22	2	64.1%
Georgia           	43	22	6	64.8%
Miami (FL)        	32	16	0	66.7%
Oklahoma          	30	12	6	68.8%
Penn St.          	39	16	2	70.2%

12 of the 25 teams (essentially half) have a close game winning percentage of 50.9% or worse. I would have expected more of these teams that have consistently won over the last 30 years to individually have succeeded in these types of games during the same span. The data still shows a fairly strong central tendency towards 50.0%, even for the very good teams.

jay santos · 01-26-2007, 05:01 PM

Quote:

Originally Posted by Indy Coug

Here are the 25 winningest teams (1975-2005)

Code:

Team	              W	L	T	Win%
Nebraska          	309	73	2	80.7%
Florida St.       	289	83	4	77.4%
Michigan          	282	88	7	75.7%
Miami (FL)        	274	92	0	74.9%
Ohio St.          	278	92	7	74.7%
Oklahoma          	272	95	6	73.7%
Penn St.          	269	102	2	72.4%
Georgia           	260	105	6	70.9%
Alabama           	262	112	2	69.9%
BYU               	270	116	2	69.8%
Tennessee         	257	109	8	69.8%
Florida           	257	111	6	69.5%
Texas             	256	111	5	69.5%
USC               	255	112	8	69.1%
Notre Dame        	248	116	4	67.9%
Auburn            	241	117	7	67.0%
Texas A&M         	244	123	3	66.4%
Boise St.         	242	123	2	66.2%
Washington        	239	123	3	65.9%
Troy              	227	119	4	65.4%
Clemson           	234	122	8	65.4%
UCLA              	225	130	9	63.0%
Virginia Tech     	226	132	4	63.0%
Arkansas          	225	136	6	62.1%
Nevada            	226	138	1	62.1%

Here are their records in games decided by 3 points or less during that span:

Code:

Team	             W	  L	T	Win%
Florida St.       	14	23	4	39.0%
Virginia Tech     	19	26	4	42.9%
Alabama           	24	32	2	43.1%
Florida           	17	23	6	43.5%
Nevada            	21	25	1	45.7%
Michigan          	31	35	7	47.3%
Troy              	28	29	4	49.2%
Nebraska          	16	16	2	50.0%
Notre Dame        	29	29	4	50.0%
USC               	27	26	8	50.8%
Boise St.         	29	28	2	50.8%
Texas A&M         	27	26	3	50.9%
----------------------------------------------------------------------
Washington        	29	25	3	53.5%
Ohio St.          	25	21	7	53.8%
Clemson           	34	28	8	54.3%
UCLA              	39	30	9	55.8%
Arkansas          	31	21	6	58.6%
Auburn            	35	23	7	59.2%
Texas             	34	19	5	62.9%
Tennessee         	39	21	8	63.2%
BYU               	40	22	2	64.1%
Georgia           	43	22	6	64.8%
Miami (FL)        	32	16	0	66.7%
Oklahoma          	30	12	6	68.8%
Penn St.          	39	16	2	70.2%

12 of the 25 teams (essentially half) have a close game winning percentage of 50.9% or worse. I would have expected more of these teams that have consistently won over the last 30 years to individually have succeeded in these types of games during the same span. The data still shows a fairly strong central tendency towards 50.0%, even for the very good teams.

Of course it's going to be close to 50%. There's no question randomness will come into play in 3 point or less games.

A lot of those games were against each other. Some of those teams weren't very good in some of those years. And still, the average is 55%.

Archaea · 01-26-2007, 05:12 PM

Good teams probably play fewer close games to increase their chances of winning. If you're in too many close games, you will lose some.

Indy Coug · 01-26-2007, 05:14 PM

For the 25 winningest teams of the last 30 years:

Games decided by 4 points or more - win% of 72.0% (7,821 games)
Games decided by 3 points or less - win% of 54.8% (1,442 games)

That's a sizable gap in probabilities, given the sample size. Certainly a helluva lot closer to 50% than 72%.

If all I knew about the outcome of the game was the point spread, and knew absolutely nothing about the teams involved, I would make a lot more money (or lose less money) betting with a coin flip.

01-26-2007, 05:12 PM	#19
Archaea Assistant to the Regional Manager Join Date: Aug 2005 Location: The Orgasmatron Posts: 24,338	Good teams probably play fewer close games to increase their chances of winning. If you're in too many close games, you will lose some. __________________ Ἓν οἶδα ὅτι οὐδὲν οἶδα

01-26-2007, 04:03 PM	#14
Indy Coug Senior Member Join Date: Jan 2006 Location: Between Iraq and a hard place Posts: 7,569	Well, there are many teams that consistently win year after year after year. You typically don't see 20 to 25 win teams drop off the face of the earth next year. The reason they win year after year after year isn't because they consistently win a bunch of close games, it's because they beat the crap out of most of their competition, so the close games are less frequent.

01-26-2007, 05:14 PM	#20
Indy Coug Senior Member Join Date: Jan 2006 Location: Between Iraq and a hard place Posts: 7,569	For the 25 winningest teams of the last 30 years: Games decided by 4 points or more - win% of 72.0% (7,821 games) Games decided by 3 points or less - win% of 54.8% (1,442 games) That's a sizable gap in probabilities, given the sample size. Certainly a helluva lot closer to 50% than 72%. If all I knew about the outcome of the game was the point spread, and knew absolutely nothing about the teams involved, I would make a lot more money (or lose less money) betting with a coin flip.