I created a simulation of the OHSAA State Tournament

Back earlier in the year, I was trying to find a way to calculate win percentage before a game is played. And after a lot of deliberation, I was able to find what I thought was a good way to figure out the win percentage.

I did this by using myhockeyrankings.com.

If you don’t know, myhockeyrankings.com ranks thousands of youth hockey teams based on the results of their games that coaches put into the website. They rank the teams on a scale of 99.99 through whatever the lowest number is, which can be as low as you can think. Normally, in high school, it doesn’t go below 70.

Through a lot of trial and error, I was able to find a good way to figure out the win percentage between two high school teams in the OHSAA by using these numbers.

I tried this out for a weekend, it was MLK Day weekend so everyone was in tournaments. Using this formula, out of 53 games that I looked at, I got around 65% of the games correctly. However, this counts games that could end in ties, as my formula doesn’t account for ties, it only counts for if someone wins.

If we remove the ties, it gives us 49 games instead of 53, I got 36 of the games correct, which gives us a percentage of around 73%, almost a 10% greater difference.

So just using that weekend, I figured that it would be a good test to try and predict who would win games.

Using Excel, I created the playoff bracket for the OHSAA. This didn’t take too long to create, a lot of copy and pasting but other than that not a lengthy process.

The last set of rankings that I used was the rankings from February 12, which was before only a handful of first round wildcard games in the Northern part of the state.

To do a simulation, I used a random number generator in each of the games, and based on their win percentage, the random number will decide who wins. (Team A has a W% of 72% Team B has a W% 27%, any number 27 or below Team B wins, any number above 27 Team A wins.)

Using 4800 simulations, the team who was predicted to win the tournament was St. Ignatius, as they won 16% of the time in the simulations, followed by St. Francis in Toledo with around 13.8%.

The four teams who made the Final Four were St. Ignatius, St. Francis, Upper Arlington, and Mentor. All of these teams had the highest predicted percentage to make the final four in their region, except for Mentor.

Mentor’s win in the Regional Final was seen as a big upset in the state tournament, in this simulation they made the Final Four 16.1% of the time, 4th in their region behind Gilmour, Walsh Jesuit, and front runner University School.

Other than Mentor, it correctly predicted ¾ teams to make the Final Four. 

The only issue with my simulation, however, was with the Final Four. This year, the OHSAA Tournament decided to rank its teams in each region based on their rankings in another site, MaxPreps, rather than a coaches vote which is what they have done for a while now.

I did not use MaxPreps rankings, which also makes it interesting because my rankings don't necessarily align with the seed. But, in the Final Four, the final 4 teams were reseeded based on MaxPreps rankings. Because I did not use those rankings, the Final Four I knew would be hard to do.

Also, because their seed in their region did not necessarily align with their MaxPreps rankings (i.e. a team with a 4 seed in their region does not mean their overall MaxPreps ranking was lower than another team who was 2nd in their own region).

So, instead, I just reseeded it based on their MyHockey Ranking, as I hoped they were close enough to the MaxPreps rankings so it would be mostly accurate.

Other than that, I think it ran smoothly. The real winner was Upper Arlington, who placed 6th in the state with a percentage of winning around 5.94% of the time, and the highest in the Columbus Region.

Upper Arlington beat St. Ignatius in the Final, so I got one of the finalists right.

Now this win percentage formula doesn’t account for anything else other than a team’s ranking compared to the rest of the league, but one day I hope to add more variables to it to make it more accurate.

While I only used a certain ranking system to create this formula, I think that it could work for any ranking system.

But, I think it is a good start in a journey, and it does a good job of predicting winners.