Tourney Wrapup II — The 3 Point Shooting Teams

By A.J. Coltrane

For purposes of Vegas/March Madness this year I thought it might be helpful to consider how 3-point dependent individual teams were during the regular season. Some of that screwing around was discussed in this post.

Below are the top 10 teams in what I called “Vol” (for volatility) in that post. Below it’s called 3DEP (3 point dependency), which is what I’ll stick with going forward. 3DEP is simply 3 point percentage (x) % of 3 pointers attempted. At least in theory these ten teams rely upon the 3-pointer for a large percentage of their offense, so I’d kind of expect their performances to be very “up and down”. (To have a wider than usual deviation in their level of performance.) If the 3-pointer abandoned them, they were screwed. The flipside is that maybe they can get hot and beat a better team.

Look what happened in the game that knocked them out of the tournament:

Team (Seed)	3DEP	3P%	2P%	3PA%	3P% in Loss	2P% in Loss
Creighton 7	1667	42.1	56.4	39.6	10.5	41.2
Iowa St. 10	1624	37	52.1	43.9	48.0	40.7
Florida 3	1539	38.1	54.9	40.4	20.0	45.7
Valparaiso 14	1493	37.5	56.1	39.8	28.0	41.4
Belmont 11	1485	37.6	57.1	39.5	29.6	50.0
SD St 13	1434	39.4	50.3	36.4	34.8	50.0
St. Mary’s 11	1421	37.3	53.2	38.1	20.0	37.2
Davidson 14	1409	36.6	52.2	38.5	35.3	44.8
Arizona 6	1376	36.3	50.3	37.9	33.3	50.0
Boise St. 13	1362	38.7	48.2	35.2	38.9	54.3
Total		38.1	53.1	38.9	31.0	45.4

Well, duh. They didn’t shoot well and they lost the game. They didn’t shoot well from 2 either. I’m sure that happened to a lot of the losers, regardless of how many 3’s they attempted. If something hadn’t gone wrong they’d still be playing. We don’t really learn anything from that.

Continuing to mess with data though — as a group those are some low seeds — 7 of the 10 teams are double digit seeds. They’d all be underdogs starting with their first game. Look at when they got knocked out of the tournament:

Team (Seed)	Lost To	Round?
Creighton 7	Duke	Round of 32
Iowa St. 10	Ohio St	Round of 32
Florida 3	Michigan	Elite 8
Valparaiso 14	Michigan St	Round of 64
Belmont 11	Arizona	Round of 64
SD St 13	FGCU	Round of 32
St. Mary’s 11	Memphis	Round of 64
Davidson 14	Marquette	Round of 64
Arizona 6	Ohio St	Sweet 16
Boise St. 13	La Salle	Play In

In aggregate the underdogs went 2-4 in their first “real” game (St Mary’s won their play-in game.) Given a fat enough money line of between +150 and +200, maybe they’d be worth betting on. Hmm…

Here’s 2012:

2012 Team	3P%	3PA%	3DEP	Round?
Florida 7	38.0	44.6	1695	Elite 8
Vanderbilt 5	38.8	41.2	1599	Round 32
Iowa St. 8	37.3	41.6	1552	Round 32
Belmont 14	38.1	40.1	1528	Round 64
Wisconsin 4	36.8	41.3	1520	Sweet 16
Missouri 2	39.8	37.9	1508	Round 64
Creighton 8	42.4	35.0	1484	Round 32
South Dakota St. 14	39.0	37.8	1474	Round 64
Duke 2	37.1	38.6	1432	Round 64
Nevada Las Vegas 6	36.7	38.9	1428	Round 64

Note the two upsets — both the #2 seeds lost. Is that useful? Maybe. But I am taking a note on it for later.

As far as the underdogs: This year there were only two. And they both lost their first game. Our underdogs are now 2-6.

One more year. Here’s 2011:

2011 Team	3P%	3PA%	3DEP	Round?
Belmont 13	37.8	42.3	1599	Round 64
Wisconsin 4	37.4	41.2	1541	Sweet 16
Virginia Commonwealth 11	37.0	41.2	1524	Final Four
Michigan 8	35.3	43.0	1518	Round 32
Northern Colorado 15	38.3	39.4	1509	Round 64
Notre Dame 2	38.6	38.6	1490	Round 32
Louisville 4	36.2	40.8	1477	Round 64
Arizona 5	39.7	36.1	1433	Elite 8
Richmond 12	39.0	36.6	1427	Sweet 16
Boston University 16	35.5	40.1	1424	Round 64

I wouldn’t bet a 16 seed to beat a one seed regardless, so we’re tossing that one out. That leaves four teams who combined to go 2-2 in their first game. All three years combined for 4-6 (not counting St. Mary’s play-in game).

If we restrict it to only the #10, #11, and #12 seeds the record is 3-2. Maybe there’s something there. I may have to do more digging. Coincidentally, my only money-line bet this year was on #12 seed Oregon at +160, and that one worked out. But Oregon didn’t rely on 3’s — they were 138th out of 150 during the regular season…

The never-ending quest for an angle continues.

Tourney Wrapup

by A.J. Coltrane

Before the NCAA Tournament I thought it would be fun to get a few people, give them $100 in Monopoly money, and have a blind bid on the tournament teams. The entry fee was 1 “twinkie” per person. The high bidder on the team that won the tournament would receive 3 twinkies, and the 2nd highest bidder would break even and gets his/her twinkie back.

I thought it would be an interesting experiment in game theory — do you put all of your money on a big favorite? There’s a very real possibility that everyone else will want to do that too, so you’d better bid a lot or risk wasting your investment. Alternately you could spread your money around, but what would be the minimum to “claim” each team?

As an added twist, it was possible to bid on “The Field”. “The Field” was defined as every team “leftover” that nobody had expressly bid upon. So — do you bid a lot on The Field, or do you value certain teams enough to put a nominal amount of money on them and hope you don’t get outbid?

The Outcome Of The Bidding:

Team	CW	BD	AJC	AS	Total	Result
Louisville	30	30	12		72	Win Champ
Ohio St	20	10	6	20	56	Elite 8
Georgetown	20	10		20	50	1st Round
Florida	20	10	16		46	Elite 8
Kansas		20	22		42	Sweet 16
The Field	10	10	2	20	42	Final Four
Duke			16	20	36	Elite 8
Indiana		10	12		22	Sweet 16
Miami			2	20	22	Sweet 16
Michigan			2		2	Champ Game
Michigan St			2		2	Sweet 16
Syracuse			2		2	Final Four
Wisconsin			2		2	1st Round
Gonzaga			2		2	Round of 32
Pittsburgh			2		2	1st Round

I [AJC] went with the strategy of trying to get as many teams as possible. I feel like I did pretty well for myself — I “won” Kansas, Indiana, Michigan, Michigan State, Syracuse, Wisconsin, Gonzaga, and Pittsburgh. If Florida, Duke, or Miami won the tournament I’d still break even. Overall it represented two #1 seeds plus a bunch of good quality to back it up. That “quality” wound up being half of the Final Four teams (Michigan and Syracuse.)

But what was I thinking with Pittsburgh? Bleh.. at least they were cheap.

Ultimately my strategy didn’t work out, of course. The #1 overall seed in the tournament was Louisville, and they did indeed win it all. (Though I was feeling pretty smart when Michigan was up 12 in the first half of he championship game.) Louisville was the team that we bid the most upon in total, and there’s definitely a good correlation between where we chose to invest and how wells the teams actually fared. The biggest “overperformers” were Michigan and Syracuse, while Georgetown was the biggest letdown.

Anyhow, I had fun with it. Hopefully somebody will want to try it again next year.

Vegas, Briefly

by A.J. Coltrane

I’m back, and recovered, from Vegas. I can talk like a normal person again. I went 29-19 this year (60.4%), betting almost every game. That figure includes money lines that I won, which raised my “effective percentage” to 63.7%. That’s pretty good, though it’s about on par with how I’ve done each of the last two years. I’d estimate I’m now “effectively” winning about 65% over the last three years.

To avoid TL;DR, here are a few bullet points about what worked, what didn’t, and some other general observations and thoughts.

1. I got in the neighborhood of “paying for the trip” this year. To actually succeed in paying for the vacation I’m going to need to lay off the stuff I don’t have a strong opinion on, and (probably) increase the size of the bets somewhat. I feel comfortable with the idea of increased bet size, though I’m not sure that everyone that I travel with shares that opinion. I may have to make little bets on the “unsure” pile, since I’m not in Vegas *not* to gamble.

2. While watching the games we usually play video poker at the bar. One of our traveling companions was dealt a Royal Flush. Dealt! No “holding” cards.. just lots of hearts. She said she just saw “heart, heart, heart, heart, heart” and held them all. At that point she realized she had a Royal Flush and had won $1000! I’ve seen people hit Royals before, but never someone we traveled with. According to at least one “odds” site, the probability of being dealt a Royal Flush (without a re-deal) is one in 649,740. I’ll need to wait several lifetimes to see it again.

3. The betting public had no idea what to do with VCU all weekend, and *that* turned into absolute no-brainer profit on both games. Here’s why: VCU’s entire offense and defense is/was predicated upon creating turnovers and scoring off of turnovers. In the first game they played Akron. The line had VCU favored by about 8. Akron’s problem is that their star point guard had been arrested (and suspended) for trafficking (5 pounds(!)) of marijuana. For whatever reason, the public didn’t notice, or didn’t react strongly enough — VCU forced 21 turnovers, gave up 5, led 50-25 at the half en route to winning 88-42.

In the next game VCU was a slight underdog against Michigan. Michigan had the lowest turnover percentage in the country this year. Final: Michigan 78, VCU 53. Thank you VCU!

4. I took the Oregon money line at +160 against St Louis. There were a couple of Vegas natives sitting to our right betting the games as well. When I told one I had the Oregon money line he gave me the “You’re smoking crack” hand gesture. Final score Oregon 74, St Louis 57. That one felt pretty good. (Nice guys, actually.)

5. The worst losses are the close losses. Temple started well against Indiana. Temple looked “longer”, quicker, and skilled enough to play with Indiana. The halftime score was Temple by three. The sportsbook published a second-half line of Temple +8.5. I couldn’t believe it — Temple didn’t even have to win, they just had to lose by less than 6 (or win)! It seemed too good to be true, though it also looked really really strange. I paced back and forth a few times from the book to the bar, and eventually asked our bartender what he thought of that bet. He said that if it were him, he couldn’t make that bet fast enough. I went ahead with it…

Indiana trailed by four with 2:56 to go. That means I was “10 points up” on that bet with less than three minutes to go. Indiana outscored Temple 10-0 the rest of the game, and I lost by 1/2 point. Temple did literally everything wrong, and Indiana did everything right. What a crusher.

6. I thought Colorado State would be able to take care of the ball against Louisville and keep it competitive. Colorado State had a low turnover rate this year… Wrong, wrong, wrong. Louisville 82, Colorado State 56.

7. Creighton met their “better” doppleganger in Duke. Creighton had been a great 3-point shooting team all year, but they went 2-19 against Duke and lost 50-66. The line was 6, so that was an easy one to pick. After the tournament ends I’m going to do a post about the heavily 3-dependent teams and what happened to them.

—

Syracuse-Indiana post-game thought: Cody Zeller does not have a strong lower body. It showed against Temple, and it really shouldn’t have come as a surprise that Syracuse swallowed him up. He got his shot blocked about six times against Syracuse. (He finished 4-10 against Temple and 3-10 against Syracuse, though he shot 62.3% during the season.) I think he may struggle at the NBA level, unless he gets a *lot* stronger.

In a not-unrelated note: Kelly Olynyk went 8-22 while losing to Wichita State. (Though he did score 26 points.) Olynyk made 62.9% of his shots during the season.

Bracket Busters and History

By Blaidd Drwg

Is your bracket busted yet? It feels like there are a ton of upsets this year, but if you actually look at the numbers, percentage wise, it isn’t that many. If you don’t count the 8-9 games, which are two supposedly evenly matched teams (even though the 9 seed has historically won something like 54% of these games), higher seeds are 20-8 in the tournament, or have won 71.4% of their games.

This year, it was not a good idea to be a #5 seed as 3 of the 4 teams lost their game with VCU the only team in that seed to survive.
What does this actually mean historically? Well, I took a look at the last 5 NCAA tournament first rounds (including this year), and found that the higher seed (1-7) has won exactly 71.4% of their games. So this year is an exactly average year for upsets. Just for fun, I decided to look at the results by seed for the 5 year period:

Seed	Wins	Losses
1	20	0
2	17	3
3	18	2
4	15	5
5	10	10
6	10	10
7	10	10
8	12	8
9	8	12
10	10	10
11	10	10
12	10	10
13	5	15
14	2	18
15	3	17
16	0	20

What does it all mean? Well basically it means that any games involving seeds 5-12 are a toss-up and you could probably expect to see a 13, 14 or 15 seed knock off a higher seed at least once in the tournament. The key is picking the right upset.

The Cheap Seat Eats Bracket Of Peril (Well, let me have just a little bit of peril?)

by A.J. Coltrane

It’s the Cheap Seat Eats (hosted by ESPN) tournament bracket! Match wits with the CSE writers! It’s free!

Up to three brackets per entry.

Link

Bracket Name: Cheap Seat Eats

Password: TakeMeOut (As in, take me out to the ballgame, note that it’s case-sensitive, and there are no spaces.)

The prize is absolutely nothing, which is what sir Galahad gets in this scene from The Holy Grail:

Narrative And Commentary

by A.J. Coltrane

Ken Pomeroy made some great points in February of 2012 (subscriber only):

…It’s the manufactured stories that attempt to explain the often-unexplainable variability in a team’s performance that I take issue with. Some team salvages its season by going on a late winning-streak and the origins of the streak are explained by a players-only meeting or the team captain stepping up and being a leader, or a renewed emphasis on defense, etc. When in reality, the causes of the change may have been more complicated that anyone could truly understand. (Naturally, this xkcd comic comes to mind.)

Murray State’s loss last week provided one of the clearest such examples of this method of analysis. The general assumption after the loss was that the Racers cracked under the pressure [(1), (2), (3)] of their unbeaten record. Even the coach said so! The thing is, Murray never reached a point during the season where they were better than a 50% proposition to go unbeaten in conference. You play enough games in which you are heavily favored, and you are going to lose eventually. Put more precisely, a team that plays ten games as a 90% favorite is expected to lose once during that span, and the Racers have played a lot of such games this season, including the game against Tennessee State…

and

…There’s lots of unexplained variance in a college basketball game. The Vegas line has long been proven to be the best predictor of outcomes, and while it has the reputation among some of being scarily accurate, the average error in the Vegas line is 8.4 points. And, with all due respect to other prognosticators out there, that’s the best we can do.

Keep in mind that 8.4 points of unpredictable variability is the combination of the variability of the two teams involved in a game. When Duke unexpectedly won at North Carolina last week, was it because Duke played better than usual and UNC played worse than usual, or was it because UNC played better than usual and Duke played much, much better than normal? Or was it because both teams played worse than usual, but UNC just really played badly? I think it’s nearly impossible to disentangle the two. If one team shoots poorer than expected, is it because their form was off or because the defense was better than usual? It is difficult to determine the answers to questions like this without some serious video breakdown…

I love the comic that Pomeroy linked, it’s exactly why I don’t listen much to the talking heads:

What Does That Mean, Really?

by A.J. Coltrane

From left to right: Team Name, Kenpom Rank (minimum, top 50), 3-point percentage and rank, free throw percentage and rank, 3-point attempted quantity and rank. The ranks are relative to all 347 Division I teams.

Then it’s me screwing around with numbers. The “Vol” is the Volatility of the team. It’s 3P% x 3PA%. These are the teams that rely upon 3-pointers for a lot of their offense. If they get hot they can beat anybody, and if they get cold they could lose to anybody. Conversely, a team with a low “Vol” doesn’t take many 3’s and isn’t good at shooting them. A team with a low “Vol” isn’t likely to shoot itself out of, or back into, a game.

The “VR” is that team’s “Vol” rank among the top 150 teams according to Kenpom. (I didn’t care about the Weak Sisters Of The Poor.) “Vol” wound up ranging from 1658 to something like 800. The lowest team on the “top 50 Kenpom list” is NC State at 920. I think it’s telling that the top 50 teams can all shoot at least a little, and with some volume.

The teams are sorted by Q-E, explained below.

Team	Rank	3P%		FT%		3PA%		Vol	VR	Q-E
Cincinnati	37	31.8	266	64.6	305	36.7	75	1167.1	75	-191
Illinois	36	32.8	211	71	116	40.9	28	1341.5	37	-183
Wisconsin	5	34.7	121	62.4	328	39.3	44	1363.7	30	-77
Iowa	34	30.4	304	72.1	77	31	230	942.4	122	-74
Akron	46	34.5	134	66	268	37.3	65	1286.9	45	-69
Wichita St.	35	34	164	68.6	197	35.8	101	1217.2	59	-63
Virginia Commonwealth	22	34.5	129	68.6	199	37.1	69	1280.0	48	-60
Nevada Las Vegas	30	32.7	214	70.4	137	33.8	156	1105.3	89	-58
Denver	42	36.7	53	74.2	34	45.2	3	1658.8	1	-50
Baylor	49	33.8	173	68.5	203	34.6	134	1169.5	73	-39
Iowa St.	38	37.3	42	72.9	61	43	12	1603.9	5	-30
Marquette	24	29.7	317	73.8	41	27.2	298	807.8	140	-19
Arizona	19	36	73	74.1	35	37.7	59	1357.2	33	-14
Oklahoma St.	15	32.4	229	74	38	31.3	219	1014.1	106	-10
Syracuse	9	32.2	240	68.4	205	30.9	235	995.0	110	-5
Mississippi	41	34	163	69.4	168	33.5	161	1139.0	83	-2
Florida	1	38.4	22	68	217	41.4	25	1589.8	6	3
San Diego St.	26	33.5	183	68	218	32.5	190	1088.8	91	7
Louisville	3	32	251	71.1	114	29.9	259	956.8	116	8
Oregon	39	31.3	283	71.8	91	27.8	292	870.1	138	9
New Mexico	28	34.7	120	72.7	66	34.5	136	1197.2	66	16
Belmont	32	38.6	15	72.7	65	40.2	32	1551.7	8	17
Missouri	27	33.5	185	75	23	31.8	205	1065.3	98	20
St. Mary’s	18	38.6	15	72.9	59	38.4	51	1482.2	16	36
Creighton	20	41.8	2	74.8	25	39.4	41	1646.9	3	39
Miami FL	13	35.5	90	65.9	270	34.1	147	1210.6	63	57
Minnesota	17	32.6	219	68.9	186	29	278	945.4	121	59
St. Louis	25	35.7	81	70.1	145	34	150	1213.8	61	69
Connecticut	47	35.9	77	75.1	21	34.2	146	1227.8	56	69
Michigan	10	38.5	20	70.8	122	35.3	106	1359.1	32	86
Georgetown	14	35.7	83	68.1	215	31.8	204	1135.3	84	121
Kansas St.	33	35.8	80	69.2	176	31.9	201	1142.0	81	121
Colorado St.	23	34.1	158	71.7	94	27.8	293	948.0	120	135
Colorado	43	34.5	136	67.3	240	28.4	286	979.8	113	150
Ohio St.	11	36.1	71	70.1	150	30.7	242	1108.3	88	171
Indiana	2	42.6	1	74.8	27	32.9	173	1401.5	25	172
Notre Dame	45	37.4	40	70.1	149	31.2	223	1166.9	76	183
Michigan St.	12	35	110	70.3	139	27.1	300	948.5	119	190
Kansas	8	35.9	75	72.4	68	29.4	271	1055.5	101	196
Duke	6	41.6	3	72.2	76	32	200	1331.2	39	197
North Carolina	31	36.3	66	65.8	276	29.5	269	1070.9	95	203
UCLA	48	34.5	130	71.6	97	23.3	335	803.9	141	205
Pittsburgh	7	35	108	66.3	260	25.6	317	896.0	131	209
Gonzaga	4	37.9	33	71.9	87	30.5	246	1156.0	77	213
Virginia	16	40.4	6	71.5	100	31.1	227	1256.4	53	221
Kentucky	29	36.6	58	64.8	301	28.9	282	1057.7	99	224
Memphis	40	36.5	59	67.7	230	27.6	296	1007.4	109	237
Middle Tennessee	21	38.5	19	69.1	181	26.8	303	1031.8	102	284
North Carolina St.	44	40.2	7	68.2	210	22.9	339	920.6	126	332

Q-E is pretty self-explanatory: It’s 3PA% rank minus 3P% rank. (The Excel columns were “Q” and “E”. Yeah, I’m a big thinker.) As an example, Cincinnati is the first team listed with a -191. They’re 75th in 3-pointers attempted, and 266th in accuracy. 75 – 266 = -191. My suspicion is that the teams with large negative numbers are going to shoot themselves out of the tournament at some point. Offhand, I’d also guess that those teams lack a guy who can create his own shot, or slash and kick, leading to some desperation heaves late in the shot clock. I’ve seen Cincinnati. It doesn’t feel like an unrealistic set of assumptions. Watching them the other night I kept saying “They look really disjointed…”

On the other hand, the strongly positive teams may benefit from shooting more 3’s. That might bode well if they happen to fall behind and are forced to try to catch up quickly.

It’s interesting to me that there are only two “top-50” teams at worse than -100 “Vol”, but twenty-two teams at better than +100. A large percentage of the top teams are relatively judicious with their 3-point attempts. My guess is that the unwashed masses are shooting a lot of 3’s that they shouldn’t be, but they’ve got to create offense somehow, so that’s how it goes.

If I was smarter I’d be able to do a regression of 3P% vs FT%, I’d expect that there’s some positive correlation to it. At the very least, it illustrates that Cincinnati can’t shoot free throws, either. Poor free throw shooting may do them in, too.

Final note. I like Florida, but their 6th-ranked “Vol” score kinda gives me the heebies.

Burn After Reading mini-spoiler below.

The New NBA CBA

by A.J. Coltrane

Zack Lowe of Grantland/ESPN on the impact on the trade deadline of the new NBA Collective Bargaining Agreement.

I don’t think he has a high opinion of Dwight Howard:

People seem to forget this, but we also had three massive deals between August and late January, two of which — for Rudy Gay and James Harden, respectively — were motivated at least in part by the new CBA. (The third was the Dwight Howard deal, motivated by candy and whatever else motivates Howard’s “camp.”)