This website was originally devoted to a mathematically objective method of ranking Kentucky high school soccer teams. The rankings were started in 2010 on a trial basis for girls high school soccer. Once parameters were tweaked to reach a consistent 75% game prediction rate, the results were sent to a wider audience for review. Posted on this website are the 2011 girls rankings. Boys and girls rankings are included for 2012, and 2013.


     Soccer is a growing sport in Kentucky, and the skill-base is not evenly distributed throughout the state. The rankings give teams a chance to see how they compare, and who to select for future opponents for a quality match. It also generates more and better discussion about teams across the state among fans. All of these benefits will promote growth, and broaden competition in soccer across Kentucky, giving more kids the chance to enjoy the game. Lastly, it is a lot of fun to track the progress of your favorite team.

How are the rankings calculated?

    The system is based on the assumption that a team’s performance on a given day will vary with a normal distribution. The mean of these performances will more or less be centered around that team’s ability level over the long term. Basically, teams will have good days and bad days, fair results and unfair results, but given enough data points, the average of all performances should be a close representation of their quality.
This ranking system is purely mathematical, using only game scores and home field advantage as inputs to rank all Kentucky High School soccer teams. The system is based on the relative strengths of the opposing teams in each match. The chance of winning a game against a certain opponent depends on the strength of your team and the strength of your opponent. The system compares these chances with the actual result, and uses the difference to adjust each team’s ranking heading into the next game. 
Of course, a team’s strength depends on your ranking and the ranking of the opponent, which depended upon the rankings of all of their opponents, and the results of their games, which depended on, the rankings of their opponents and the results of their games, which… well you get the idea. Because of this, it takes a many iterations to get it close.
The most important factor is winning, followed by the score, followed by home field advantage. It is possible to win a game and decrease your scaled rating, if your opponent was much weaker. Conversely, it is possible to lose a game and increase your scaled rating.
Historically, the system has performed with an 80-88% prediction accuracy, measured over 500 game stretches. Accuracy generally increases throughout each season, as shown on the top pages for each respective sport.
The rankings only include in-state Kentucky  and Ohio High School games.

Looking Forward

  With the success of the system in 2013, as measured by a game prediction rate in excess of 88% over the last 500 games of the season, a system for ranking Ohio high school soccer is now under development.  Please read, enjoy, comment and criticize. Our philosophy is that criticism can only make the rankings better, we never take it personally!
 Starting in 2015, the Maher Rankings are now making their first foray into Kentucky High School Basketball Rankings. Please check them out, and let us know what you think, by leaving a comment!

61 thoughts on “About

  1. Do you have a means of calculating total strength of schedule? I coach a high school boys team and have been using your system more than the coaches poll over the past several years as was intended, to assess strength of each opponent. However, this year, we are in a down year as I graduated 12 seniors and our girls team is doing very well. This is causing some of the normal comparisons girls and boys teams from the same school make, and while I assure them it will always be like comparing apples to oranges, I did compare both teams schedules based on your rankings. If you take your teams ranking and do a plus x however many spots an opponent is above you or minus x for however many spots an opponent is above you, you get a general number. I classified all un ranked teams as 151 and for example, the boys total composite schedule strength worked out to roughly a + 257 while the girls worked out to a – 309. I assume the maximum would be 149 x however many games? I’m not sure how to quantify what would be the strongest or weakest schedule but if the worst team played the best team every time over 10 games I assume it would be + 1490 or – 1490? Just thought it was worth a mention? Maybe another feature to add when considering schedule strength?

  2. Just an FYI…on your ribbon on the main page you have a typo as it scrolls… It says “About the Marher Rankings” noticed only because my best friends last name is Maher and I noticed the extra “R” in front of the “H” .

    have a great day….

  3. Big fan of your ranking system. Thanks for doing all the hard work. A common critique seems to be starting rating of teams based on last year’s finish. I know collecting outside data is impractical. But is there another way to handicap or adjust the start? Would starting with 80% of last year’s ending value instead of 100% bring everyone a little closer to start and then allow separation going forward? Just a thought. Keep up the good work! Its much appreciated.

  4. Could you please update East Jessamine’s record? we actually 6-8 (not 3-8 as it’s reflected). It may or may not make a difference in our rankings, but it may and would be greatly appreciated.

    Where do you pull the teams results from? KHSAA?

    thank you for your assistance


    • I think two things are going on here: 1)The cutoff for last week’s post was 9/21, so your win against Somerset is not yet reflected in the rankings, and 2)we don’t yet count out of state games, and you have two wins against Indiana teams. This is assuming you are the boys coach?

  5. Olentangy Liberty is listed as 16-1-2. Their record is 18-1-2.

    I have read that Out of State games do not count.

    Liberty is 1-0-1 in OoS games, with a win over Trinity (KY) and a tie vs St. Xavier (KY).

    Their In State record is 17-1-1.

    Also, if Out of State results do not count, why is St. Ignatius listed as 18-2-1 when they have are 2-0-0 Out of State? They should be 16-2-1, no?

    • I am afraid I don’t know the answer to this question. I don’t track individual players stats – biting off a little more than I can chew for now. Though I must admit, I have always wanted to “weight” a player’s goals based on the quality of their opponent – much like this system rewards each win or loss, based on the quality of the opponent. And then calculate the true, best goal scorer in the state. The leading goal scorer may only be that because their team played in a lot of 10-0 games, it is hard to tell. I believe Nate Silver does this for FIFA players, and I don’t think it would be too hard, at least in KY where many teams post individual stats. I don’t know if Ohio does that. Just another idea to add to the list though.

      Let me know if I misinterpreted your question.

  6. I disagree with the biggest upset of the week. North Hardin (ranked 113 defeated Central Hardin ranked 33. Seems a bit more significant to me.

    • The Upset of the Week in the last posts were for games from the 10/4 thru 10/10. The game you mention was on 10/15. You are right though, I will be shocked if there is a bigger upset than that, that’s huge!

  7. given one spot notwithstanding, how do you rank a team ahead of another with the same scaled ranking and twice the SOS number….. They are percentage points different. Also how does a team get an SOS of 22 playing 4 teams in the top 5, 5 in the top 10 and 12 in the top 20…. Who did these other people play?

    • If one team is ranked ahead of another and they have the same scaled rating, it must have come down to the third decimal point, which we normally don’t report. Essentially, this is “in the noise” and the two teams are tied in the rankings. For the next question, first, 22 is a pretty good SOS, there are over 500 teams in the state. However, what you are describing could happen if your team played a couple of additional games against teams ranked very low. Let us know the team, we can take a look into it.

    • Their game against Campbell County (KY) was out of state and therefore wouldn’t be counted. Thanks for keeping the eyes peeled though, I’m sure their are still some missing scores, keep them coming!

  8. highlands defeats nda, ncc, owens cath, w jess, only ky team to beat S.H. ties MND that defeated common opponents in top 10 and ranked ahead of all ky teams but 1. Has 10 in strength of sched, better record than mentioned teams. but is ranked behind all mentioned teams. only losses are to 2 top ranked teams

    • Actually, I tend to agree somewhat with the Highlands complaints, at least from the common sense perspective. Highlands resume this year is flat out impressive. Let’s compare it to Sacred Heart’s at the time of the complaints, about 10/1. The numbers below represent how each team has done against Ohio/Kentucky combined competition.

      Sacred Heart Highlands
      vs. Top 20 3-0-0 1-1-1
      vs. Top 50 4-2-1 6-2-1
      vs. Top 70 8-2-1 6-2-1

      Even though Highlands beat Sacred Heart, you’d probably give a slight nod to Sacred Heart for the complete body or work. However, Sacred Heart was #2 at the time, and Highlands was #10 by the numbers – probably a little harsh on Highlands. To analyze this perfectly, one would want to do this for all of the teams mentioned, quite a time consuming process, and by the way, what the mathematical system attempts to do.

      I suspect at the root of this issue is going to be the preseason estimates. The table below shows where each of the teams started the season. The preseason estimates are based on historical performance, but predominantly where they finished the prior season. The system continuously corrects for discrepancy between current rank and each individual performance, but for a team that ended last year at #71, and is now arguably a top 5 team, it takes a while to adjust. Such a change in ranking is highly unusual. First of all, how have they done it?!

      Rank Team Scaled Rating
      3 Sacred Heart 16.87
      5 Owensboro Catholic 16.84
      18 West Jessamine 16.62
      19 Notre Dame 16.60
      46 Newport Central Catholic 16.15
      71 Highlands 15.73

      One final point to make. This is an issue that we have had historically. The algorithmic changes we made this year, as detailed in the Preseason post, were made to improve this bias toward historical results, and the response speed to radical year to year changes. This being the first complaint we’ve received along these lines, I believe we’ve made an improvement. There’s always a trade off. The trade off in this case is that teams playing weaker schedules can also move up the rankings more quickly. Until now, we thought that we had perhaps over corrected, allowing teams that had relatively weak schedules move up a little too high. With this case noted, perhaps not! Perhaps we’ve struck the proper balance. In point of fact, once you start talking about squeezing from say 87% accuracy to 90%, or alternatively arguing about 3-4 ranking places for an individual team out of 200 teams, you are talking about balancing of trade offs – you can’t get it perfect. There is no perfect! Anyway, we appreciate the feedback. It always helps to make us better!

    • Perhaps. Here are the questions we ask when considering whether or not to rank a new group of teams:
      1) Is anyone else already doing it, and doing it well?
      2) Are the scores readily available?
      3) Is it contiguous to our Ohio/Kentucky epicenter?
      The answer to 3 is yes, maybe you can help answer 1 and 2?

      • The answer to 1 is definitely no as far as I can tell. The answer to #2 is “kind of”. does report many statewide scores but sometimes there are ones that fall through the cracks. Very similar to the way that scores are reported for Ohio soccer on, kind of hit or miss at times.

  9. I thought this might be credible but when I see you select Medina as the #1Ohio you lost all credibility. They just Austin a scrimmage 6-1to Ignatius and we’re over match the entire game. Medina will be lucky to win their conference. The graduation of their seniors last year really hurt this team.

    • We don’t select teams as #1 or #anything. We run a mathematical algorithm to calculate the rankings based on scores and home field advantage. The system is mathematically sound, proven and credible – not perfect. The preseason rankings are based on historical results, since we can’t possibly chase after info like the loss of graduating seniors for a couple thousand teams. After just one game, the time at which you made your post, a team’s ranking would still be heavily biased toward historical results. After week two Medina has dropped to #7 in Ohio. The system will continue to hone in on each team’s true ranking as the season progresses.

  10. For the 9/9/17 Boys Rankings under the “Games of the Week”, I noticed that all of your locations for the Kentucky games are incorrect. For example, for the Lexington Catholic vs. Paul Laurence Dunbar game, you have that the game is being played at Lexington Catholic, but it will be played at Dunbar. The other KY games are all incorrect as well. I can’t speak for Indiana and Ohio games.

    I don’t think anyone goes to your website as a source for the time and location of a game. However, I’d for this to be considered a factor in your process and the wrong team be considered as the home team.

    Thank you for what you do. I sure it takes a lot of work, but soccer fans really love the discourse that ensues because of the rankings!

  11. It looks like you have Covington Catholic’s record as 8-7-4. They should be 8-5-4. Losses at Summit Country Day (Cincinnati), Home to Anderson(Cincinnati) and East Central (Indiana) and neutral site (Louisville Trinity) losses to Columbus St. Frances DeSales and Cincinnati Moeller.

    • Apologies. ossca has the date of the DeSales game reported incorrectly, thus fooling our de-duplication algorithm and double counting the loss. Also, khsaa reports the 9/23 loss as being against McNicholas, not Moeller. Again, this fools the de-duplication algorithm, and you get stuck with losses against both. has the latter score reported correctly against Moeller, but does not indicate home field. Do you know where this game was played?

      If not, we will assume a neutral field and have it updated for this week’s post.

  12. CVCA has 2 Varsity teams. You have a mix of their top team and their 2nd team. Their top team is Varsity Blue. Their second team is Varsity White. They play completely different schedules and one is much harder than the other.

  13. I see one of the upsets of the week in Ohio is Triway over Cuyahoga Valley Christian Academy. Triway beat their “B” team. I am not certain, but it appears their B team plays their conference games. This may be having an impact on the rankings involved between their A and B teams. Thank you for work on the rankings

    • Thank you. We caught the CVCA B team trap for their prior games, but not for this one. It will be corrected this week.

  14. Covington Catholic went to penalties with Lexington Catholic on 8/18/18 in Colonels Cup Final. Regular Time ended 0-0. Lexington Catholic won in penalties. I believe the score should be recorded as 1-0 not 3-1.

  15. I think you are getting caught in the CVCA B trap again. They are currently 7-3 and not 9-2. Nothing to for the program to be disappointed with for they lost to a terrific teem in Illinois, while their B team is 2-0.

  16. Curious if you cap any reduction to a team’s ranking in blowout wins? For instance our girls team had a game against a very low ranked school, where the difference in ratings points suggested a 13-point win would be expected. It was 3-0 after 5 and the girls stopped trying to score at 7-0 (and the girl that scored #7 got an earful as it wasn’t our coach’s desire to humiliate a team in the early days of establishing a program. Would this be considered an ‘under-performing’ result or does the model account for sportsmanship decisions like this?

    • There is no cap, however the model does what you are looking for by heavily (un)weighting a game that is won by such a huge favorite. In your example, a 13 goal win would gain the favorite 0.0016 scaled rating points while costing the losing team the same. The 7 goal win would cost the winning team 0.06 scaled rating points, while gaining the losing team the same. The model reverses things, of course, if the underdog wins. Then the game gets weighted really highly. The model also starts to weight things more highly if the underdog can manage to keep things competitive, say in the 2-3 goal ball park.

      In any case, I should give you a sense of how little 0.06 scaled rating points is, and why this isn’t worth it. You can look at it two ways. 0.06 scaled rating points is loosely equivalent to 0.06 goals, so a team you used to be tied with in the rankings, now you are a 0.06 goal underdog. Another way to look at it is this: if two teams were separated by 0.06 scaled rating points, the higher ranked team would have a 50.8% chance of winning.

      One final point to make is that I would assume most coaches would make a similar choice in that situation, so that the loss of the 0.06 points in a relative sense would cost you nothing.

      Really, the model incents top teams from avoiding games like this where ever possible. This is as it should be since these games really don’t do anyone any good.

  17. Thanks for the detailed response, seems very fair. And agree we’d LOVE to avoid games like this, but it’s a league game so not possible…

  18. Thanks for the detailed response, seems very fair. And agree we’d LOVE to avoid games like this, but it’s a league game so not possible… absolutely love the site, by the way. It’s a great tool, especially this time of year when matchups range further afield!

Leave a comment

Your email address will not be published. Required fields are marked *