Team Scotland - How the squad was selected for Eurobowl

The first blog lifted the lid on what sort of things a coach looking to gain their first cap should consider undertaking to improve their chances of selection. Here I want to be open and transparent about the change of approach I wanted to deliver when selecting Team Scotland. 

With the permission of the coaches who applied, I will provide open access to the data the Selection Committee relied upon to inform its analysis. The analysis from a group of trusted coaches aided me as Captain to select the best Squad of twelve from the coaches who applied to represent Scotland in Budapest.

Team Scotland at UK Games Expo, June 2024

Trusting a Captain over a System of Automatic Selection

In recent editions of Eurobowl, post-covid, Scotland has had a system of automatic places awarded to coaches based on their performance within the Scottish Championship (a NAF Series), but there are a number of problems with this approach. In short:

• Scottish coaches not resident in Scotland didn’t have an equal opportunity to gain a place in the team.

• A system which governs and confers automatic places, however it’s designed, is only ever going to be based on very limited data, making it a blunt instrument at best, removing the subtlety of context, whilst open to being gamed.

• The Scottish Championship, largely made up of one-day tournaments, in a small self-contained pool of Scottish coaches fails to highlight a coach’s true potential. Likewise, it ultimately fails to encourage coaches to really stretch themselves to improve in preparation for one of the strongest two-day tournaments on the Blood Bowl calendar.

I’ve always contended that we should trust a Captain to undertake an honest endeavour to assemble and ready the best team they can to represent Scotland at Eurobowl. 

Certainly over favouring the ‘transparency’ of an ill-fitting pre-defined system that removes agency from a Captain for fear that they’ll be tempted to form an old-boy’s network rather than trying to put the best team out on the field to represent Scotland.

A Transparent Selection Criteria

That said, I do believe to encourage engagement from the community, the Captain has a responsibility to outline an open and transparent Selection Criteria so coaches know where they stand. 

During the election for Captain I outlined a detailed Selection Criteria I would rely upon to make my decisions. Without regurgitating it word-for-word, the headlines from the selection criteria:

• The process was to be driven by an open application that enabled any Scottish coach to apply and raise their hand to be considered for selection.

• Utilising both (the limited) quantitative data and qualitative answers from coaches in their application to evaluate their suitability for the squad in the round.

• A Selection Committee of experienced coaches would be formed to advise me as the Captain, to help me avoid blindspots in my analysis and selection.

• Key metrics for coach performance were identified and published in advance of the selection process.

• Post-selection a detailed and open summary of how the selection decisions were reached (this blog!)

It should also be noted that Ant made some initial changes to the selection process, but he maintained, and I honoured, the three automatic selections from last year’s system to be included in the squad of twelve for Hungary as we evolve to all squad places being awarded based on application from Poland onward.

How did we assemble and evaluate the metrics we’d chosen

There are two layers to the metrics we used - the obvious things like win rate, loss rate etc, that gave us the individual data points we could measure from the NAF database. However more importantly these data points were reviewed when we considered the results achieved in the order of importance: 

1. Against coaches with a 200+ Elo rating

2. Results that a coach achieved with their declared preferred races and other races we felt they could potentially play

3. Overall results of all a coach’s games, screening out results from stunties or races they’d played very little with i.e. one, 1-day tournament with Nurgle

Then a time frame of results from the last twelve and twenty four months were considered to see what improvements a coach was making over time.

Snapshot view of some of the data that underpinned analysis for selection

Obviously when the squad plays at Eurobowl they’re far more likely to face coaches of significant experience, typically over 200+ Elo (the accepted benchmark that intimates a very good coach). So understanding how a coach fares against top class opposition is most important. 

Although typically the number of these games tend to be quite limited with only six coaches playing double digit matches against 200+ coaches and 50% of the twenty-six applicants playing less then five matches against 200+ Elo coaches in the last two years. 

(As an aside that’s why I’m keen to encourage our coaches to travel to big tournaments outside Scotland and very supportive of ChrisRaff88 and Amsfeld as they look to grow Mulligan’s, Scotland’s NAF National, and the Scottish Team Championship which in turn should attract top coaches to come and play in Scotland)

So with very few data points against 200+ Elo coaches the next layer that we considered is what was a coach’s performance like on the races they considered in effect their best races that they had applied with. This increased number of data points improves the confidence that we can have, but again it’s still limited and so the final layer was to look at all of a coach’s performances, screening for races like stunties.

However it should be stated here, loud and clear, that even when evaluating all the numbers involved, due to the limited data pool, they could only ever be used as breadcrumbs for the Selection Committee to follow to build up a picture of a coach, rather than taken as hard and fast and definitive. 

Publishing the data sources compiled

So far I’ve talked in the abstract. Now is the time for you to openly see and interrogate the data for yourself.

Firstly though a very big thank you goes to Sann0638. Mike, simply put, is Scotland's data guru who compiled all of the NAF data in a way that matched the Selection Criteria I had proposed and openly displays these Scotland’s Metrics that the selection committee relied upon on his Tableau, which has been viewed over 600 times. (most of them probably by me!) 

Even the process of working with Mike to create this view sharpened up my thinking and added in things like the difference in a coach’s performance when playing in two-day tournaments compared to one-day tournaments. So I’m eternally grateful to him for the work he put in to help me in my role of selector.

The Tableau Space Station built by Sann that hosts the Scotland Metrics with over 600 views

The other data source we pulled upon was the Danish Blood Bowl Elo system that they use to select the Denmark Squad for Eurobowl, so thank you to them also. I collected Scotland’s Top 40 coaches on a monthly basis and integrated that into the mix.

A summary of the data resources and links to them listed below:

• Scotland’s Metrics Folder that contains all of the data used in selection

• Scotland Coach Metrics in Tableau - the underlying data source from the NAF database, which underpinned the analysis

• Scotland Coach Metrics - pulled from Tableau for data manipulation and analysis

• Denmark’s Coach Rating system

• Month-by-month recording of Scotland’s Top 40 on the Danish coach rating system

• Individual Coach Analysis - a review of each coach’s NAF record looking at every game played to identify notable results against known strong coaches

The principal sheet worth studying is the Scotland Coach Metrics in sheets - this has everything including a detailed cover sheet and legend to aid navigation.

Legend within the Scotland Coach Metrics to help highlight strong scores in key metrics.

Lifting up rocks and seeing what was underneath

If you’ve been waylaid by looking at all the data then you’ll know there is a lot of information there to be deciphered and made sense of. To do this we undertook the following steps:

1. Compiled all the information in one master view

2. Broke down the overall view into individual races to identify the best coaches who could conceivably play a specific race

3. Identified a few key metrics that were felt to separate coaches in the margins

In looking at specific races it meant we were comparing and contrasting just a few coaches rather than trying to decide which twelve coaches out of the twenty-six applicants were the ‘best’. 

This was due to the simple rationale that you can’t have five Undead coaches (how many coaches we actually considered for Undead), and whilst a coach may be great at Undead, if they’re not the best coach for that seat then they wouldn’t be selected for that race.

What made life more complicated was the Hungary ruleset - it was felt that the meta changed significantly enough which meant some formerly established assumptions could no longer be taken as read.

With regards to fine tuning there were a few key metrics that helped separate some close selection picks:

• Loss Rate and average TDs conceded per match

• Average Opponent Elo Rating

• The percentage of games played in two-day tournaments and the differential rates for things like win rate when comparing two-day tournaments with one-day tournament performances.

When coaches talk stats in Blood Bowl they usually talk about their win rate. However with Eurobowl as a Team Tournament arguably it's more important not to lose a game within the eight boards as it is to win and thus a coach’s loss-rate is something I prized over winrate. By extension the lower the average number of touchdowns a coach concedes indicates the likelihood of losing a match is reduced.

Equally not all win rates are ‘equal’ - one of the important caveats is what kind of coach is someone beating or losing to. So the Average Opponent Elo Rating intimates how experienced the coaches that a Squad member is playing against to give context to all the other stats.

Finally, it struck me to look at how many of the matches played were under the conditions of a two-day tournament and whether there was a difference in the metrics when comparing with one-day tournaments. Eurobowl is a two-day, six round affair so it makes sense to see how coaches stand up to that format of tournament. 

Obviously the greater the percentage of games played under the condition of two-day tournaments potentially the better the quality of games a coach is playing. Likewise there is an underlying assumption that playing rounds four to six exposes a coach to pressure and better coaches. In general terms across the board there were a significant number of coaches who played less well in two-day tournaments - an area for ongoing improvement.

Data isn’t everything

As I’ve mentioned previously there are limitations to the quantitative data on account that there's simply not enough of it to be confident in it without having any doubts. Equally the variance of the game means you can’t just look at a result on a NAF Coach page and understand the underlying nature of the game that was played out.

So if looking at raw data is not enough what else did the Selection Committee review to help me reach the final decision:

1. We asked coaches to outline the value they could offer the team and what was motivating them to want to represent Scotland, taking account of qualitative information directly from them - giving them a chance in effect to make their case.

2. I went through every coach’s NAF record and evaluated individual games looking for a pattern of results against known strong coaches

3. Consider the chemistry and value each coach could bring to the table beyond mere results. Eurobowl isn’t just a weekend, but a year-long endeavour of preparation and it should be fun. So trying to build a Squad that gets along and enjoys each other’s company is as important as getting strong coaches in the squad.

   
   

Selection is not an exact science, but this Squad has got the right stuff

It’s my first time round as Captain so it's as much a learning process for me as it is for the coaches who’ve been selected in the Scotland Squad. With the advice from the Selection Committee; thank you to Purdindas, Loki and ChrisRaff88 for the time they took, I’m confident I’ve selected eleven other coaches who will join me in the pursuit of our ambitions in Budapest in October.

I hope you'll join me in congratulating the coaches on their selection and cheer on their charge in Budapest to do Scotland proud and bring home Eurobowl Hosting Rights.

Hopefully this has given some insight into the thought out process we followed with diligence and in good faith. I leave an open invitation to any Scottish coach to chat with me about Team Scotland, the process of selection and if you applied and weren’t successful on this occasion what steps we could take together to prepare you for a future Eurobowl. Likewise, have you spotted a gap, a flaw in the plan where we could improve our approach - always up for feedback, drop me a line! Thanks for reading.

BB_Jock

Team Scotland Captain for Eurobowl XVI