STEELE versus LUNARDI

BRACKETOLOGY COMPARISON

    Three weeks ago when I decided to try my hand at "bracketology", it was primarily a reaction to media "experts" expounding on which college basketball teams deserved invitations to the NCAA basketball tournament.

    With winter's grip still on the Midwest, golf was not an option; and with no work, I had plenty of time to research the relative merits of the teams with any chance of making the field of 68. To my way of thinking, anyone with too much time on their hands could probably come up with a "mock" bracket that would compare favorably with those of the talking heads and the tournament selection committee.

    First, I compiled a list of potential tournament teams. Then, using national rankings, public rating percentage index and available strengths of schedule, I began sorting candidates according to their "power rating", the sum of RPI and SOS. Of course the ratings changed almost daily; but as the regular season wound down, the list of candidates for at-large bids began to take shape.

    Two weeks before Selection Sunday, I constructed my first mock bracket. I remodeled the bracket over the next two weeks, ultimately building the final version published here last Sunday. As previously posted, my final bracket predictions compared fairly well with the official NCAA bracket: I correctly picked 65 of the 68 teams invited to the tournament; 46 of which were on or within one line of the committee's seeds; 19 were placed into the same regions as the committee's; disregarding regional placement, seven pairings in the Thursday's and Friday's games are identical to those of the committee. Based upon a scoring system (see "THE NCAA SELECTION PROCESS-BRACKETOLOGY SYLLABUS" posted February 27) that awards one point for each team invited, one point for each correct seed, two points for each correct region and three points for each correct pairing, I scored 170 points.

    By comparison, Joe Lunardi scored 180 points. He, too, correctly named 65 teams, 55 of which were seeded on or within one line of the official bracket seeds. He had 21 selections in the same regions in which his picks will play; and he had six pairings that mirror scheduled games.

    As to the three teams we did not have in our brackets, Lunardi and I both had Colorado and Virginia Tech in the final field. He picked St. Mary's to go, while I had Boston College "in". Neither of us had Virginia Commonwealth. He missed UAB and Georgia, which I had "in"; and he put Penn State and Southern Cal in the field, but I didn't.

    To further compare our pre-tournament choices, I've gone a couple of steps further: Lunardi had 31 teams on the same line as the official bracket, while I had 28. Deducting a point for the difference between actual and predicted seeds (Seeding Differential), Lunardi scored a 46; and I missed the lines by a total of 75. He missed only one team (Utah State) by as much as four lines; but I missed seven teams by four lines. (I attribute this statistic to the fact the weather turned nice enough to play golf last Friday, Saturday and Sunday; and I didn't spend much time refining the 8-12 lines.)

    To my credit, I correctly predicted all the #1 Seeds and had them going to the regions where the committee sent them. Lunardi had Notre Dame instead of Duke on the first line, with only Kansas and Ohio State in the correct regions. Ten of the first 16 seeds (1-4) in my projected bracket were identical to the official seeds. My Seeding Differential (SD) for that group totaled seven points. Lunardi hit eight of those teams on the nose, missing the others by just one line for an SD of eight points.

    Although my mock bracket lacks some of the refinement of Lunardi's, I am rather pleased with the results. Perhaps my time, effort and research will pay off with a winning entry among the five submittals in the three pools I entered this year.

    If not, and if I'm still breathing good air this time next year, I will have plenty of spreadsheets ready for "Build-a-Bracket 2012".