Machine Division: Through the Friday games, according to the NCAA's RPI and taking into account game locations, there were three upsets: Vandy over Ohio State, Wake Forest over Georgetown, and Clemson over Alabama. In other words, the NCAA's RPI already has missed 3 games. According to the NCAA's Non-Conference ARPI, there only were two upsets: Vandy over Ohio State and Clemson over Alabama. So, it has missed 2 games. According to the 5 Iteration ARPI and Massey, the only upset was Vandy over Ohio State. They're both doing better than the NCAA, each having missed 1 game. But, there's an interesting note about game locations. Past precedent has been that, for games between unseeded teams, the team with the better RPI (that has bid and has a qualifying field) always gets to host the first round game. The Committee appears to have changed that this year, assigning the Hofstra v Auburn game to Auburn (Hofstra had the better ARPI), the Georgetown v Wake Forest game to Wake Forest (Georgetown had the better ARPI), and the Alabama v Clemson game to Clemson (Alabama had the better ARPI). Both Hofstra and Georgetown bid for their games, I don't know about Alabama. I don't know whether Hofstra, Georgetown, and Alabama have qualifying fields. The change in game location to Auburn, for the Hofstra v Auburn game, changed the likely winner from Hofstra to Auburn according to the ARPI and Massey (but not according to the NCARPI and the 5 Iteration ARPI). The change in game location to Wake Forest, for the Georgetown v Wake Forest game, changed the likely winner from Georgetown to Wake Forest according to the ARPI and NCARPI (but not according to the Iteration 5 ARPI and Massey). And the change in game location to Clemson, for the Alabama v Clemson game, changed the likely winner from Alabama to Clemson according to the ARPI and NCARPI (but not according to the Iteration 5 ARPI and Massey). The bottom line is (although there's ambiguity) that the Committee, in its game location designations and from a statistical perspective, may have made itself a factor in turning these games into possible upsets.
Thanks, this is a good explanation. So bottom bottom line is the RPI is terrible and needs to be changed and then we can look forward to the day where we don't even have to play any games we can just run computer models. ;-)
Actually, his analysis shows that the bottom line is that the RPI is better than last night's results reflect. His statistics show, that had those 3 matches been played at the home field of the higher RPI opponent, it is likely that the higher RPI team would have won. In those examples, the lower RPI team was given home field advantage, and beat the higher RPI team. I've never thought "home field" advantage meant a whole lot ....but this makes one go "hmmmm"...
Happily, the computer models can't predict upsets, so we can't do it. And, in soccer there are lots of upsets -- about 25% of games that statistically should be wins turn out to be losses or ties. It's part of what makes soccer the beautiful, and sometimes agonizing, game.
First Round scoring is complete. Human Division 29 WENDY’S 28 DMTHOMAS49 28 SOCCERHUNTER 28 CRANE65 28 EUROPE FB FAN 28 OLELALIGA 28 McSKILLS 27 NORFOLK 27 MPR2477 27 JUSTDOIT 27 INGOLDSBY 27 GOT JUKES? 27 HEADSPUN 27 PLAYSIMPLE 27 TIME&SPACE 26 MERLIN13 26 WALSCD 26 ZIGGY1010 25 GLOVE STINKS 25 KURT KLINE 25 CASTATE 24 BIGWEST 24 KICKITHARD 22 RTD! Wendy's sits alone with the top picks -only missing three out of the 32. But essentially at this point most everyone is still theoretically in the running as a five or even a seven point spread most of the time does not decide the eventual winner. There was a different set of leaders before the results of the 0-0 tie between Butler and Northwestern was decided by a shoot-out in Northwestern's favor. The Butler pickers all fell down a slot. (It was a classic soccer game where Butler led on the stats for shots on goal, corner kicks (and fouls) but NW got the win with more saves AND a 3-2 advantage goals from the spot.) Over in the Machine division, CP's Interation 5 Adjusted RPI and Massey are leading the pack for the moment.
The second round scores are in. On the human division side, the top three had now separated themselves from the pack a little. JUSTDOIT, MERLIN13, and EUROPEAN FUTBOL FAN are 1,2,3. The separators were: --- the NC State - Princeton game (Decided on PKs in very dramatic fashion in that NCSU was up 4-3 with either of the next two shooters able to take if for the Wolfpack, but both missed and Princeton advanced 5-4 in the 7th PK round.) 14 of the contestants lost three points to the 6 who picked Princeton. ---the Washington State - Tennessee game where Ziggy and Glove Stinks put themselves in 4th and 5th place by being two of the only three that called Wazoo to go this far. ---the Notre Dame - TAMU game where 8 contestants gained three point over the 24 who picked the Aggies. ---and the Virgina - Pepperdine match where 5 loyal Pepperdine fans mpost three points to the Virginia pickers. However, there are still four rounds to go with lots of surprises remaining, and 3/4 of the contestantes have a realistic chance at this point of ending at or near the top of the standings. 63 JUSTDOIT 62 MERLIN13 61 EUROPE FB FAN 59 ZIGGY1010 58 CRANE65 58 GLOVE STINKS 58 KURT KLINE 58 McSKILLS 58 OLELALIGA 58 SOCCERHUNTER 57 GOT JUKES? 57 INGOLDSBY 57 KICKITHARD 57 NORFOLK 57 TIME&SPACE 56 WALSCD 56 WENDY’S 55 CASTATE 55 DMTHOMAS49 54 MPR2477 51 BIGWEST 51 HEADSPUN 51 PLAYSIMPLE 49 RTD! Over on the Machine/Poll side, things tightened and Bennet Rank picked up three points on last weeks leaders by picking Santa Clara. 60 BENNETT RANK 59 ITRATN 5 ADJ RPI 59 MASSEY 59 UNTD SOC COACH 57 ADJ NON-CONF RPI 57 ADJUSTED RPI 54 TOP DRAWER RANK As a final note, Bennett Rank came in dead last in 2016 )and currently is in first and in the Human division, Walscd who is in the middle of the pack now, almost won the contest last year.
Just did the little calculation. As always, the trend is starting to show that the Machine entries will average better than the Human entries, but that the top entries from the Human division will be better than the highest Machine division score. As of the second round, Humans have the top scores of 63, 62, and 61 while the Machine top scorer is 60. But the average Machine score is 57.85 while the average Human score is 56.5.
Human Division The third round certainly cleaned out the field at the bottom end. Those of us who voted wit our hearts got clobbered and those of you who stuck with convention are looking pretty good. I went from a fairly respectable top half position only 5 points from the leader to out of the running entirely. And poor Ziggy who had the courage to pick a contrarian bracket got totally cleaned out. But it was Kurt Kline who had the biggest jump. He leap frogged over everyone to the top spot on the back of getting 6 of the elite 8 correct. After the fourth round, the top players will all be in eyeball range. If past experience is a guide there'll be only a half dozen or so in the running. It will be interesting to see how it pans out because there are * contestants with the exact same final four. And of those 8, 4 chose Sanford, and 2 each chose UCLA or Duke to take it all. What this means is that one or two points will likely separate the winner from the runner-up. In other words, if the winner comes from that group of 8, then the first round with only a single point per game will be the deciding factor. 91 KURT KLINE 90 JUSTDOIT 89 WALSCD 88 EUROPE FB FAN 85 CRANE65 85 McSKILLS 85 OLELALIGA 84 KICKITHARD 84 NORFOLK 84 TIME&SPACE 83 WENDY’S 82 CASTATE 82 DMTHOMAS49 81 MPR2477 79 GLOVE STINKS 78 GOT JUKES? 78 INGOLDSBY 73 SOCCERHUNTER 72 HEADSPUN 72 PLAYSIMPLE 71 MERLIN13 70 RTD! 66 BIGWEST 56 ZIGGY1010 Over in the Machine division. Things remain pretty stable with Bennett still on top. But I'll venture a guess looking at the picks to come that the eventual winner will be the Adjusted Non-Conference RPI. 90 BENNETT RANK 83 ITRATN 5 ADJ RPI 83 MASSEY 83 UNTD SOC COACH 81 ADJ NON-CONF RPI 81 ADJUSTED RPI 78 TOP DRAWER RANK Cheers!
After the UNC-Princeton game I thought the ANCRPI might come out on top, as it picked Princeton. Interesting because statistically, it's the least reliable of the RPI-related systems and Massey, by quite a bit. TopDrawer bringing up the rear of the machine and poll group is no surprise.
I may be in the Human division, but I have been known to have some machine-like qualities. And how did Ziggy lose points? From 59 to 56?
I agree about TopDrawer. It's a sensationalistic "publication" that is a step above the supermarket tabloids. It's total garbage that appeals to parents that yearn to see their kids mentioned in it. TopDrawer and Holmes do have something in common, though. Both are best ignored.
I have a more serious thought about the Adjusted Non-Conference RPI. I have thought, and some evidence suggests the Committee thinks, that the ANCRPI is a very weak tool when it comes to NCAA Tournament seeds and at large selections. Princeton may be one case, however, where reference to the ANCRPI in the decision-making process is important. That is a case in which a mid-major team has a really outstanding non-conference record and also does well within its conference. Doing well within a mid-major isn't going to be terribly impressive, but coupling that with a very impressive non-conference season can identify the team as one that should receive consideration as a serious contender. The ANCRPI would be one of the indicators of a very impressive non-conference season. Princeton's ANCRPI rank was #8.
OK I've spent 30 minutes investigating such an error and have determined what it is..... Some how on the spreadsheet the scores were correct on the second round, but when I added the third round scores in and re-calculated, the 3 points for the UCLA win in the second round got dropped from everyone's score. So below are the correct third round scores, with everyone 3 point higher. Ziggy had 59 in the second round and has not been lowered!. (The rank order is unchanged.) Thanks Kurt for your sharp eye! 94 KURT KLINE 93 JUSTDOIT 92 WALSCD 91 EUROPE FB FAN 88 OLELALIGA 88 McSKILLS 88 CRANE65 87 TIME&SPACE 87 NORFOLK 87 KICKITHARD 86 WENDY’S 85 DMTHOMAS49 85 CASTATE 84 MPR2477 82 GLOVE STINKS 81 INGOLDSBY 81 GOT JUKES? 76 SOCCERHUNTER 75 PLAYSIMPLE 75 HEADSPUN 74 MERLIN13 73 RTD! 69 BIGWEST 59 ZIGGY1010 Also in reviewing all of the score entries, I caught another error. In the Machine Division, the Adjusted Non Conference RPI was missing 6 points for the Princeton win, bumping that calculation up to second place. Sorry. Late night unexplained error which probably has something to do with human error.
Final Four (Comments later) 126 KURT KLINE 120 CRANE65 120 OLELALIGA 119 NORFOLK 118 WENDY’S 117 JUSTDOIT 113 INGOLDSBY 109 CASTATE 109 DMTHOMAS49 108 WALSCD 107 EUROPE FB FAN 107 HEADSPUN 107 PLAYSIMPLE 105 GOT JUKES? 103 KICKITHARD 103 TIME&SPACE 100 MPR2477 97 RTD! 90 GLOVE STINKS 85 BIGWEST 84 SOCCERHUNTER 74 MERLIN13 59 ZIGGY1010 Machine Division 119 ADJ NON-CONF RPI 107 ITRATN 5 ADJ RPI 107 MASSEY 107 UNTD SOC COACH 106 BENNETT RANK 105 ADJUSTED RPI 94 TOP DRAWER RANK
Go UCLA! I actually think Stanford is the better team and beats UCLA 8/10. I am hoping for one of the two. Still have to get past notable competition. Will be in Florida with the kiddie soccer. Hoping to see the final
I am so sorry.... A bad cut and past job before I ordered the data. (Tried to get to you last night but the hotel wifi was not secure, and it turns out that Big Soccer will not let one log on in that situation. As you may have surmised you are tied for second place!
Correction to scores and comment. I've added McSkillz back into the score sheet with apologies. Kurt Kline is on top, but is vulnerable. He is among a minority who have picked Duke to win. If they do he may win it all, but if Stanford wins it will be decided among a half dozen. Olelaliga and McSkillz are in the group of second place finishers who if UCLA takes it all ill tid for the championship . Norfolk and Crane65 will tie for the top spot of Stanford beats Duke in the final. However, if South Carolina beats Stanford, other combinations will have to be considered. 126 KURT KLINE 120 CRANE65 120 McSKILLS 120 OLELALIGA 119 NORFOLK 118 WENDY’S 117 JUSTDOIT 113 INGOLDSBY 109 CASTATE 109 DMTHOMAS49 108 WALSCD 107 EUROPE FB FAN 107 HEADSPUN 107 PLAYSIMPLE 105 GOT JUKES? 103 KICKITHARD 103 TIME&SPACE 100 MPR2477 97 RTD! 90 GLOVE STINKS 85 BIGWEST 84 SOCCERHUNTER 74 MERLIN13 59 ZIGGY1010
Thanks for doing this! This was my first time participating in NCAA and strangely this prediction thing has sucked me in even more but that's what makes these contests fun I suppose. At this point, I have no idea what the results are going to be but obviously I'm hoping UCLA luck out and win over Duke who are on paper slated to be the favorites. Unfortunately I'll be away on business in this whole week without a television so I won't be able to follow the matches live but best of luck to the remaining teams and I'm glad I got to go to most of UCLA's games this season and post-season before they headed off to Florida.
"On paper," Duke wins over UCLA and Stanford wins over South Carolina and Stanford wins over Duke. But, this has been a year of some pretty good upsets.
OK It's down to the wire! The UCLA fans who voted with their heart are looking good! McSkillz and Olelaliga have jumped into a tie for first place (and the other two UCLA fans jumped up some notches too.) Who would have thunked that only four lonely fans would have called that match correctly? So on the human side, the prognostication at this point distribution going forward is much simplified as only 2 (the current front runners) have picked the Bruins to take it all. If the Bruins win then McSkillz and Olelaliga get to celebrate and have the prognosticators bragging rights until next year at this time. If Stanford prevails, then Wendy's (another of the UCLA pickers, but who switched to Stanford for the last match) will win it all followed by Crane65 who will be joined by McSkills and Olelaliga. We shall see..... The overall trend continues with the Machine division maintaining a lead over the humans in average score (114.4 vs 111.5) but the extremes of the humans are greater whose top and bottom scores are 136 and 59 while the Machines have a range of 127 to 102. 136 McSKILLS 136 OLELALIGA 134 KURT KLINE 134 WENDY’S 128 CRANE65 127 NORFOLK 125 JUSTDOIT 123 HEADSPUN 123 PLAYSIMPLE 121 INGOLDSBY 117 CASTATE 117 DMTHOMAS49 115 EUROPE FB FAN 111 KICKITHARD 111 TIME&SPACE 108 WALSCD 105 GOT JUKES? 105 RTD! 100 MPR2477 98 GLOVE STINKS 85 BIGWEST 84 SOCCERHUNTER 74 MERLIN13 59 ZIGGY1010 Over in the Machine division, things are holding steady with the adjusted Non Conference RPI entry flying high with an insurmountable lead. The fight for second will be a tie between two or three entries. If Stanford wins, then Massey and the United Soccer Coaches Poll will be second 12 points behind Non Conference RPI, but if UCLA wins, then Iteration 5 RPI will join Massey and USC Poll in a three-way tie for second place. 127 ADJ NON-CONF RPI 115 ITRATN 5 ADJ RPI 115 MASSEY 115 UNTD SOC COACH 114 BENNETT RANK 113 ADJUSTED RPI 102 TOP DRAWER RANK