To start off, I have a few disclaimers and assumptions made for this data set:
District events are only hosted at high schools with teams. This is done to reduce the number of distance checks done against the google maps API.
The district championship is hosted at Williams Arena with 60 teams.
The world championship (and associated travel distances) is outside the scope of this analysis.
Teams are only the host for 1 district event. This simplifies and encourages the search to spread events out.
Teams would only be going to two district events. Third district events are outside the scope of this analysis, but there will be extra slots.
This analysis uses a combined dataset from 2023 and 2024 registration for a team count of 202, requiring 11 district events (and therefore 3 fields). This analysis uses Minnesota, North Dakota, and South Dakota teams as recommended by the FUM PDO committee.
This analysis uses 50 miles as a hard cutoff between a “nearby” and a “travel” event for teams, and a flat cost estimate of $6,000 per travel event per team. The details of team travel are far too complex for this to fully simulate, but this was chosen as an approximation.
The 2024 data on the graphs is not yet representative of a full season because many of our teams (including my own) are still on waitlists. I expect the total travel to be even higher than 2023.
The simulation picked the event locations by finding the teams with the most number of “nearby” teams (tiebreaker being a lower team number), choosing that as an event with up to 40 nearby teams, then repeating the process until all 11 event locations were selected. From there, the simulation picked the closest event for the teams who had not yet been placed in an event, then picked remaining placement to reduce the distance of travel. The district championship teams were selected by those with the highest normalized EPA for 2024 (as estimated by Statbotics). By this selection process, here are the 11 optimal district events in FUM:
Hosted by 3630 in Minneapolis with 40 teams (40 nearby & 0 travel)
Hosted by 2177 in Mendota Heights with 40 teams (40 nearby & 0 travel)
Hosted by 6175 in Eden Valley-Watkins with 40 teams (23 nearby & 17 travel)
Hosted by 2512 in Duluth with 20 teams (11 nearby & 9 travel)
Hosted by 3691 in Northfield with 40 teams (34 nearby & 6 travel)
Hosted by 5913 in Pequot Lakes with 24 teams (9 nearby & 15 travel)
Hosted by 3212 in Granite Falls with 32 teams (9 nearby & 23 travel)
Hosted by 3278 in Detroit Lakes with 19 teams (8 nearby & 11 travel)
Hosted by 4181 in Blackduck with 11 teams (7 nearby & 4 travel)
Hosted by 2847 in Fairmont with 23 teams (6 nearby & 17 travel)
Hosted by 5464 in Cambridge with 40 teams (21 nearby & 19 travel)
Note: Events with less than 40 teams have extra slots for third plays.
For those of you (like me) who don’t have the city names of Minnesota memorized, here they are in an image (or in google maps):
However, the most interesting results are those coming from an analysis of the simulated results (raw data available here, and the notebook used for this analysis here).
The move to districts almost doubles the number of event slots (286 in 2023 to 500).
It also halves the total travel distance required (53,400 miles in 2023 to 22,500 miles).
While travel costs are not linear with travel distance, the total travel costs did go down ($1.32mil in 2023 to $1.128mil).
I’m not going to claim to know all of the roadblocks and steps required to get from here to there, but I wanted to have the data so we can make informed decisions and know the benefits.
I strongly hope that following up on this community feedback meeting that there are additional “town hall” meetings in which information like this is broadly communicated teams. The benefits seem so overwhelming, especially in light of this data.
First off, great work in getting something that begins to make sense for a spatial optimization, not always the easiest.
Would there be any way to use highschool size (or town size) as an initial first guess at where to place events then optimizing based off that? Or at the very least a check on the size of highschool (i.e. number of students, size of town, etc)? There are a lot of small highschools in the state that can host small single day off season events, but are likely not going to be capable of larger events that span 2 days.
The assumption of “team closest to the center of a geographic cluster” is an EXTREMELY hard sell for me. I.e. Bemidji will almost certainly be a better option than Black Duck, or St Cloud over Eden Valley -Watkins.
Apply some principles of central place theory to the python script and you are likely to get a result others will find useful.
Finally, and perhaps most importantly, FUM is more then MN. By taking the center of a cluster you are always is going to push it to the middle of the study area. For growth in team numbers in the district it makes sense to support the less saturated edges of the area (i.e. Arrowhead, Red River Valley)
Fundamentally I have the belief that the locations of events are rather self explanatory and will handle themselves. The real key is optimizing which weeks they occur on.
Fwiw if the double deccer became the st championship I would not be mad. Seems like it would continue to be a great venue and has a lot of potential as a state comp. All the logistics are there, only thing that is a major change is the date.
I don’t think the nuances of precise district event locations dramatically changes the major takeaway that the district system will not dramatically increase overall travel costs for teams in the region as has largely been speculated.
One goal I had in throwing together this analysis in time was to only use data I could easily grab from a public API. That’s why I used EPAExpected Points Added to determine district championship teams instead of historic state championship rankings. That said, I think that if FUM was going to move towards a district system, the locations would definitely need to be refined based on not only the viability of schools to host events, but also for towns to host the event’s teams.
I actually think this can be a great strength of the district system, by placing events in strategic cities, it becomes easy to expand the program in that local area. Most of the events outside the metro area are currently targeting 5-10 nearby teams, and the exact placement of them doesn’t affect the results of this analysis too much, so shifting them around to target specific areas of growth is important.
That’s where the problem lies in my eyes and many others: cost.
We are talking less than a 20% decrease, this is all modeled (and correctly indicated ) that assumptions were made) with 50 mile cutoff for local and $6k for travel. If that shift in location is results in an increase of teams going an extra 20? miles (which it almost certainly will, as distance was optimized for) then the amount of teams that are within that 50mi cutoff changes a lot.
Totally fair
Yes, this needs to be explicitly called out for future growth plan imo. Initial placement of events and future events. Once an event gets established it is likely to be at that location for at least a few seasons.
Were we looking at the same graphs that Tyler made? Changing the exact locations for events could dramatically increase the overall distance traveled by teams (in fact it could double) and we’d still see less travel than we see today.
Point is there is a lot more into where events are places than optimizing for number of teams that are close enough to commute.
I appreciate tyler’s work for sure), but the assumptions that are being made are far from convincing in my eyes.
going through the takaways:
Yes, dependent on the number of events of course, which in turn is decided by the natural distribution and demand.
Yes, this was optimized for. so in a perfect world it is halved.
Reduction of ~200k over hundreds of teams. Definitely an impact (i will gladly take a few hundred more in robot/travel budget). But with a swap to districts now there is the possibility of a state championship and all that travel time/cost.
My general MN district argument takeaway:
Districts only work well where there is high team density. Either the district needs to encourage rapid expansion (FIM) with sustainable teams (maybe not FIM) or it will suffer (see CHS this year)
I like the idea of the DCMP rotating between Duluth and Minneapolis, and maybe throwing Grand Forks in there. Duluth is the obvious choice if there’s more than one division, though.
For sure, this is an optimal situation and reality is not going to match it. However, I do think this analysis shows that it’s not going to be a huge increase in costs to teams (on average). I think that the most important number to focus on when picking actual locations will be the number of teams who do not have a “nearby” event. This optimal set gives us only 35 teams without a “nearby” event, but I expect a realistic picking of locations to be in the 40-50 team range.
This is some great work. I’d love to see it extended to include direct team to team comparison with past years. For example, my team has historically attended a local event and a travel event - that doesn’t change under this hypothetical event distribution (in fact, since we wouldn’t be paying for a second regional, our costs would go down). Then we have teams who have never travelled before, but would need to under this system - for example, 7850, whose costs would go up. While the aggregate costs over the entire region are useful, understanding them on a team-by-team basis is perhaps more important, to understand where we would be saving money, and where we’d be spending more. I think it’s important to recognize that those teams who would need to spend more under this model would be placed in a difficult position.
For the Championship, I don’t know how realistic a 1-division, 60 team championship is. We had 190 teams last year (189 the year before, and currently at 184 with a number of teams we still hope to register late), which puts us very close to Ontario’s 198 teams - and they had 2 divisions of 40 teams each. So, you may need to add a few more teams to the championship. As for venue, Minneapolis would probably be the best option from an overall travel cost perspective, but Duluth would definitely be the best venue, being able to fit everything under one roof - with Minneapolis, you could have them across the street from each other (the traditional North Star/10k split), but it’s not as convenient. Plus we have the history of the Double DECCer winner-take-all match to continue
I think offering a travel stipend to teams who are put into challenging travel situations would be a great solution for this. Knowing that overall costs go down is important as we’re all drawing from the same sponsor pool year after year. Allow teams to apply for “hardship” travel stipends which would need to cone out of a centralized fund. I believe some other districts have a similar setup.
Instead of just complaining about my understanding how about you lay out your viewpoint for me @Ryan_Swanson .
The way I see it:
There is a optimal solution that Tyler laid out. A lot of assumptions, but an optimal solution.
Solutions are highly spatially dependent and have optimized for distance (closely related to cost) and therefor show a best case scenario and assume an event can be hosted there.
There are a lot of unknowns, and while a best guess has been made with the assumptions there is generally a better distribution of events with districts leading to less travel for those events .
District championship is still a thing that costs a bit of money and has to be somewhere in a large area (although moving this around on a multi-season rotation seems like an ok way to handle this on the surface).
All of the above assumes teams are operating under the same economic constraints, some optimized solutions for districts may really hurt specific teams sustainability.
I fail to see where we are having a major disconnect. As we move out of the optimized window for logistical event hosting reasons travel distance and costs will rise. How much? I am not sure.
For teams that currently only go to one event, their travel cost will go up only if they are one of the 35 teams who do not have a nearby event, otherwise they now have a nearby event to go to and still have a travel event.
For teams that currently go to two events, their travel cost will go down. For most metro teams, they now have two nearby events to go to. For non-metro teams, they likely now have both a nearby event and a travel event, compared to two travel events previously.
I didn’t put much time into an analysis of the district championship, and honestly it was mostly an afterthought to analyze it. I felt that 60 teams for a district championship was more than the current MSHSL state championship hosts, but also a good balance against how many teams would need to travel to it.
Until team density rises this is very important. As Jon said more granular analysis is probably need than in the black or in the red when it comes to district/region-wide financial impact. If a swap to districts placed undue hardship for a certain geographic region (or teams outside of a certain distance ) then other mechinisms may need to exist.
i would hate to see a team in a district-underserved area to not be able to make it to the state championship or worlds due to an increased burden. Especially if it were a team that may not normally be competitive and this is the one opportunity.
Not every team that only attended one event had to travel to it.
It might make sense to just pick particular cities as hosts (Duluth, Minneapolis, St. Paul, St. Cloud, etc) instead of specific host teams, as an alternative look. How does it all get impacted if we move an event to Rochester, Mankato, Fargo, Bimidji, etc? I think there’s probably enough knowledge floating around to know if there are reasonable venues in some of the bigger towns around the state to use them as a reasonable guess over an idealized projection.
PS. Visitation (2177) would have a nice set up for an event, and if it came down to it I think we have the connections and support from the right individuals to be able to make it happen. Parking might be an issue if it overlaps with the school day. There are two gyms with the STEM center (and associated machine shop) right between them, they can be locked off from the rest of the school to keep everything separated, and would have 3 decent options for load-in/out paths. So, plenty of space for field, pits, and on-site machine shop!
The “optimized” solution isn’t that far off of a “realistic” solution. The two locations that are squared I struggle with a bit since I’m not super familiar with the local schools… I think Cambridge (5464) can probably host, but not 100% confident in that. Generally not super familiar with the facilities in SW MN so I don’t know the best spot for that one. I’d bet there’s something viable in that area though.
This analysis is obviously only looking at a small piece of the District puzzle… but to me it seems pretty clear that overall travel costs aren’t going to be a major hurdle. Costs will go up for some teams and down for other teams. The teams where costs go up will need to be supported and that will need to be a part of the conversations moving forward. There will need to be a budget for that… but it’s a solvable problem.
Unfortunately I’ve run through my free credit for the google maps API and don’t have the time to rewrite it all to use openstreetmap. If we instead picked a set of host teams rather than cities, it might be possible with the data I have already pulled from google maps.
I don’t think anyone here would disagree with this statement. From what I’ve heard no team in MN has ever not attended the world championship due to hardship. I have no reason to believe the same support would stop happening for teams traveling to a district championship.
