51
Goko Dominion Online / Re: gokosalvager?
« on: July 28, 2016, 05:03:58 pm »
Is there an alternative way to look up logs for specific games besides GokoSalvager?
Buying provinces at 22+ coins seems suspicious. It seems like some other condition is being met coincidentally at the same time and it may be worth looking at sample game logs to work that out.
I'm not sure why an engine would want to build to, of all numbers, $22 to start greening. If anything you would want $16 (or $24). Engines should draw their whole deck every time ideally, so you shouldn't need excess money.You can improve the win rate by changing to a rule about being having enough money to buy 2 provinces and having sufficient money (Festivals) in deck. For example buy 2 provinces with 16+ in hand, 2+ buys, and 24+ in deck increases win rate by a few %. You might also include buying Gardens, as the larger number of late game buys compared to Smithy-BM gives an advantage with more victory cards available making the engine more powerful. You can also tweak a few things like buy a Festival before gold, buy a mid/late game 3rd gold in certain situations when it is helpful, more complex strategy with aggressive trashing combined with buying silver when certain too low money in deck conditions are met, changing the order of early Festival/Smithy/Village buys slightly, modifying the Dutchy/Estate buy rules slightly, choosing to buy Dutchies (or Gardens) over Provinces in certain cases when behind, etc. However, all of this adds very few percent to the win rate at the cost of increased complexity. If the OP or others using it want to keep things simple or only need a core framework and will fine tune while playing, I think the engine is fine as is.
I think Library / Festival rules are actually pretty easy to code...I believe the problem with coding some of the discussed more complex strategies is the simulator has default Library play rules about which action cards to keep that cannot be changed. It also has default play rules for Chapel and Throne Room that are not ideal in many situations (particularly with Throne Room) and are relevant to this discussion. Library-Festival... could possibly be a better choice than the discussed engine in real play, but I don't think it will be best in simulation, if the metric of success is beating a Smithy-BM player. Both will lose to the Bureaucrat/Market/Festival rush strategy mentioned in my earlier post by this metric.
<player name="Bureaucrat/Gardens + Market"
author="NolanA"
description="">
<type name="Competitive"/>
<type name="UserCreated"/>
<type name="Combo"/>
<type name="TwoPlayer"/>
<type name="Bot"/>
<type name="Province"/>
<buy name="Gardens">
<condition>
<left type="countCardsInDeck" attribute="Bureaucrat"/>
<operator type="greaterOrEqualThan" />
<right type="constant" attribute="5.0"/>
</condition>
</buy>
<buy name="Province">
<condition>
<left type="countCardsInSupply" attribute="Gardens"/>
<operator type="equalTo" />
<right type="constant" attribute="0.0"/>
</condition>
</buy>
<buy name="Duchy">
<condition>
<left type="countCardsInSupply" attribute="Gardens"/>
<operator type="equalTo" />
<right type="constant" attribute="0.0"/>
</condition>
</buy>
<buy name="Estate">
<condition>
<left type="countCardsInSupply" attribute="Gardens"/>
<operator type="equalTo" />
<right type="constant" attribute="0.0"/>
</condition>
</buy>
<buy name="Market">
<condition>
<left type="countCardsInDeck" attribute="Market"/>
<operator type="smallerOrEqualThan" />
<right type="countCardsInDeck" attribute="Bureaucrat"/>
</condition>
</buy>
<buy name="Bureaucrat"/>
<buy name="Silver"/>
<buy name="Copper"/>
</player>
It sounds to me like you need to move on from the base cards as soon as possible. Never mind the rankings and win rates. Dominion is an incredibly rich game, and I've noticed that the more complex later sets start making using the simulators as guides much more difficult, if not impossible.I am still a relatively new player, so I still find the base set interesting and enjoyable, and still regularly come across strategy improvements or new strategies that I hadn't considered, like the thief onslaught strategy mentioned above. I am planning to gradually move on from the base set, but by playing kingdoms with a couple new cards from another expansion set (probably Intrigue) and most of the cards being ones from the base set, like occurs in the base set campaigns. I may start the next time there is a sale or free trial period on an expansion set (not sure if this happens). Once base set + Intrigue starts getting old and familiar, then move on to the next set, and so on. I think this would work better than jumping in to a game with all unfamiliar cards, and likely spending enough time reading/thinking about the cards prior to my first move for my opponent to either leave the game or complain. It also gives me the chance to experience new cards gradually for months or even years, if I choose, gradually introducing new aspects to the game that make it more interesting and give more play time before becoming bored of the game. With other board games, I often get tired of games quickly for reasons similar to the ones you discussed. For example, I became bored with Machi Koro after playing just 2 games and haven't played it again since then. I think Dominion will be different, with me gradually adding expansions as I described... hopefully allowing me to enjoy the game for years. I am not in a rush.
The engine involves opening Chapel/Workshop, then going for Village, Smithy, Throne Rooms, another Workshop or two, and 3-4 Thieves which get throned each turn. Maybe add Spy if you don't have anything else you need.This is exactly what I was hoping for when I started the thread. This seems like an effective Thief strategy that is also fun and interesting to play, and something I've never seen anyone do in all of my games. In my simulations of Thief Onslaught vs Smithy-BM, Thief Onslaught obliterates Smithy with a ~98% win rate, without using Throne Room or Spy, and the game almost always ends with the Smithy-BM player having no treasure at all! I can reach ~93% win rate without using Smithy or Workshop (only Village/Chapel + Thief), although I see why you need something to better draw your deck when playing against several other types of strategies besides big money variants. These are higher win rates than I thought would be possible with a Thief centered strategy or almost anything in the base game vs Smithy-BM. I believe the 98% vs 2% is the highest win rate I've ever seen against Smithy in the base game, so I saved a screenshot (attached). Requiring so few cards to be effective also opens the door to a wide variety of thief uses.
The strategy is more complicated here too. Open Moneylender/Chancellor (5/2 is pretty bad; I'd probably get Festival/-), get a Village or Festival on shuffle 2, and build from there using Throne Rooms, Villages, Festivals, Libraries and Chancellors on spare $3s.In my sims, this strategy loses to a simple Library-Festival. I may have implemented details differently than you were thinking. Simulation code would be helpful.
Chancellor beats Silver in draw-to-X engines with sufficient actions available. So something likeI'd probably choose Library-Festival in this kingdom. In a quick sim without any optimization, Library-Festival without Chancellor beats Library-Festival + Chancellor in this example, as well as tested Chancellor options without Library. I'll review in more detail later. What type of Chancellor strategy were you thinking of in this case?
Festival, Village, Library, Chancellor, Moneylender, Mine, Thief, Spy, Throne Room, Bureaucrat
certainly uses Chancellor in the optimal strategy.
While one can build a strategy not based on money with a Thief, would it be better than the alternative in your example kingdoms? I think the answer depends on what your opponent does. For example, if your opponent is dead set on using a Lab with Moneylender terminal strategy and will not change it regardless of what you do, then a Spy-Thief works well. However, if your opponent is using a Lab with Workshop terminal strategy, then I'd expect Spy-Thief to lose 80%+ of the time. Lab-Workshop would also beat Lab/Spy/Thief. It could be useful to switch your terminal card to Thief, depending on what your opponent does and how the game is going. It seems risky to start out with a Thief strategy in these cases, especially with the Smithy example.QuoteSo yeah, I can design a kingdom in which buying Thief is the correct move: Cellar, Chapel, Feast, Mine, Moneylender, Spy, Thief, Throne Room, Village, Workshop. If your opponent doesn't buy Thief, they will probably lose.I win by a large margin with Mine as well as with Workshop (generating silver) against Thief or Spy/Thief in this kingdom. It would be difficult to find a kingdom where there is nothing better than Spy/Thief.
The answer is again in the engine. In the above kingdom, we probably want to add some draw to make a better engine (Lab is good, Smithy probably works too, there are enough actions). Replace Mine by that. The engine should beat any strategy based on money, and the only way to build a strategy not based on money is Thief.
Moneylender is a lot better than "rarely buy". It's certainly better than Moat, Feast, or Bureaucrat, especially in the absence of other trashing.The main time I'd get Moneylender is when all the elements are available for a great engine except for a better trasher, like Chapel or Remodel. An example is Village/Smithy + Festival + Workshop + Moneylender. I'd much rather make this engine if Remodel or Chapel is available, but in their absence Moneylender is better than non-engine alternatives, like Smithy-Big Money. This situation is rare, so I rarely get Moneylender. A list of more common types of games with the base set and why I don't get Moneylender is below. In what common situation would you get Moneylender in a base set game to make it more than a "rarely buy"?
Thief's main utility isn't very big, but in 4 player games it is a good Gardens enabler as it lets you gain up to 3 cards with one play. With later expansions, Thief lets you take key treasures like Platinum and Fools' Gold.That's a good point with 4-player games that I had not considered. Thief does get more valuable as number of players increases and probably has far more useful situations in 4-player games than 2-player.
Something to keep in mind is that a base only player at 4500 on the ladder is going to be a lot worse than an all cards player on the same spot on the ladder. After all they are only playing other Base only players.It's been my experience that all other persons I play against who are ranked 4500+ frequently play with more than just base set, as listed in log files. However, I usually leave "allow cards I don't own" unchecked, so when we play it is just base game cards. I'm sure they'd destroy me in a game where most cards are ones I am not familiar with, but I hold my own in base set games, probably winning the majority of games by a small margin. I expect the larger number of cards in additional sets makes it easier to distinguish players at this rank level. This probably partially relates to why there are few base game only players at this ranking. An opponent ranked 1000+ less is likely to be familiar with simple base game strategies, like Smithy-BM or Workshop-Gardens and has a good chance to win in such games, causing a big rankings decrease. However, the 1000+ less player is less likely to be familiar with the best strategy for an all sets game, so the skilled all sets player is less likely to see a rankings decrease.
I also would love to see the buy rules for the Spy/Chapel that wins by a small margin - something seems off about that and unlike the other results I have been unable to reproduce it.I get a 55%/40% win over Chancellor with the simple Spy/Chapel rules below (open with Spy/Chapel, then buy unlimited gold and silvers). I spent less than a minute writing this, without attempt to optimize or utilize additional available cards. I'm sure you could get a larger win rate with such improvements.
<buy name="Province">
<condition>
<left type="getTotalMoney"/>
<operator type="greaterThan" />
<right type="constant" attribute="12.0"/>
</condition>
</buy>
<buy name="Duchy">
<condition>
<left type="countCardsInSupply" attribute="Province"/>
<operator type="smallerOrEqualThan" />
<right type="constant" attribute="3.0"/>
</condition>
</buy>
<buy name="Estate">
<condition>
<left type="countCardsInSupply" attribute="Province"/>
<operator type="smallerOrEqualThan" />
<right type="constant" attribute="2.0"/>
</condition>
</buy>
<buy name="Gold"/>
<buy name="Duchy">
<condition>
<left type="countCardsInSupply" attribute="Province"/>
<operator type="smallerOrEqualThan" />
<right type="constant" attribute="4.0"/>
</condition>
</buy>
<buy name="Spy">
<condition>
<left type="countCardsInDeck" attribute="Spy"/>
<operator type="equalTo" />
<right type="constant" attribute="0.0"/>
</condition>
</buy>
<buy name="Chapel" strategy="aggressiveTrashing">
<condition>
<left type="countCardsInDeck" attribute="Chapel"/>
<operator type="equalTo" />
<right type="constant" attribute="0.0"/>
</condition>
</buy>
<buy name="Silver"/>
Cellar, Chapel, Chancellor, Feast, Mine, Spy, Thief, Throne Room, Village, Workshop.On the board you listed, I think Mine would be a better alternative than Chancellor. With a simple strategy of buying 1 mine, I get a 49% Mine / 42% Chancellor. I also beat Chancellor with a more complex Mine/Chapel/Spy/Village setup that has different rules for buying relating to the chapeled deck. I beat Chancellor by a small margin with Spy/Chapel alone (no Mine) as well.
Spy/Thief money is the best strategy not involving Chancellor, gets beat by Chancellor money 55% to 39%, which is fairly significant - the best strategy against Chancellor is Woodcutter money, which loses 50 - 43. It is a small, but certainly significant, benefit to be able to play your newly bought cards sooner.
In this case, I feel like this is more of an example than a specific kingdom, but is enough to show that Chancellor given terminal space can easily be better than Silver. The article on the wiki is still fairly relevant for this difference.
So yeah, I can design a kingdom in which buying Thief is the correct move: Cellar, Chapel, Feast, Mine, Moneylender, Spy, Thief, Throne Room, Village, Workshop. If your opponent doesn't buy Thief, they will probably lose.I win by a large margin with Mine as well as with Workshop (generating silver) against Thief or Spy/Thief in this kingdom. It would be difficult to find a kingdom where there is nothing better than Spy/Thief.