Dungeonquest: A Statistical Analysis
Digging into the data I've collected on the greatest board game ever made
Hey y’all. As always, thanks for reading! I appreciate you. I’m taking a fairly significant change in direction this time out by doing something I’ve been wanting to do for a long time, which is data analysis for “fun” (instead of for money/at work).
While the general theme in this blog is music, the mission statement remains “it’s me writing about things I love” and there’s few things I love more than Dungeonquest, the most fun, most maddening and most beloved board game I’ve ever played. In the Dan Sartain Greatest Hits post I wrote that:
I can confidently say that I everyone I have ever played his music for has enjoyed what they heard.
That also applies to Dungeonquest. It’s The Best. So, here’s a deep-dive into the data I’ve been collecting over 10 years plus worth of games, to try and determine who the most valuable Dungeonquest-er is.
What is Dungeonquest?
Dungeonquest1 is a board game from the 1980s that was published by Games Workshop. It’s based on a Swedish game called Drakborgen. Essentially, the folks at Games Workshop got a promo copy of Drakborgen in ‘86, fell in love with it and made a deal to publish an English language version, which they released in ‘87.
The game itself is a relatively straightforward fantasy setting: in the villages outside Wyrm’s Crag mountain, people tell stories of the treasures that can be found inside Dragonfire Castle, the dark fortress atop the mountain, which has laid dormant since the death of Evil Wizard T’Siraman a thousand years ago. The players take on the role of four classic fantasy archetypes:
Sir Rohan, the Knight; with his shining armor
Ulv Grimhand, the Barbarian; with his huge axe
El-Adoran, the Ranger; with his deadly longbow
Volrik, the Brave; a swaggering adventurer
Like most things in this game, the character you get is chosen randomly. Each have pros and cons, e.g. Sir Rohan is as tough as an elephant, but also as agile as an elephant.
We don’t really need to get too far into the game play or mechanics, as so much of it is random choices, but there are some key things to know:
The goal is to enter, get some treasure and try and get out alive.
The highest value treasure is found in the central Treasure Chamber, but so is the sleeping dragon that guards that treasure. Can you sneak in without waking him?
The types of rooms you encounter - and their contents - are randomly selected and vary wildly, from an empty room to a bottomless pit that you have no choice but to try and jump across.
Many rooms will contain monsters that must be fought and defeated. The strength of the monsters vary, as does the amount of damage each character can sustain before dying.
There are also traps, portcullises, doors, cave-ins, rotating rooms, crypts, corpses and all kinds of other inconveniences to deal with. As the room tiles are random, you can even die by getting stuck in a dead end (which has accounted for 5% of the deaths in our data set!)
Perhaps most importantly, the time in the castle is limited to daylight hours, limiting players to 26 turns each. Staying in the castle after the sun goes down is so deadly that the game designers simply state that “it’s better not to ask what happens to those characters…”
Why do people love it?
I queried some of the players whose valiant efforts make up the data set we’re going to be looking at. Here’s what some of them said:
Alex: “…it's easy, quick and frustratingly fun, nothing is really the same each game, its all about those tiles and pushing your luck against the sun.”
Rebecca: “I love Dungeonquest because you're not really playing against the other players, you're just waiting in delight to see how they'll die”
Ed: “it’s the intersection of prat falls and comradery through shared suffering”
Paulie: “Your first move can be into a revolving room, straight into a bottomless pit and if you survive that, follow it up with a dead end and game over. Few games can be so hilariously unfair and get away with it.”
For me, the main reason this game is so fun, and why I’ll never be tired of playing it is that the overall deadliness of the game makes those brief occasions when you get a victory all the sweeter. As Paulie noted above, the mechanics offer a lot of different outcomes, so even when you die its usually funny in some way or other. Like dying in a bottomless pit on your first turn (which has happened at least twice, sorry Nic.)
There’s also an unquantifiable element to the game that the data could never capture, which is that it more often than not offers up at least one good adventure (of the 4) per game, most often a race against the clock or maybe a player limping over the line with 1 hit point. It’s like the game somehow has an in-built dungeon master, pulling the strings behind the scenes. Every game has a story.
How do the Game Designers define success?
In the rule book for the game, the game designers let you know - in the first page of rules, no less - that you’re going to die a LOT:
After long experience and extensive playtesting, we know that your character has about a 15% chance of survival…
They also define how to win the game:
The winner of the game is the player who…escapes before the sun goes down…with the highest total value of treasure.
Sometimes a player will win simply by being the only person to get out of the dungeon alive, even if they have…no treasure at all!
They go on to suggest (and this is why we are here today)
Keep track of the total amount of treasure each player has managed to get out of the dungeon in all the games they have played to see who the champion Dungeonquest player is in your group….You might even keep a ‘league table’ - divide a players total treasure by the number of games played and you will have a number which reflects how good a player they are.
“Basic Metrics” (using the game designers suggestions)
First, a quick note on the data. I recorded 97 games over the years, all on paper. I moved those into Excel, cleaned up the data and everything you see below comes from that spreadsheet. Although 97 games is a good sample size overall, unfortunately at a player-by-player level we have some small samples. Part of this is my fault because for the first 20 games or so I was only recording my own achievements2.
1) Survival Rate
As noted, the Games Workshop play-testing gave a 15% chance to escape. I suspect, as gaming professionals, they played more aggressively than most of us, because our escape percentage is 26% - 11% higher.
When looked at per player, we see a range from 50% to 0%3, with four players above our survival rate of 26% and all but one player above the game designers rate of 15%. Not bad.
2) Highest Total Value of Treasure
Overall, we have 25 games (of our 97 total games) where a player escaped. There are no games escaped without at least some gold, ranging from a mere 10gp all the way to a massive 9,060gp. Here’s the Top 15, which all those where someone made it out with more than 500gp4:
There’s a clear winner here, and it is I, your humble author. I will accept your praise in the comment section. (Don’t be afraid to drop that gold medal emoji in there.)
3) Average Treasure per Games Played
This one is a little tricky, only because the game designers were ambiguous in this statement. I’m interpreting this as the total amount of gold across all games where the player escaped, which is not the same “by the number of game played”.
This chart kind of combines the two above to tell an interesting story. While I may have had the single best game, I’ve also had a lot of low gold-yielding escapes and end up below the overall average here. The samples here are small - escaping is hard, after all - but let’w take a look at all the games for the leader here, Ed, who has 5 total games and while he escaped only once, he escaped on his best game.
With the small sample sizes, I think this metric maybe be overly influenced by luck but that is a vital element of this game, so I’m ok with that.
Advanced Metrics
While working through the data, I found a couple of elements that made me want to take a deeper dive, and in the spirit of every amateur baseball statistician5, create some new metrics. What I wanted to account for were:
What if you are good at getting treasure but bad at escaping?
Does the character you play with make a difference?
What elements would combine to determine the overall most valuable player.
So for #1, let’s look at the average amount of treasure amassed per game, regardless of whether the player lived or died. This should:
Reward the aggressive treasure hunter, which is the truest spirit of the game
Account for the variances in playing with certain characters
Even out the elements of luck involved in living or dying
This chart makes perfect sense to me, because the player in first place, Raulo (aka Paul6), is the most aggressive Dungeonquest player that I’ve ever seen. Additionally, Nic is a more conservative player7, so this does seem to add further merit to this metric.
Weighted Escape Score
Moving on to the impact of which character you play with, let’s go back to escape percentage. Our average is 26%, and it turns out the character you play with does make a difference:
Ulv, our barbarian friend, is the most well rounded player, closely followed by Volrik and this seems to be advantageous. But, while El-Adoran and Sir Rohan are behind that escape percentage average, you may have noticed that the 5 best single games were all made with them, so this suggests that should you avoid death while playing them, you might do rather well.
For example, Sir Rohan dies a lot in bottomless pits, due to his woeful agility stat8. Perhaps a key to escaping with him is to get lucky and avoid jumping over stuff?
Given the significant variance in characters escape rate, I decided to weight the escape rate per character. To do this, I calculated a weighted escape value for each character, based on how far their chance to escape varies from the overall average9.
From there, I compiled how many times each player escaped with each character and applied the escape weight for each character to give total weighted score. I then divided that score by the number of escapes that player had, to give a Weighted Escape Score per Player. The benchmark here - for all escapes by all players with all characters - is 1, so being above that is better than average and below is worse.
What we see here is that Jane leads - by escaping once ever, with Sir Rohan, the hardest character to escape with. On the other hand, I have 11 escapes, but 7 of them are with Ulv, the easiest character to escape with, hence I land way below average.
Overall, this metric really rewards those players who have defeated the odds and made it out with the toughest characters. (But it suffers from small sample sizes.)
Finally, for our MVP Award, let’s combine all of these other metrics into one. To do this, I took the rank each player achieved in each of the five metrics above, added them together and the player with the lowest overall score is our MVP.
Actually, this might be more like an “all-around” contest in gymnastics, as to score low you have to have done reasonably well in all five categories:
Escape Percentage
Best Single Game
Average Treasure Value per Escapes
Average Treasure Value per all Games
Weighted Escape Score
And our MVP10 is….a tie. There’s a myriad of tie-breakers that could be applied, but despite what American Sports want you to believe, a tie is always a valid result11. Congratulations to Alex and Ed!
Anyone that wants to see the Excel file and maybe even play around with the data can find it here.
There’s a separate, newer version of the game, published in 2014 by, Fantasy Flight. They modernized many aspects of the game mechanics, but they also lost the soul of the original game. Dragonfire Castle is supposed to be a cruel mistress in which death lurks around every corner. That game gives the players more ways to avoid dying and feels like playing a video game with cheat codes: you can do it, but it’s not the authentic experience. Then again, you can buy a copy in a store for a reasonable price, so it does have that going for it.
Yes, I am an only child. How did you guess?
This is where that per-player sample size hurts, both the 50% and 0% scores are from two games played per player.
500gp is a good dividing point between the most amount of treasure you can get without going to the dragon’s lair and the least you can get if you did go in.
Advanced metrics in baseball being best understood by the beginner by reading (or watching) Moneyball.
He’s also one of the two people that introduced me to this game, along with the man at the other end of this chart, Widge (aka Dan.)
In her defense, dying on your first turn in a bottomless pit - twice! - will have that effect.
Maybe don’t wear full plate armor if you wanna jump over stuff.
The math here is: average escape percentage divided by characters escape percentage. Volrik’s escape percentage is 26.1%. The overall average is 25.8%. Hence, Volrik’s escape weight is 26.1/25.8=0.988
Wait a minute, this MVP deal is a waste of time. Let’s give the trophy to the guy with the single best game. Who was that again?
I’ll do this again in another ten years and hopefully have bigger sample sizes.