RPI vs KenPom

[Jamar Howard 4 President]

But you have Kenpom as being better 20 times and not better 20 times. The sum of the parts says they are the same.

"Not better 20 times" is a very strange way to say it. Also, a very misleading way to say it.

KenPom was better 20, equal 19, lesser once.

[wufan]

"Not better 20 times" is a very strange way to say it. Also, a very misleading way to say it.

KenPom was better 20, equal 19, lesser once.

I feel like you eyeballed this one with no criteria. I expect more from you. Just for instance Eastern Washington (definitely RPI advantage)...KP and RPI are much closer to about equal when I eyeball it.

[wufan]

	Dec	29		Jan	21
Team	RPI	KenPom		RPI	KenPom
VCU	3	14		4	18
Wichita State	8	13		10	11
Northern Iowa	10	26		24	20
Old Dominion	12	62		29	43
George Washington	17	38		45	63
Colorado State	19	55		37	66
LSU	20	54		41	40
Buffalo	21	82		47	77
Louisville	26	6		32	13
Texas	30	10		16	12
Oklahoma	36	12		25	10
Georgia Tech	38	89		87	94
Wofford	39	88		50	108
Incarnate Word	40	174		148	179
Penn State	41	84		101	122
Eastern Washington	47	104		61	119
High Point	58	148		85	148
Stony Brook	59	150		94	132
Minnesota	62	27		107	53
Gardner Webb	63	167		100	174
Ohio State	65	11		60	22
St. Francis PA	67	143		124	177
Lafayette	68	142		118	172
UConn	80	32		70	47
Syracuse	81	31		58	62
Texas Southern	91	178		116	153
Florida	94	14		75	39
Radford	97	173		187	167
Notre Dame	98	23		39	15
Texas Arlington	99	188		139	187
Indiana	104	44		43	54
Creighton	126	76		151	110
Wyoming	143	56		86	72
McNeese State	144	302		219	264
New Orleans	158	305		252	301
Fairleigh Dickinson	168	294		246	322
Delaware State	170	248		205	254
North Dakota	185	319		268	316
NJIT	197	258		184	208
Southern Illinois	321	202		289	237

I bolded the ones that I felt the RPI on Dec 29th gave a better representation of the current RPI than did KenPom. I came up with 17.

[Jamar Howard 4 President]

wufan, what in the world did you use as your criteria? For example, how do you possibly say that the RPI did a better job with Buffalo than KenPom did? Your selections make no sense to me.

[wufan]

RPI today for Buffalo is 47. RPI predicted 21. That's a difference of 26. KP predicted 82. That's a difference of 35. Since the selection committee uses something very similar to the RPI to seed, I'm using the RPI as the gold standard. As best I can tell based on current data, KP was wrong on the 29th and they're just as wrong today.

[Anthroshock]

You know the NCAA has an RPI that is freely available...

http://www.ncaa.com/rankings/basketball-men/d1/ncaa-mens-basketball-rpi

[Jamar Howard 4 President]

RPI today for Buffalo is 47. RPI predicted 21. That's a difference of 26. KP predicted 82. That's a difference of 35. Since the selection committee uses something very similar to the RPI to seed, I'm using the RPI as the gold standard. As best I can tell based on current data, KP was wrong on the 29th and they're just as wrong today.

You have apparently missed my entire point. I'm not interested in predicting what the RPI will be in a few weeks, or at years' end. Atleast not in this thread im not.

I'm trying to show that more often than not, as differences between RPI and KenPom converge as the season progresses, it is the RPI coming toward what KenPom already said, not the other way around. Im trying to show that November and December KenPom ratings are better than RPI ratings at the same time.

Teams like North Dakota are a perfect example. On 12/29, KenPom and RPI disagreed significantly. Now, they are much closer, despite the fact that KenPom's ranking of North Dakota remained virtually unchanged. It was the RPI that adjusted to match what KenPom has been saying all along.

I've already admitted my analysis isn't perfect, but from a big picture perspective, I think some obvious conclusions can be made.

[wufan]

You have apparently missed my entire point. I'm not interested in predicting what the RPI will be in a few weeks, or at years' end. Atleast not in this thread im not.

I'm trying to show that more often than not, as differences between RPI and KenPom converge as the season progresses, it is the RPI coming toward what KenPom already said, not the other way around. Im trying to show that November and December KenPom ratings are better than RPI ratings at the same time.

Teams like North Dakota are a perfect example. On 12/29, KenPom and RPI disagreed significantly. Now, they are much closer, despite the fact that KenPom's ranking of North Dakota remained virtually unchanged. It was the RPI that adjusted to match what KenPom has been saying all along.

I've already admitted my analysis isn't perfect, but from a big picture perspective, I think some obvious conclusions can be made.

This is what you said at the beginning of this thread:

"I would say superior could be determined a couple ways. One of my expectations is that the computers (RPI & KenPom) will spit out numbers in March that are fairly close to each other, and that KenPom today will be closer to that future rank than the RPI of today."

You said they'd be virtually the same by the end of the year. If KenPom supports KenPom by not changing and you think that's superior then good for you.

[Jamar Howard 4 President]

Sometimes, a discussion just needs to end with a friendly "you're crazy" and both parties move on. Nice chatting with you wufan

EDIT: One last point that might help bring us together. You are focusing on what the RPI says now as if the current RPI is the gold standard. The committee does indeed use the RPI, but they use the FINAL RPI, not the December, or January, or February RPI. My argument is that the RPI is on a trajectory toward being useful. Although it is better today than it was in December, it is still not done self-correcting as the sample size is still too small. Only at the end of the year is it anywhere close to being "the gold standard". I'm happy to argue whether RPI, or KenPom, or Sagarin, or the pattern of drool left on Fever's pillow in the morning, or anything else is the best tool on Selection Sunday. That is a good debate. However, my purpose in this thread is not about which ranking system is best at the end of the year. My argument is simply that the RPI needs all year before it gets into the vicinity of being the best, and during the season itself, KenPom is much better.

Let's compare the 12/29 numbers of KenPom and RPI to the RPI on selection sunday. I think at that point you and I will be able to gain some traction in our conversation.

[Jamar Howard 4 President]

	Dec	29		Feb	19
Team	RPI	KenPom		RPI	KenPom
VCU	3	14		13	26
Wichita State	8	13		17	13
Northern Iowa	10	26		15	11
Old Dominion	12	62		53	88
George Washington	17	38		86	94
Colorado State	19	55		26	67
LSU	20	54		55	33
Buffalo	21	82		58	71
Louisville	26	6		20	18
Texas	30	10		34	19
Oklahoma	36	12		16	10
Georgia Tech	38	89		110	75
Wofford	39	88		40	97
Incarnate Word	40	174		156	197
Penn State	41	84		100	87
Eastern Washington	47	104		73	138
High Point	58	148		93	128
Stony Brook	59	150		118	145
Minnesota	62	27		99	64
Gardner Webb	63	167		162	211
Ohio State	65	11		36	14
St. Francis PA	67	143		182	226
Lafayette	68	142		154	213
UConn	80	32		80	60
Syracuse	81	31		65	56
Texas Southern	91	178		153	217
Florida	94	14		77	31
Radford	97	173		143	174
Notre Dame	98	23		27	20
Texas Arlington	99	188		152	168
Indiana	104	44		29	44
Creighton	126	76		135	109
Wyoming	143	56		79	121
McNeese State	144	302		281	269
New Orleans	158	305		328	310
Fairleigh Dickinson	168	294		308	323
Delaware State	170	248		240	265
North Dakota	185	319		275	311
NJIT	197	258		176	177
Southern Illinois	321	202		280	230

Green = KenPom was clearly better on 12/29 than RPI
Red = RPI was clearly better on 12/29 than KenPom
White = Debatable who was better

I think it has become very clear that KenPom was the much better ranking system to use on 12/29. Whenever there is a big discrepancy between KenPom and RPI in November, December, or January, the majority of the time KenPom is the one you should trust.

[Cdizzle]

I think it has become very clear that KenPom was the much better ranking system to use on 12/29. Whenever there is a big discrepancy between KenPom and RPI in November, December, or January, the majority of the time KenPom is the one you should trust.

Did you have a set criteria or just kindof eyeball it? This is an interesting exercise. Difficult to take in all the data, especially without knowing your cutoffs.

[dregn]

I don't think kenpom should get credit for being more accurate when both did an awful job (i.e George Washington) or if they were both off equally (i.e. VCU), although VCU is even more unique with their injury situation.

[Jamar Howard 4 President]

I don't think kenpom should get credit for being more accurate when both did an awful job (i.e George Washington) or if they were both off equally (i.e. VCU), although VCU is even more unique with their injury situation.

Regarding VCU, I agree their injury situation makes them a unique situation. Throw them out if you like. However, if we are going to make a judgement at all, here is why I say that KenPom was clearly better.

12/29, VCU 3, KenPom 14
02/19, VCU 13, KenPom 26

If you believe VCU is the #13 team in the country, then KenPom was much more accurate on 12/29.
If you believe VCU is the #20 team in the country, then KenPom was much more accurate on 12/29.
If you believe VCU is the #26 team in the country, then KenPom was much more accurate on 12/29.

Any reasonable rank of VCU today leads to the conclusion that KenPom's rank of 14 was more accurate than the RPI's rank of 3.

[Jamar Howard 4 President]

Regarding GW, of course both rating systems were way off. However, it is not fair to say that KenPom did equally as bad. His system ranked them 38 back on 12/29. Clearly, that is bad considering they are now hovering around 90. However, RPI had GW at 17 on 12/29. Both overrated GW, but at least KenPom never had them anywhere close to the top 20. Neither system did well, but still, KenPom did better. That fact is indisputable.

[AndShock]

I would be more interested to see how KenPom performed against RPIForecast