Leveling up

“I’ve been playing for a while now but I’m stuck on level X on the leaderboard. How do I get to level Y?”

This is a common question on the Dominion strategy forum and I’m sure all players have struggled with it some time in their playing career. There are a lot of tips flying around: play more games, analyze games you lost for mistakes, watch top players play on twitch, read strategy articles,…

Today, I’m going to use an interesting Kingdom to illustrate some of the common mistakes made by (in)experienced players and the impact of those mistakes on your win rate.

HeraldQuarryMarketGoonsNoblesAmuletSquireGreat HallTrade RouteFeodum

SeawayRaid

This board featured as Kingdom of the Week on reddit. The discussion that followed revolved around the matchup of the Feodum/Raid combo vs the Goons mega-engine? Most posters thought the Goons/Herald engine was stronger than the Feodum/Raid combo, but couldn’t quantify how much stronger it was. Then top player Mic Qsenoch chimed in:

“Only thing here is Goons engine, Feodum is a joke.”

A good player will be able to see the possible decks that can be built, how good a target deck is at aquiring points, the path to constructing the desired deck and how fast it can be set up. To Mic Qsenoch it’s clear a big Goons engine can be constructed which will easily score +100 points and do that in a reasonable number of turns. The support for this archetype is pretty good; there’s Quarry to pick up multiple actions easily, strong trashing (Amulet) to make the deck consistent, card draw to draw the entire deck (Nobles + Herald) and Villages (Squire, Herald, Nobles) to allow multiple Goons per turn plays while drawing the deck.

The other target deck is the Feodum/Raid combo. Raid and Amulet will allow Feodums to reach Colony status (10 VP). Since the deck is going to get flooded with Silver there’s no way to incorporate this into an engine approach. This means we’re probably going to buy only 1 Feodum each turn which means 8 turns of greening + X turns to flood the deck with Silvers. X is probably going to be around 10. 18 turns means you’re giving the engine player a LOT of time to set up. The -1 Card Token from Raid will slow it down, but probably not enough.

Let’s simulate

It’s time to fire up the old simulator. I first had to code the new Adventures cards/Events which was easy enough. My first attempt at the Goons bot beat the Feodum bot 65% to 35%. Then I tweaked a lot of knobs and eventually ended up with following (close to optimal) bots:

Goons engine bot (uncontested):
Open Quarry/Amulet (trash agressively)
Get 2 Goons asap
then focus on Nobles and Heralds (don't get a second Quarry)
when you start to draw your entire deck get more Goons
don't buy Silvers, instead get Great Halls with spare buys
don't overbuy Squires, just enough to support the terminal actions
buy Copper when you have 3 or more Goons in play

This will result in a few big Goons turns eventually three piling Heralds, Nobles and Great Halls for the win with an average of 150 points in 19 turns.

Feodum/Raid bot (uncontested):

Open Amulet/Amulet
Buy Raid and Silver exclusively
Once you have 24 Silvers start emptying the Feodums

This deck can achieve 67 points in 19 turns.

FirstGraph

So the Goons engine crushes the Feodum collector 94% to 5%. The games won by the Feodum player only happen when the Goons player gets awful draws and the Feodum player can 3-pile Feodum, Herald and Nobles/Great Hall. To combat this I adjusted the Goons engine so it protects itself from getting 3-piled and instead goes for Provinces in its big turns. This approach leads to a 99% win rate for the engine.

The less experienced player might be tempted by the Feodum/Raid combo while the seasoned pro will always go for the engine because he realizes the huge potential of the Goons engine. So choosing the wrong strategy here is a fatal mistake.

Engine mirror

Now what happens when two good players face off on this board? They’ll both go for the engine of course. Then it becomes a question of who sets up the deck the fastest, knows which piles to fight over and plays the inevitable endgame point dance the best. Let’s try to see what happens if players building the engine stray from optimal play.

An inexperienced player might open Quarry/Squire instead of Quarry/Amulet, then copy his superior opponent for the rest of the game:

FourthGraph

As expected this is a big mistake with win rates 80% – 19%. A card that trashes is almost always part of a good opening.

Here’s what happens when a player gets a second Quarry:

SecondGraph

A second Quarry results in probably losing the Herald split and means having one more non-action card in the deck and this lessens the ability for big turns. The win rates are 57% – 42% so a second Quarry is a small but significant mistake.

The key card in this kingdom is Herald. It’s going to be important to try to win the split. Now let’s examine what happens when a player doesn’t fight over the Heralds and will buy Nobles over them:

ThirdGraph

Losing the Herald split isn’t that bad if you win the Nobles split. So it’s only a tiny mistake with win rates 51% – 49%.

In my experience choosing the right opening buys is the most important aspect of the game for learning players. Mistakes made later in the game are less crucial. This is due to the snowball effect in building Dominion decks: a mistake made in the first turns of the game will have an impact on each following turn.

Buy ALL the cards

A mistake I often see even advanced players make is overbuying. They seem to do everything right: identify the engine and its components, choose the correct opening buys, fight for the important splits, but carelessly overbuy. The trap in this kingdom is overbuying Squires early on. With Quarry in play, a Squire is essentially free, so a lot of players just use up their buys and flood their deck with cards that don’t draw (ie stop cards). The sim proves overbuying Squires early on is a mistake with win rates 45% – 54%.

Eleventhgraph

Tactics

Now let’s examine tactical play. Strategy is all about having a plan, while tactics is about executing the plan. I consider the way you play your cards as tactics. Some examples of tactical decisions are returning 2 Coppers over 1 Estate with Ambassador, not playing a Smithy to avoid a bad reshuffle, choosing which card to Haven to the next turn, etc. This kingdom features Amulet which has a lot of modes to play (6 total) so a lot of room for tactical mistakes. To illustrate this, I kept the Goons engine bot that uses the Amulet very agressively and trashes every chance it gets. Then I made another bot that will prefer to use the +$1 mode if it means it can buy a better card (higher on the buy priority list). This last bot will still often trash Estates, but not Copper. Here’s the result:

TwelvethGraph

The agressive trasher comes out as the clear winner with win rates 73% – 26%. Bad tactical play in this kingdom is a big mistake.

Deck strength

Another skill in the game is estimating deck strength. Let’s illustrate this with our Kingdom. Suppose we remove the trashing cards from the kingdom (Amulet and Trade Route). So we open Quarry/Squire. Is the engine still strong enough to beat the Feodum player?

FifthGraph

Turns out no trashing hurts the engine a lot. It still wins, but the Feodum player will get lucky 1 in 5 times. Win rates 79% – 19%

Let’s remove the big drawing card Nobles. Now what happens to the engine? Can it still draw itself while playing enough Goons?

SixthGraph

Not being able to draw itself consistently is a big issue, but it still beats the Raider. Win rates 71%-28%

What if there’s no Goons? Does the engine still have enough payload?

SeventhGraph

This required a little refactoring because we need a different payload. Turns out we now do want a second quarry and focus more on Markets instead of Heralds. The win rates are 78% – 21%.

Finally we remove both the Nobles and the trashing. That should do the trick, no?

Eighthgraph

Yup, this kills the engine big time. Win rates are 11% – 86%. So choosing the engine  strategy in this case is a fatal mistake. (note that this engine will still beat a normal deck like Smithy Big Money 99% of the time so it’s still very strong, just not Feodum/Raid level).

You might wonder how strong the Feodum/Raid combo is. Here’s how it fares vs Big Money Smithy:

NinthGraph

Feodum/Raid (no Amulet) wins 94% of games, so it’s a very strong combo.

It’s hard to imagine stronger decks than the Goons engine, but let’s replace Trade Route and Feodum by Ironworks and Island. Does the Ironworks/Island/Great Hall rush have the speed to beat the Goons?

TenthGraph

The uber-rush dominates the Goons engine with win rates 62% – 36%. Turns out 11 turns is just not enough time to set up an engine.

Having a good feel for deck strength is a very important but hard skill to master. Because of the big variance in Dominion a very strong deck can have bad draws while a weak deck can have great draws and win. So it’s hard to draw meaningful conclusions from the outcome of a single game. You either need to play the same kingdom repeatedly to be sure or trust an experienced player’s opinion with thousands of games under his belt.

Wrapping up

As should be clear by now, not all mistakes are equally bad. In this kingdom choosing the wrong strategy proved fatal, while not fighting over a split was only a minor mistake. In other kingdoms this might be the other way around. If you want to win more often, you should try to identify and quantify these mistakes. Or you could just post your game log on the strategy forum and ask for advice.

Until next time and may your plays always be optimal!

This entry was posted in Uncategorized. Bookmark the permalink.

7 Responses to Leveling up

  1. yariv says:

    Is Seaway useless in this deck? It seems it would be useful to use it once (for Nobles, Goons or maybe Herald). You pay a little extra on that turn, but the additional (5 or so) buys on a big Goons turn are probably worth it.

    It’s not part of the simulated strategy, is this for simplicity or because it was tested and was not good enough?

    • Geronimoo says:

      Seaway only added 1% to the win rate, so I left it out for simplicity

    • Anonymous says:

      If you have 5 coins early, I think buying a seaway to get a herald is the right move, but it’s probably not necessary, because the goons give you +buys anyway. The seaway is a nice addition, and can minorly improve your deck, but its not all that necessary.

  2. dbergan says:

    Fantastic write up. My gut said to trash Squires into Goons, did that factor in at all? Like, if Quarry weren’t available, would that be a faster way to get them?

    • flies01 says:

      Early on, you want to play Goons once per turn just for the attack and the buy — additional plays are only useful if you need the buys. You shouldn’t aim for points until you’re drawing your deck, basically. So trashing squire into goons is something that might become useful, but only to get the first one or toward the end-game.

  3. Filipe Molina says:

    Woo! Very good article!

  4. luser says:

    That engine does not feel right, one thing is that with amulet you could convert squires to goons early quarry does not look that important, I would open double amulet and get quarry and squire on second shuffle to convert it into goons. Also I bet that second quarry doesnt help because bot is misplaying this kingdom, for example it should prefer markets versus great halls as 1vp is nothing and use herald overpay to setup turns.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s