Root Analysis

Part 1: intro

The game Root is one of my favorite board games, and because of that I wanted to look at data relating to it. Root has a cute aesthetic, but the game play is anything but cute as Root is an asymmetric war game, where every faction has wildly different mechanics and point scoring. As the players Race to get their faction to 30 points first this leads to a game filled with alliances betrayal and cut throat game play as the players squabble to see which faction gets on top.

When looking at root I was interested in seeing how the faction’s performance change based on the map, setup, and deck used.

Data Source

The primary data set is Swuft’s Root Community Data Sheet, the data is a collection of root games played by people over the course of many years. The data in the sheet is self reported by the people who played the game and then submitted it into a goggle sheet which can be found here: Swuft’s Root Community Data Sheet

This data did require some cleaning so I could use, because I want this to be easily readable and concise I will not include the process of cleaning in this report. However you can view the clean data set here: Cleaned Data

Summary of the Data Set

Data Dictionary:

birds: The amount of points the Eyrie Dynasties scored in a game.

cats: The amount of points the Marquise de Cat scored in a game.

WA: The amount of points the Woodland Aliance scored in a game.

otters: The amount of points the Riverfolk Company scored in a game.

lizards: The amount of points the Lizard Cult scored in a game.

moles: The amount of points the Underground Duchy scored in a game.

crows: The amount of points the Corvid Conspiracy scored in a game.

seekers: The amount of points the Keepers in Iron scored in a game.

warlord: The amount of points the Lord of the Hundreds scored in a game.

vagabond1: The amount of points the Vagabond scored in a game.

vagabond2: The amount of points scored by the second vagabond in a game if there are more than one vagabonds.

setup: Whether or not the game used advance setup as well as whether or not the game was played on root digital.

map: What board the game was played on.

deck: What deck was used during that game.

hirelings: A logical variable that is true if hirelings were used during the game.

landmarks: A logical variable that is true if hirelings were used during the game.

Experience: How experienced the players are with root determined by how many games of Root they have played before.

number_rounds: How many rounds the game lasted.

number_players: How many players were in the game.

Winner: Which faction won the game.

Win Frequency of all the factions.

	total_games	win_rate
cats	471	22.9%
birds	288	23.3%
WA	249	28.9%
otters	169	27.8%
lizards	186	20.4%
moles	115	30.4%
crows	108	22.2%
seekers	9	22.2%
warlord	11	36.4%
vagabond1	226	31.4%
vagabond2	10	30%

Here we can see the win rates of the factions. We get win rate by dividing how many times a faction wins by how many times a faction is played in a game. One thing of note is that the seekers and warlord do not have many games played, this is because they are the most recent faction and data was added far less frequency after they came out.

Part 2: Descriptive Analysis

The first thing we are going to do is see the win rates of all the factions that have a large amount of total games.

#|echo: false
my_colors <- c("cats" = "#FF8F00", "birds" = "#0D47A1", "WA" = "#4CAF50", "otters" = "#00BCD4", "lizards" = "#CDDC39", "moles" = "#E5D1B1", "crows" = "#4A148C", "seekers" = "#B0BEC5", "warlord" = "#D32F2F", "vagabond1" = "#9E9E9E", "vagabond2" = "#424242", "vagabond" = "#9E9E9E")


win_rates %>%
  filter(faction != "warlord" & faction != "seekers"& faction != "vagabond2") %>% 
  arrange(desc(win_rate_prop)) %>%
  mutate(faction = factor(faction, levels = faction)) %>%
  ggplot(aes(x = faction, y = win_rate_prop, fill = faction)) +
  geom_col() +
  scale_fill_manual(values = my_colors) +
  labs(
    title = "Win Rate by Faction",
    x = "Faction",
    y = "Win Rate"
  )

We can see here that vagabond has the highest win rate, and moles and the woodland alliance have the second highest and third highest win rate respectively. This makes sense as those factions are considered the best factions by the community. What is interesting is that lizards have the lowest win rate, because even though they are hard to win with they are not considered the weakest but rather the cats and the crows are considered the weakest.

As we see here factions all score a some what similar amount of points although lizard cult’s distribution is far lower than other factions, where vagabond, the faction with the highest win rate has a fairly small distribution and scores a lot of points, one interesting thing is that even though otters have a similar point distribution to vagabond they only have the fourth highest win rate this could be because they only have 18 points they can score on their board and have to get the rest of their points through crafting, where as the vagabond always has a wide variety of point scoring opportunists at every point in the game.

We can See that the maps have a big impact on the performance of the factions, with Woodland Alliance specifically not doing well on Lake and Mountain, this could be because lake is less connected so spreading sympathy becomes harder due to the clearings being more spread out, and in mountain it is most likely the cost of dinging tunnels is too much for the Woodland Alliance as they need cards for their supporters deck so they can spread sympathy better. Another thing of note is how much worse the cats do on the lake map this is most likely due to the low amount of building slots in that maps clearings.

In this graph we see how the two decks you can play with effect point distribution. The two decks here are the base game deck and exiles and partisans deck. The big difference between the decks is that the E&P cards are less expensive to craft. This tends to lead to all of the factions having more similar and smaller distributions of points leading to games where everyone is more even.

This graph shows the final score for a faction as well as how many rounds the game went. I chose to cut off games that were shorter than 6 rounds or longer than 10 rounds, because those games are exceptionally rare. One surprising observation is that the length of a game did not effect the final score as much as I thought it would, with the only factions that improved during long games being moles and lizards, however the moles don’t have many observation’s beyond 10. when lookign at the lizards it seems that lizard cult do better across longer games. This would make sense because lizard cult usually are not scoring more than 4 points a turn and lack the ability to score a lot of points that other factions have.

Part 3 Secondary Data Source

The secondary data source I chose was the root 2020 winter tournament data. I found this data at this website: Root 2020 Winter Tournament Data

Much like with the main data set I used this one needed cleaning which again I will not show here to make this report easier to understand. You can find the clean data here:Tournament Data Clean

Distribution of Points in Tournament Play

Warning: Removed 1 row containing non-finite outside the scale range
(`stat_boxplot()`).

When Comparing performance in casual and competitive play it can be seen that in casual play the distributions are in general much more even. It can also be seen that the vagabonds did notably worse, however this is most likely due to the fact that the vagabond often won with dominance in this tournament and we are not looking at dominance wins. Another thing of note is how the moles now seem to be the best scoring faction where as before it was more of a competition between the other top 4. This is likely because the moles are hard to shut down and because everyone knows what they are doing in competitive play it means that their players are able to use them to their full ability.

Conclusion

All in all when looking at this data it can be seen that the same four factions seem to preform well. However it seems that in the competitive play the factions that do better tend to do more consistently better in competitive play while the factions that do worse seem to in general do more consistently worse in competitive play.