Saturday, February 28, 2015

Hearthstone February Results

The 11th Hearthstone ranked ladder season ended at 4am. I ended up finishing at 968th place, and I'm pretty sure there were over 2k people who made it to legend so I was in the top half of legends. The claim is that only .5% of people made it to legend in the first few seasons, so it's entirely possible that I'm what, 1 in 400? Not too bad.

Of course, 968th place is worth no qualification points towards worlds, so it isn't actually very good. My goal is to get top 50 in a month, or top 100 in two months, and I now have two failed months. (One because I didn't play, and this one.)

My peak ranking was actually 40th place. But I got to that point on the back of a 10 game winning streak. I then told myself I wouldn't play unless I dropped out of the top 50. Enough people played and passed me that I fell down to 58th and then I lost my next game and tumbled out of the top 100. I never made it back up that high.

The really sad thing is I ended up going 79-78 after the 10 game winning streak that put me at 40th. So going better than 50-50 still meant falling from a good standing to middle of the pack...

The whole thing is reminiscent of the old Magic rating system where I was qualified for Nationals and GP byes and the like but was prevented from playing most games for fear of losing those rewards. Card games simply have too much variance to make it worth playing against people significantly below your rating. And in Hearthstone it seems like they're more interested in getting a fast match going than in pairing you up against an even match. (If my matches were close to even then my ELO should have remained similar after going 79-78 instead of falling so much I fell from 40th to 968th.)

On the plus side I now have 2 decks built and I'm learning more and more about the format every day. So I should be able to do better in future months! Maybe?

Friday, February 27, 2015

Hearthstone: Quest Rerolls

Sceadeau clued me in yesterday that there's a feature in Hearthstone to swap out one of the daily quests if you don't want to do it. I feel a little silly for not knowing about such a feature for so long, especially since it turns out to be pretty simple, but now I know. And other people should know too!

Basically each day you get assigned a quest at random. You're allowed to store up 3 quests at a time so you don't need to play every day to keep on top of your quests but they also don't just stack up forever. Quests can be for a variety of tasks and can be worth either 40, 60, or 100 gold. I've always just done them when they show up, or shortly thereafter.

It turns out, however, that each quest in your log has an X on it in the corner. I assumed this was to delete the quest in case you hit your limit of 3 and wanted to get a different one the next day. That's not quite how it works. Instead it deletes the quest and immediately gives you a different one. It also removes all the Xs from all your quests for the rest of the day.

So it would seem that each day if you get a 40 gold quest you should send it back. If it comes back as a 40 gold quest then you should just leave it sitting there and send it back the next day, too. (Days change at midnight PST, so 4am out here.) And then only worry about doing a 40 gold quest if you've got a big backlog of 40 gold quests.

Realistically I'm not going to go through all that effort. If I have a 40 gold quest for casting spells I'm not going to play games without playing spells all day. I'm just going to get my 40 gold and move on with my life. But it does mean that any time I get a quest I will be sending it back for a chance at a bigger gold one. And if I get a quest to win twice with a class I don't like then I can just ignore it and reroll it again the next day too instead of forcing myself to draft warrior.

Thursday, February 26, 2015

Hearthstone: Sweet Patch Change

A new patch for Hearthstone went out earlier this week. It didn't have very much in it. No card nerfs or anything like that. The card back reward for playing constructed this month, a new animation that makes me want to resolve Mimiron's Head proc at least once to see what it is, and a bug fix for my old mech deck. (Cogmaster wasn't working properly when you switch his stats with a spare part.) But then there was also a set of quality of life client changes that I really like!

They added a client state that basically makes it so spectators exist between games. If you're watching someone play, their game ends, and then they start a new game you'll automatically start watching that game too. They also removed the idle timeout for people who are watching a game. So now I can start watching Sceadeau or Matt play a draft deck and just leave it running in the background to pop in from time to time and see what's going on. It's pretty sweet!

(PS: My Battle.Net tag is Ziggyny#1233, add me so I can watch your draft games and point out when you miss a point of damage!)

Sunday, February 22, 2015

Isaac Rebirth: Stupid Builds

I've started playing The Binding of Isaac: Rebirth again recently. I'd finally gotten around to figuring out where the saved files were stored so I could salvage my old save file from my dead hard drive... And then I actually loaded the game and found out Steam seems to store the saves online too since my old game was just sitting there before I moved any files around. Hurray!

The first thing I did was complete the last challenge that I'd not done, purist. This challenge starts you with no items and doesn't spawn any treasure rooms the whole game. My solution was to restart until I got a curse room with a reasonable start (I think I kept a guppy piece and some soul hearts) and then I got super lucky with devil deals. I ended up with one of my most stupidly powerful builds despite the limited access to items. I had a knife, brimstone, guppy, dark bum, and mapping. It was glorious. And now that challenge is finally done!

Just now I finished a run with the stupidest build I've ever used. I don't mean stupid in the same way I did above, where I just combined a bunch of absurdly powerful items and murdered everything. I mean stupid in the sense that it shouldn't be possible and no one in their right mind should ever try to do it. It took me barely under 2 hours to clear the dark room with Maggy (giving me two more unlocks) and I was both unkillable and unable to take actions for the vast majority of the run. I combined an item that makes you invincible if you didn't move with two items that do passive damage. (A giant fly bomb that moves around the room, detonates on contact with an enemy, and respawns 10 seconds later and a leech that aggroes things that get near you and then does miniscule amounts of damage to them.) Then I would walk into a room, put the controller down, and watch a League of Legends stream. Or an episode of Heroes. Or Sceadeau playing a Hearthstone draft. Eventually everything in the room would be dead and I could repeat.

There was no way I could have won without cheesing it on this particular run. I had pretty much no damage items drop for me the entire time, and I'm simply not good enough at dodging to beat things with low damage. So maybe it's good for the game that I had an alternate win condition available? I feel like I would have had more fun if I'd just died and started over though.

I do like abusing game mechanics though, so I guess being able to do it once was fairly sweet. But I have no intention on doing it again! (Except maybe if I ever unlock The Lost...)

Saturday, February 21, 2015

Hearthstone: Value

The other day when I was helping Sceadeau understand what I think about when I draft I kept using the term 'value'. It's something I've picked up from watching streamers of the game and it's a pretty developed concept I think, but as someone new to the game Sceadeau didn't really know what I meant. I mean, obviously you want to take cards with higher value than other cards... That's what the word means... But how do you weigh the value of different cards in a draft? He asked and I answered as best I could. I've been thinking about it more lately, and here's where I've arrived.

The first thing to understand as someone who comes from a game like Magic is creatures in Hearthstone are designed to die. Creatures in Magic heal to full at the end of each turn which means it's entirely reasonable to expect a good durable creature to live several turns and eat several cards from the opponent in the process. Creatures in Hearthstone rarely heal, and never automatically. The permanent damage on them means that you play a creature, do some damage with it, and then watch it die.

The end consequence of this is the default assumption for a creature is you're going to trade it for a creature that costs the same amount from your opponent. So the basic idea of value that I'm using is that it's what you expect to have left over after trading with an average card of the same cost from your opponent.

For example, how good is spider tank? It's a 3/4 for 3. If you trade it for a 3/3 taunt guy you get to keep a 3/1 and they keep nothing. That's pretty good! Harvest golem? Spider tank remains as a 3/2, golem leaves behind a 2/1. Tank has the edge here, but if nothing else goes on the 3/2 just trades for the 2/1. Dalaran mage? Spider tank straight up stays as a 3/2 against nothing. Harvest golem stays as a 2/1 and then another 2/1. Even the taunt guy sticks around as a 3/1 taunt. Dalaran mage is terrible under this comparison.

Always comparing with something of the same cost isn't exactly fair. Fighting up or down one in cost is a pretty standard situation too, and it does help to think about those situations. There are also hero powers that can be relevant. Sticking around as a 3/1 is a lot better against a warlock than it is against a mage, for example!

You shouldn't go too crazy with comparing up or down though. Obviously an archmage trades very well with a wisp, but it also costs 7 more mana!

What sort of things are typically found on cards I'd consider to provide high value? Big stats compared to other creatures is a good start. 4/5 yetis for 4 are surprisingly good even when they have no ability. It's just super hard for any creature of equal or lower value to kill a yeti in one round, and similarly hard for one of those creatures to survive a round with the yeti.

The stat mix is also really relevant. You want fairly balanced stats, with toughness actually being better than power most of the time. Lost tallstrider is a 5/4 for 4 and it's actually a lot worse than the 4/5 yeti. The problem is the 5th power tends to be wasted against creatures around the same cost. They don't have 5 toughness, so in a trade the extra power isn't used. Having only 4 toughness increases the number of creatures that will kill it is one shot. Especially against a mage which can tag in an extra damage for 2 mana if they need to do it. But while 4/5 is better than 5/4, 1/8 is not better than 4/5.

The 5th toughness is especially important because of the existence of flamestrike. Having creatures you can play before the mage's turn 7 that live through flamestrike is clutch.

Creatures that are 'sticky' can also provide good value for their cost. Stickiness refers to a creature that generates an extra body in some way after it 'dies'. This can come in the form of divine shield, or a deathrattle effect which generates an extra creature, or even a battlecry which generates an extra creature. The battlecry ones tend to be weaker since mass removal can hit both creatures at the same time, but it is still a potential source of value.

Or maybe it's a creature that generates card advantage in some other way. Maybe it has a good sized body for the cost and lets you draw a card when it comes into play or dies. Or maybe it does some bonus damage at some point. Demolisher is a 1/4 for 3 which is a terrible body, but if it stays in play it gets to do 2 extra damage each turn. It's generally pretty terrible value, but I had one game today where I had enough removal via weapons that I was able to keep a demolisher alive for many turns and it just kept picking off small paladin creatures. It was pretty good value, that one time.

It feels like creatures often pay some amount of power or toughness to get an ability on them, and then that ability just doesn't do much. Spellpower, windfury, and taunt are some of the big ones. When they work out they do great things, but if you spend 6 mana for a 4/5 windfury and the enemy just trades for it right away you never get to attack twice and you ended up spending 2 more mana than a yeti for just a yeti. That sucks. The 2/2 spellpower guy for 2 can be good when you follow him up with a swipe or something, but if instead he just gets eaten by a 2/3 then he was low value for you.

Friday, February 20, 2015

They ALWAYS Have Fiery 'Win' Axe

Watch practically any Hearthstone constructed streamer for very long and you're apt to hear them complain about their warrior opponent playing the 2 casting cost card Fiery War Axe on turn 2. It lets the warrior attack for 3 damage twice, which is a huge setback for most aggressive decks. It makes it tricky to decide what creatures to play early when I have a choice in the matter. Heck, if I knew for sure that they had a fiery war axe then I might even decline to play a creature right away. Especially if I had an unstable portal in my hand. But if they don't have one then leading with a mechwarper is way better...

So I've been wondering... What are the actual odds that they have a fiery war axe in hand to kill my 2 drop? To give them the best chance at having one you'd assume they mulligan away every single card that isn't a fiery war axe. I'm not even sure that's an unreasonable assumption to be honest. It's so much better than any other card they could have early and pretty much ensures they'll get to the mid/late game in order to use their other more powerful cards. Maybe they'd also keep something like armorsmith against a known aggressive deck? This is one of those places where I wish I had the cards to play around with more decks so I could get a feel for what their decisions would actually be like.

If you're going second then you start with the coin and can play your 2 drop on turn 1. This means the warrior gets to look at 3 cards in their initial opening hand, and then 3 cards after they mulligan them all away and 2 draw steps. So they need to whiff on a 3 card hand and then whiff on a 5 card hand. This nets out to them having a 44.4% chance of having a fiery way axe on their turn 2 to use against my mechwarper.

If you're going first then you no longer have the coin and can only play your 2 drop on turn 2. This means they get to look at 4 cards in their opening hand, then 4 more cards and 2 draw steps. So they'd need to whiff on a 4 cards hand and then whiff on a 6 card hand. This nets out to them having a 52.6% chance of having fiery war axe on their turn 2 to use against my mechwarper.

So, does a warrior ALWAYS have fiery win axe? No. No, they really don't. If they have the coin then they're a little bit of a favourite to have one, and if they don't have the coin they're a little bit of a favourite to not have one. Overall they'll only have one a little under half the time.

I can see why it's frustrating though. Personally I've won 58.5% of the time against control warrior with my mech mage deck and my feeling for how that match goes is I win if I get an early rush in, or if I get a stealthed archmage antonidas. And I lose if the game goes long and I don't get infinite fireballs. Them having the war axe means my early rush is almost certainly destined to fail, but when they don't have one, and they don't rate to have one, I get my rush on and get a strong edge. They can stunt me with an armorsmith into a coined death's bite too, so it's not the only determining factor, but I suspect if I'd actually been tracking turn 2 FWAs it would be positively correlated with their chance to win.

Thursday, February 19, 2015

Hearthstone: Spectating

An interesting feature of the Hearthstone client that I stumbled on last week is the ability to spectate games being played by people on your friends list. It turns out this lets you watch their game, in real time, including being able to see the cards in their hand. It even lets you see their mulligan decision if you start spectating fast enough.

This is a really useful learning tool, for both people. It lets the spectator see what decisions the player is making in terms of cards to play and whatnot. It lets the player ask for suggestions from the spectator without needing to provide extra information about game state. It's especially useful when using an external voice chat program like Skype.

I used this one time with Sceadeau as he asked me for some quick tips on arenaing. Being forced to explain why I would do what I do certainly helped him out, but it also helps me out too because it gives me a chance to figure out why I do what I do. Being questioned gives me a chance to reinforce my choices as right ones, or it gives me an opportunity to fix a leak in my game I didn't know existed. Win-win!

I don't have many friends after my adventures with being hacked late last year. My tag is Ziggyny#1233 if anyone wants to add me.

Wednesday, February 18, 2015

Hearthstone: Revising 'To Legend' Numbers

Sthenno pointed out that starting my simulation off at 0 stars isn't terribly realistic, especially now that I'm actually at legend this month. When the ladder resets for the next month you don't get set back to the very bottom of the ladder. Instead you get to start with a number of stars equal to the number of ranks you gained in the previous month. This means that I'll start off at 25 stars next season instead of at 0 and will therefore only need to gain 71 stars to hit legend.

Now, this isn't quite as good as it seems. The first 10 stars are practically free in that losses don't cost you stars down at that level. The last 26 stars are harder to get because the '3 wins in a row' bonus is removed at that point. So you're shaving off easy stars and still have the big slog at the end, but it should still make things a little faster. But how much faster?

I'm on my real computer now so I bumped the size of my spreadsheet up to a million runs, coded it to start with 25 stars, and set it running. Check the same things I checked last time, which was the minimum win percentage to guarantee legend in X games... (I checked both just the first 100k and all million and the numbers were the same.)

1700 games - was 52%, is now 52%
850 games - was 54%, is now 54%
567 games - was 56%, is now 55%

So not a whole lot of change, actually. But I'm now more wondering the inverse of what I'd been checking. With a given win percentage, what is the most games it took to hit legend? Average? I mostly care about the range around 50%, and anyone who didn't hit legend at all is counted as hitting it in 5000 games just because of the way I set things up. So averages close to 5k were mostly people who didn't hit legend at all.

45% - 1428 min, 4990 average
46% - 1035 min, 4922 average
47% - 809 min, 4522 average
48% - 718 min, 3482 average
49% - 606 min, 2228 average
50% - 502 min, 1418 average
51% - 459 min, 985 average
52% - 417 min, 733 average
53% - 353 min, 583 average
54% - 305 min, 484 average
55% - 280 min, 412 average
56% - 251 min, 360 average
57% - 241 min, 319 average
58% - 221 min, 286 average
59% - 206 min, 260 average
60% - 189 min, 238 average
61% - 174 min, 219 average
62% - 164 min, 203 average

Basically it really feels like 54% win rate is a good target. Find the fastest deck that wins about that much and you're good to go. Slowing down to eke out a slightly higher win rate isn't likely to be worth it. Speeding up by losing more often than that is probably too big a hit to your actual speed to legend, especially if you're far away from the minimum values.

Tuesday, February 17, 2015

Hearthstone: Winner Winner, Legend Dinner

It took a total of 266 wins with mech mage, but a few hours ago I managed to put together the last win streak I needed to get up to legend rank. It put me in at position 168 to start off, which seems really high. I've been matched against people in the 400s I think, so that's way up on the ladder. Has it been tracking my Elo all along? Have most people just fallen below the initial Elo? Who knows!

Now the goal would be to get up to the top 50 or the top 100 to earn some of those worlds qualifier points. But I'm wondering if maybe I should work on a better deck first? I've been finding more and more I play the same people repeatedly as I got closer and closer to legend, and some of them have decks I have very little chance against. (This version of the deck has gone 2-9 against zoo warlock, for example.) So having more decks to switch between when the guy who plays zoo is around would be really nice.

But what deck should that be? My problem is I don't have enough experience with constructed to really know what beats what. I'm getting a decent idea of what beats mech mage, but that's it. Probably the right thing to do is to build a zoo deck and use it next month on my way up the ladder! Focus more on drafting for the rest of this month and grind out a bunch of mech mage games closer to the end of the month and try to eke into the top 100.

In case anyone wants to try out my current mech mage variant (not exactly ground breaking stuff here, and some decisions were based on not owning many classic cards) here's the list:

2xclockwork gnome
1xmana wyrm
2xfrost bo9lt
2xunstable portal
2xmad scientist
2xmirror entity
2xspider tank
1xtinkertown technician
2xgoblin blastmage
2xpiloted shredder
1xarchmage antonidas
1xdr boom

Monday, February 16, 2015

Hearthstone: Disenchanting Value

A while ago my brother suggested to me that maybe the optimal line of play for a 'free to play' constructed player would be to flat out ignore some of the classes. Thus far I've been only disenchanting excess cards but I could just blow up every card for a class I'm unlikely to play in constructed like hunter. Certainly with the new information that I'm going to want constructed decks for 3 or 4 classes it certainly seemed like investigating the value I could earn by sacrificing 4 or 5 of the 9 classes seemed reasonable.

I've had a spreadsheet under construction for a while now with every card list and how many I owned to make it easier to track what cards I still needed for specific decks. It was a simple matter to modify that sheet to take into account the value for blowing cards up and then break things down by class to see where I stood.

For reference I have specced out a a few decks I'm potentially interested in and they mostly cost in the neighbourhood of 5500 to 6000 extra dust to make.

Now, if someone were to step into things with no cards but infinite dust, how much dust would they need to make all the cards? Each class currently has cards that cost a total of 9440 dust to make. The neutral cards cost 90160 dust. All told it would run you 175120 dust to make all the cards. I've averaged about 100 dust per arena run with my ~67% win percentage so I would need to draft 1751 times to craft all the cards. (Fortunately most of that dust is generated by blowing up cards from packs and I can just keep 2 of each... But even then, the relevant cards are only worth 4 times the disenchant cost so it's still a silly number of drafts to get all the cards.) I could also buy packs with gold, and I do average 125 gold per draft.

Anyway, those are just some numbers with no real importance on my current dilemna. The question that actually stands is what I could get by blowing up all my cards of a class. Could I get the 18k dust I need to make 3 more decks? (Actually less than that since all 3 decks run The Black Knight which runs 1600 dust to make alone.)

The answer is a resounding no. If I were to blow up every single card I owned I'd only get 5850 dust out of it. And that includes blowing up Dr Boom for only 400 dust when he also goes in one of those decks! Choosing to sacrifice all of my druid cards would only get me 135 dust! I could get almost 700 out of my mage cards, but I'm using many of those in my only deck! Heck, I have 750 dust on hand as it is, and I've spent 3200 making cards already.

So while sacrificing a class or two is an interesting idea, it actually isn't very useful for me. It wouldn't get me any of the decks I'd want, and it would certainly set me way back in the long run. You never know when I'd suddenly want to play a shaman deck and be sad that I didn't own any shaman cards at all. It would also take a lot of clicking to blow up all those cards (there's a button to automatically blow up all the excess ones).

There is the option of manually picking out the epics and legendaries I own that really suck and blowing those up. I have a Bolvar Fordragon card, for example, and I can't imagine ever putting him in a deck. But that's one card, worth only 400 dust.

Realistically there are only three options that I see to actually get a bunch of cards. Pay a bunch of money, draft a ton for dust, or spend gold on packs instead of on drafting. (Or go super infinite on drafting...) I've actually been hitting the gold earned cap in ranked play every day this week and while that only gives one pack it is a pack for a set I can't win in draft.

Touching on that last point a little... I now own every common in GvG and 93% of the rares. By contrast I own 27% of the commons and 10% of the rares in the classic set. So I actually do think I need to buy a bunch of classic packs in some way.

I don't want to spend money on packs at this point, so I think my plan is to keep playing a lot of constructed and spend some of that gold on packs and some of it on drafting.

Sunday, February 15, 2015

Hearthstone World Championship

Why does getting up to legend on the constructed ladder matter? Well, other than just wanting to be the best and making a number get bigger (smaller?) there's also a pretty majour tournament which uses ladder rankings as part of the qualifying method. They haven't announced a ton of details about the world championship itself in terms of when/where it will be but they have posted the qualification details.

As far as I'm concerned the basic idea is the top 23 point earners in the US/Canada get qualified into a tournament, with the top 8 getting some extra byes. Beyond that the winner of the fireside series, whatever that is, gets in as well. And then 16 more people who advance from a last chance tournament get in as well. In order to qualify for the last chance tournament you need to have earned at least 2 points total between the months of January and August.

Ok, but how do you earn points? Get into the top 100 on the ladder each month to earn some points. Or participate in tournaments to earn a lot of points for finishing very highly. It isn't currently very clear to me what these tournaments are or how to join them. Some of them (the fireside things) seem to be local events sort of like Magic FNMs? But the closest one on the list is in Ottawa and that's sure not going to happen for me!

There's also the problem that tournament formats require playing 3 or 4 different classes. They sound like pretty interesting formats, but my collection doesn't really allow for building real decks for that many classes right now. Even if they provide the cards (which it sounds like some tournaments do) I don't have the ability to test other decks right now. I've also lost a chance at points from January since I wasn't playing constructed at all then.

So realistically my 'in' would be to just earn 2 points sometime in the next 7 months and use that to get into the last chance qualifier. You get 10 points for top 50 on the ladder in a month and 1 point for top 100 in a month. So I need to get 50th or higher one month, or 100th or higher in two months. I feel like that's actually very plausible. Maybe not this month, but the more I play the more cards I'll get and the more I'll know about the format. I'm pretty confident I could pull this off in a few months.

As far as this month goes, I actually hit the win cap in terms of earning gold yesterday. (10 gold every 3 wins, 100 gold max in a day, so I won over 30 games yesterday alone.) I'm also up to rank 2 on the ladder after making some minor changes to my deck. I'm still not thrilled with where I stand but I have won almost 59% of the 85 games I've played since I added secrets to my deck. I have started getting absolutely dominated by fast warlock decks though... They don't seem to use any expensive cards either, so maybe I should give that a try for variety's sake. And for learning! SCIENCE!

Saturday, February 14, 2015

Hearthstone Constructed Progress

I posted a week or so ago about my Hearthstone constructed deck and the potential viability of hitting legend with it. I felt at the time like it probably wasn't doing good enough unless I had been getting unlucky in my previous games or was going to start getting lucky going forward. My deck just didn't seem like it really had enough of a late game. But I didn't have any actual stats to show any of that, I just had my gut feeling about being stuck in the rank 10-12 range for a while.

Anyway, I started keeping pretty meticulous track of my results in a spreadsheet. For a brief period of time I tracked the length of each game (13 games in 83 minutes for a little over 6 minutes per game) and other than that I tracked opposing class, deck type, and result for each game.

After 69 tracked games with my deck I was 34-35. I was actually up 7 stars over this period of time thanks to the winning streak bonus. But I still wasn't really feeling like my deck had enough oomph. My brother suggested the next card I crafted should be Dr Boom because almost every deck that runs any legendaries at all runs him. I have my eye on building an expensive paladin deck at some point and it definitely runs Dr Boom. My mech mage deck could also run Dr Boom, and I had enough dust to make another card, so I went for it. I took out the copy of Jeeves since it was another late game card, probably a better one, and having extra things that cost 7 would make Jeeves worse anyway.

I've put in another 175 games with this version of the deck and I've gone 90-85. Better than before, and I think the change was a positive one, but still not fantastic. That's not at the 52% win rate I thought I'd need with 5 minute games, but it's getting close, and I'm still learning match-ups so I am getting better. I'm up 21 stars in this range but it feels like I've gotten stuck again. This time I'm hovering around rank 4-6. Rank 5 is where the bonus stars disappear, which is surely part of why I feel stuck.

All that said, I have played 244 games in 12 days. That's nowhere near the pace of 1700 games that was my upper bound for games in a month, and I think I'm actually playing more constructed lately than would be ideal what with my hard drive dying and being unable to do much of anything else. So I probably need to find a way to up my win rate even though I'm only 27 stars from legend right now.

I do think the two versions of the deck are close enough that I can safely combine the stats from the two of them to get a feel for what other people are playing. My most played matchup is the mirror match, with 39 plays. I've won only 38% of those games. That number did improve when I added Dr Boom, but not even over 40%. I do know I'm playing an older version of the deck that I copied off the internet and I guess the changes that have been made since then have made the mirror match better for the other people. In particular they're playing mad scientists and secrets and I've lost a few games to being unable to draw a cheap dude and having to choose between giving them a good creature with mirror entity or not playing any creatures. I've tried both choices and they both end in death.

Other decks I've played against more than 10 times, in descending order, are ramp druid (57%), mech shaman (67%), miracle rogue (48%), '33 win' paladin (29%), warrior control (56%), priest control (69%), face hunter (64%), and hand warlock (77%). Ramp druid and warrior control both got worse with Dr Boom which seems a little weird but Jeeves actually went to town against both of those people. They're slow decks without good mass removal which meant drawing 4 cards a turn against them would overwhelm them pretty quickly! But I don't play against them enough to justify adding the Jeeves back in I don't think.

Breaking things down just by deck category I've played against 30% mech decks, 18% control, 15% ramp, 9% miracle, 9% '33 win', 6% face, 5% hand, 3% zoo, 2% burn, and 1% aggro. So the meta actually seems reasonably balanced with 40% control, 48% midrange, and 10% beatdown.

Actually, I'm not sure if I should be counting mech decks as midrange or beatdown. My versions seems more beatdown than others, but I am also running flamestrike and 2 other things that cost 7 so I'm nowhere near as beatdown as the hunter charge decks. I did have one opponent friend request me so he could berate me for playing 'cancer' so maybe lumping mech in with face hunter does make sense.

Anyway, I don't know that it much matters... What I really want is a deck that does well against the various varieties of mech decks since they make up such a large chunk of the meta right now. I could look at what's beating me most often... Which of decks I've played more than a couple of times would be burn mage (25%-4 games), '33 win' paladin (29%-21 games), mech druid (33%-6 games), mech mage (38%-39 games), and zoo warlock (43%-7 games). The paladin deck is the one I'm trying to save up for but I need another 5000 dust for it. The burn mage was interesting looking, but realistically what I should be doing is just updating my current mech deck and see where that takes me.

Friday, February 13, 2015

Hearthstone: Board Sweepers

One of the key concepts that needs to be learned for practically any CCG is the idea of card advantage. The super basic idea is that you want to trade one of your cards for two of their cards as often as you can. Since each player draws one card per turn if you can make more of those trades than your opponent does then eventually they'll have no cards and you'll have a couple left to use to actually win the game. It's a lot more complicated than that in reality, and especially in Hearthstone in particular, but it's still a very important thing to keep in mind.

There are some cards that just innately generate card advantage without either player having much say in the matter. Mage has a card 'arcane intellect' that costs 3 and draws 2 cards. There's not anything that either player can do about it... If the mage has time to squeeze that spell into their rotation they're just going to go up a card. Similarly, azure drake is a 4/4 for 5 that draws you a card when you play it. That's innate card advantage assuming you can get any benefit out of the 4/4 body (and you almost certainly will!) There are also weapons like the paladin's truesilver champion that basically guarantees to kill 2 of the opponent's 4 toughness creatures. You can work to minimize the value they get out of the second attack by using taunts or selectively choosing which creatures to put in play but unless you have one of the few cards that kills a weapon they will straight up get to 2 for 1 you just by virtue of drawing the weapon.

On the other hand there are quite a number of cards that may or may not trade for multiple cards, and the play of both players decides just how good they are. The biggest example here is the mage spell flamestrike. It costs 7 and does 4 damage to every creature your opponent controls. If your opponent has a single 8/8 in play then it doesn't even get card parity. If they have a full 7 creatures in play all with 4 or fewer toughness then you get a massive 7 for 1! Both players get to work to make the flamestrike better or worse with the choices they make. The mage can use freezes and stealth creatures to limit the trades their opponent can make in the turns leading up to the flamestrike, letting their opponent potentially overextend into it. The opponent can proactively make trades instead of going face in the turn leading up to 7 mana being available. They can hold onto sticky/big creatures and play them before the flamestrike to minimize the damage done and keep stuff on board after the flamestrike.

Now, if you want to properly play around these kinds of cards you need to know what they all are. This is true in both limited and constructed, with a higher priority in limited being placed on commons and a higher priority in constructed being placed on the more powerful ones regardless of rarity. Anyway, I wanted to make a list of potential card advantage removal spells that can be played around. And then Matt asked me if I had one, so I figured I really should make one. So here we go! Broken down by class and rarity after the break...

Tuesday, February 03, 2015

Hearthstone Legend Thoughts

I got quite a few interesting comments on my constructed Hearthstone post both here and on Facebook which have been running through my mind for the last couple days. One of them from Sthenno basically pointed out that if you just play enough games you'll get to legend. His number was 1700 games, which is a lot of games to get done in a month. If games take 5 minutes then it's over 140 hours to get there, and my games take longer than 5 minutes. That would actually be an advantage to playing the most mindless deck possible. I've fought a few people playing a hunter deck that can pretty much be described as mindless. It's just a bunch of cheap dudes and charge dudes. Win fast or lose and get into another game fast. I tend to be drawn to more controlly decks that have lots of decisions but even if I had all the cards for one of those decks would it even make sense to play it? How much higher would my win percentage need to be to make up for games taking 2 or 3 times as long to play?

I decided to write a spreadsheet simulation that would randomly pick a win percentage and then play up to 50000 games, stopping if it ever earned 96 stars to break into legend rank. 50000 games is actually way too many games since there are only ~45000 minutes in a month and I need to sleep and actually resolve games! But I figured it would make sure to show if a given win percentage really had a chance or not. I ran 100000 simulations with the following results...

Some of the people winning in the 44%-46% range wouldn't hit legend in 50000 games, but some of them would. But I'm not sure the number of games it would take those people make a lot of sense. Realistically if you could expect to average 5 minutes per game, and could expect to play 14 hours per day, then you'd be more like 5000 as an upper bound, not 50000. The lowest win percentage of a player who got legend in fewer than 5000 games was actually 42.7% win rate, who managed to do it in 3262 games. This is possible because there is a win streak feature in the lower ranks and anyone can clump all their losses together at the start and then wins at the end to get to legend. Any one 'can' do it, but most people won't!

Every single person who averaged a rounded 42% win rate or worse failed to hit legend in 5000 games. Every single person who averaged a rounded 50% win rate or better was guaranteed to hit legend in 5000 games. 96% of the people at 49% pulled it off, 69% of the people at 48% pulled it off, but only 32% of the people at 47% pulled it off.

Ok, fine, but I'm not actually going to play 5000 games even if I built the face hunter deck. I want to play a variety of games in a given month. Even my Hearthstone time is going to be split between drafts and constructed. So more realistically I might be able to play 7 hours a day, 20 days per month. That's still almost 1700 games! How does that shake up?

Now 45% and below are entirely failures. 52% and above are entirely successes. 71% of the 50% winners get to legend though. But realistically if I spend 1700 games and don't get to legend I'd be really unhappy, so I wouldn't be happy with a 50% deck under these odds.

What if I'm spending more time per game? What do the odds look like for 850 games? 567 games? Only caring about the full guarantee of legend I'd be looking at 54% at 850 games and 56% at 567 games. It sure feels like finding a deck I enjoy would be the big thing since it only has to be a little better to make up a huge increase in time spent in a game.

My current cheap mech mage deck is fun enough, but I'm not sure it's actually going to better than 50% once I start playing exclusively top notch decks. I have noticed that I keep losing the mirror match because other people have both Archmage Antonidus and Dr Boom while I only have one of them. Our early games are so similar that we always just trade off until we hit late game and their having two (or more) late game legendaries means they have a big advantage.

So really I need to find a faster deck, or I need to get more late game cards. Or I need to keep track of my win rate as it is and then just put in the time if I'm winning even 54% of the time.

Monday, February 02, 2015

Hearthstone: Potentially Frustrating Draft...

I just finished a draft with a druid deck that didn't seem terribly exciting. My best card was probably azure drake, which is a passable body for a 5 drop that also draws you a card. It's card advantage, and card advantage is good, but it isn't jaw droppingly amazing by any stretch. The deck was rounded out with 3 copies of starfire which is a single target burn spell that costs 6 and hits for 5 and also draws you a card. Oh, and 2 copies of swipe which hits one target for 4 damage and all other enemies for 1 damage each. So some pretty decent removal which pretty much all has card advantage baked in. Fill up with some good to mediocre creatures and I guess you have a recipe for success. I did have 4 creatures which gave spell power and at one point I managed to cast swipe with 3 of them in play to kill off a 6/7 and a pair of 3/3s which was pretty sweet.

Anyway, I ended up starting out 9-0. More than a couple of those games featured me playing starfires on turns 6, 7, and 8 on the single creature my opponent played while I attacked with a couple creatures. Huzzah!

Then I got paired up against a mage who also seemed to have an unexciting deck. He didn't play anything amazing either, though he did also play an azure drake. My biggest problem with this matchup was that he never really played anything worth hitting with a starfire. Most of his 'end game' creatures were either untargetable or created other creatures or drew cards. So his deck was a lot like mine, except with better creature quality. He crushed me.

I queued up again and was immediately matched back up with the exact same mage. He once again crushed me. The idea that you can end up playing the same person twice in the same draft is pretty ludicrous to me. Especially in a draft environment that has a very rock/paper/scissors feel to it. Many of my losses I think about afterwards and decide my odds of winning a rematch would be pretty low. Hearthstone decks are so small and drafts often end up pretty homogeneous so if a given opponent has a deck apt to beat mine and has already shown himself to not be bad enough to throw the game away I don't like my chances of winning a second time.

This got me pretty frustrated so I took off to get a drink, clear my head, and make sure I didn't face that same guy a third time. I came back, didn't face him, and finished off my draft with 3 more wins. So I ended up going 12-2 with a deck I wasn't terribly excited about, and both my losses came to the same guy.

I guess this means going forward if I lose a match deep in a draft and don't think it was just misplays on my part that I should probably wait a bit before playing again to make sure the guy who is likely to beat me gets a match against someone else first.

Sunday, February 01, 2015

Zelda Race Results

Last weekend I was up during the day on Saturday so I joined in on the second of the 'Get Yourself Speedrunning' races. It was for the SNES Zelda game, A Link to the Past. I managed to get in a few practice runs and actually knew where all the keys were located and where to go for the entire run. I didn't get lost and I only game overed once, and it wasn't even much of a time loss since I'd just entered a dungeon and dying in that game takes you to the start of your current dungeon. Since it healed me back to 4 hearts it might even have been faster than not dying!

The race winner finished in 22 minutes and 1 second. I finished in 28 minutes and 33 seconds. This means I only took 30% longer than the best person which is way better than my Mario 3 result where I took more than 4 times as long as the best person. In the Mario 3 race I did come 91st out of 140 people. In Zelda I only came 121st! Which sounds worse, except attendance for Zelda blew the attendance record set by Mario 3 out of the water. 301 people showed up! 290 of them even finished the race!

The rating system the SRL website uses is clearly flawed though, as this race showed. The guy currently in 7th place overall on the A Link to the Past leaderboard has only done one race. This one, where he came 34th. 33 people beat him, but only 6 people are ranked higher than him on the leaderboard. Presumably the system treats initial races as more important, or only cares about how many people you beat, or how many higher ranked people you beat, or something of the sort. I feel bad for the people who may care about the Zelda leaderboard since this huge race clearly messed things up.

121st place was actually good enough to get me some ranking points. Not enough to skyrocket me to the top of the charts but I have an actual number for a game now which is pretty cool. And it was fun!