Ms. Information and the Dreamer are the only villains I've found to be consistently easier with less heroes.
Something else to look at with H=3 compared to H=5 is also the scaled effect of buff powers. With H=5 Legacy's Galvanize often outshine's his other potential powers like Motivational Charge, etc. With H=3 the trade-off equivalence is less, and you are also often in that "what can you do now" decision mode where Legacy adding in a few extra points of damage is possibly more important than the +1 damage later. (E.G. Legacy is mid-team and though you'd like the +1 for Hero #1 next turn you need that damage this turn - and a point of healing would be nice too!)
This also comes into play depending on what the hero mix is in that game. I play a lot of H=3 solo and Legacy, Visionary, etc. are often important for getting their one-shot damage cards into play. And a power like Motivational Charge is also important since there might actually be very few sources of healing available to the group. With a smaller pool of heroes in a game the "best play" optimal for a hero might have to get blurred since the hero might be needed to take a wider or more flexible role in the match.
You know guys... we do have this great resource in the Statistics... Give me some paramaters to search for and i can find out the information based on what weve been collecting. If you want to know 'Dawn vrs 3 or vs 4 which is a longer game' i can probably come up with the information :p
I'm glad that the Statistics Thread exists, but I don't consider it terribly hard proof of anything. I've seen duplicate entries which haven't been purged, entries where the Comments admit to the use of a houserule, and other such anomalies that prove it isn't being maintained with scientific rigor or anything (which is hardly surprising since it's something one guy is doing in his spare time, but still). People can be playing under mistaken assumptions about the rules, they can be using various different strategies of which some are counter-productive, they can be playing ultra-diehard serious or just chilling with their friends and having a laugh as they're utterly destroyed...putting all of these results into the same data pool and then taking random, non-specific samples from the pool produces results that are more interesting than accurate. It's certainly nice to have the option but I always take the results with a grain of salt.
It's certainly better to make judgements based on lots of data (even if the data's not perfect) rather than just assessing a handful of your own games.
Mistakes and games being taken more or less "seriously" will cancel out given a big enough sample size. It's kind of why statistics are collected in the first place...
I've lost hideously against Omnitron enough times when playing face-to-face games that if I was to just look at my own non-G+ experience I'd conclude that he's much harder than Dawn. Which is obviously complete nonsense.
Even normal Baron Blade can be quite deadly. One really nasty game played Hasten Doom x2 then Slash and Burn, repeat on 2nd turn, then Consider the Price of Victory. That's a lot of damage and with 4 players that's 11 cards in the trash on his 3rd turn, 12 after his MDP is down.
The duplicates have been purged, or at least all that my tired eyes could find. Ill look through the comments - You may notice the thick grey bars every so often? Thats a note to myself that ive cleared everything before them - or where the stats were when I stopped last. There was a good 3, 4 month period where I was working 60+hr weeks and went to europe for 10 days and other things that prevented me from cleaning it up. But its working on getting there!
Also, like silverleaf says, the whole point to a massive amount of statistics is to smooth out the bellcurve. There will be anomolies, yes, but the way math works the more entries you have, the less said anomolies matter. And we are upwards of 7000 entries now, which means were probably pushing easily 2-300 for each hero and probably around 150 or so per villain.
I thought the gray bars were just some number of days worth of statistics. Live and learn.
they were about a month, cause thats about how often I was going in and cleaning it up :P
Yeah. It's true the data isn't perfect... but it is imperfect across many sources, which makes it more useful than imperfect across our own personal experience, so while I do look at it with a grain of salt, I do tend to see it as more telling than my own personal experience... which is tainted by a lot more uniform biases.
Yeah, you need a wider range of results to get a better idea of what's what - if you only look at your own playing experience, for example, you'll only have stats for the particular number(s) of heroes you tend to play with the most, plus your game results are likely to be affected by the differing playstyles of your players, especially if you play with the same people all the time - if someone likes to play a certain hero in every game and always does so in a certain way, they might achieve or miss some things they might (not) have done if playing a different way. Pretty much all my games are with H as three, so I probably show a much better win percentage against someone like Miss Information or the Dreamer than someone who regularly plays with H as five. Plus, of the two of us who play, one of us plays two heroes and he tends to prefer certain heroes over others, while I similarly tend to take certain heroes I like playing more than others.
If you have a whole load of people, all playing their own games with their own groups of people, all uploading many, many statistics like we have here, you get a much wider range of results so as described, it'll smooth itself out the more you get and you should have a better all-round idea as to how certain game setups tend to work out.
I play all heroes pretty much equally. Currently each of them has been played between 14 and 16 times, and I'm tracking the total so that I can prioritize the ones that are trailing.
plus your game results are likely to be affected by the differing playstyles of your players, especially if you play with the same people all the time
I have quite a variable playgroup. Only one person of my acquaintance gravitates toward one character with any regularly, and even he expands his horizons from time to time.
Pretty much all my games are with H as three, so I probably show a much better win percentage against someone like Miss Information or the Dreamer than someone who regularly plays with H as five.
Again, I intentionally vary this up (when I have a choice; in "real" games it's however many people we have, but in solitaire games I do about 45% 3H, 35% 4H, and 20% 5H).
If you have a whole load of people, all playing their own games with their own groups of people, all uploading many, many statistics like we have here, you get a much wider range of results so as described, it'll smooth itself out the more you get and you should have a better all-round idea as to how certain game setups tend to work out.
I'm more inclined to believe that the opposite effect would occur. You might logically conclude that all the water in the ocean tends toward its level as water is wont to do, but nonetheless, there are such things as tidal waves. If a certain playgroup routinely mis-plays a certain situation and has a LOT of games featuring it, the entire database could be distorted as a result. Since I can't personally verify the results from the statistics thread, I trust it less than I do my own experiences (and, more relevantly, projections based on apparent logic, e.g. "obviously The Dreamer is more dangerous with higher H, since she plays H-1 villain cards per turn, deals H damage to everyone with Violent Nightmares, etc.").
We've talked about this before, but that's still not a big enough sample size to draw any real conclusions.
Evisioner, go to your local community college and audit a Statistics course. (if you dont die from straight up boredom first...) youll learn a lot of cool things about sample pool sizes and what it means for the Project for having 6000 entries versus ... say the 80ish or so you have personally in the system
First off, is the Bellcurve of statistics, which (greatly simplified becaues I suck at statistics math) basically is a mathematical provable formula that extreme circumstances (Tidal waves, in your ocean example) will end up as a small percentage of the overall statistics (the two ends of a bell shaped curve (see this picture: http://andihammer.com/wp-content/uploads/2013/10/6a00d834525fff69e20133ecf35c3d970b-800wi.gif ) ) and the far majority of 'normalized' results will occur in the middle.
Then youll also learn about Error Rate - which is another pmathematical provable formula that tells you how far off your results will be. When you are in the 10, 20, 30 points of data you'll have 20 or 30% error rates. (i think... my math is hazy) and its not until you get into 500+ data points (that would be 500 times Legacy plays against Baron Blade) that you finally get down into 2-3% error rates.
More power to you that you keep your play varried. But you are still just one person, and even with your prolifferant addition to the project, we can point out some things
Your heroes - for a well seen diversion - Bunker and Tachyon are at 21 games each, but Mr Fixer is at 14 - thats 3/4ths of the size of the first.
or, in otherwords, given your 80 some games, you are 25% likely to have tachyon in your group, and only ~15% likely to have Mr Fixer - thats a 10% difference. Again, this doesnt seem like much, but at 1000 games, if your trends continue? that would 250 games for Tachyon, and only 150 games for Mr Fixer. Your own preferences slowly start to be shown.
But this also proves the other side - more data = more information and less error. Becuase if you do get 1000 games of your own and come back later, how likely do -you- (by your own points above) think youd still be at the same ratio? They are pretty close now, your tachyon / mr fixer ratios, but if you are correct (in your statement that your group is varried) then over 1000 games suddenly youd have a closer one. Maybe in reality, instead of extrapolating as I did above, youd have 250 and 230 games with each hero. Simplifying that (a LOT) means that the difference between 15% use of mr Fixer and your (possible) 23% use of Mr Fixer at 1000 games would mean at 80ish games you have an error rate of 8%. - Which means that your own, personal view of what is going on is at least 8% (probably closer to 10-20%) off from a true understanding of things in an analytical situation.
Yes the ocean is not always flat. Tidal waves exist. Tidal waves also happen once in a long while, while the rest of the time - day after day, week after week - the sea is mostly flat. Even when you take into the Up and down of waves, the mathematical funciton of a wave means that the average height (With the top being plus and the bottom of a wave being negative) ... is flat.
More data = less impact that odd stats will have.
I am aware of all these concepts. I simply do not consider them accurate or relevant. My attitude is that, whether or not statistics proves that the ocean averages out to flat, you can still get your house washed away by a tsunami, and therefore the statistics serve only to lull people into a false sense of security. It doesn't matter how unlikely a thing is when it actually happens. I've played games where I assumed a villain was harmless, and managed to get one hero or even the entire group killed by virtue of that assumption; I've also played games against villains that were supposed to be tough, and found them an underwhelming cakewalk. Therefore, I assume nothing beyond the absolutely obvious. It's obvious that, say, Spite (the original version that is) is very hard to defeat before he flips, because he's healing every time he deals damage and every time he kills a victim, and every turn besides that in Advanced. Thusly, if the statistics record "proves" that 60% of all games against Advanced Spite are won pre-flip, the only thing this will "prove" to me is that the statistics record must be flawed somehow.
(Also, for the record, I'm not the one who prefers Bunker and Tachyon to Mr. Fixer; I've actually played him more often in actual non-solo games, but a lot of other players have taken Tachyon or Bunker in games where I was playing. Not that this disproves your point or anything.)
No. That means that your games fall into the 40%. See how statistics work?
Seems the greatest danger is the assumptions.
Apart from the matter of how many times you play each hero, it's how many heroes are in the group in total, where each hero comes in the play order, which specific heroes are playing, and which villain/environment combo is being used with that particular combination of heroes in that particular order. Someone once worked out how many total possible games there are, and it was an absolutely enormous number, taking into account every possible combination of everything that can ever be played (and that was, I think, pre-Shattered Timelines so the total will be even bigger now).
And if you find that one particular hero is played a lot and also wins a lot, it might just be that you only tend to use that hero in a combination in which they happen to do best. Card luck also comes into it, of course - a Wraith who starts a game against Plague Rat or Spite with a couple of Impromptu Inventions in-hand is already in a much better place than one starting up against Miss Information or Citizen Dawn with a couple of Suture Selfs and Inventory Barrages. I tend to base my hero choice on which villain we're up against, and sometimes also which other heroes are in the group. If you randomise all your choices (we only randomise villain and environment, via dice rolls), you might end up with a lower win rate overall than someone who chooses everything, because you might roll up a combination which ends up being really difficult (lots of equipment/ongoing-dependent heroes against someone who destroys a lot of stuff, or low-damage heroes against someone with high damage and lots of minions/ongoings).
Even if you choose all of the decks being used, you can still get your arse kicked if you just don't get the right cards while the villain gets all their best stuff out and/or the environment plays only the stuff which messes with you and barely/never inconveniences the villain at all. Hopefully games like that won't come up too often, but that's what the stats are for - so you can see "Okay, Baron Blade tends to lose nearly all the time but he's had a few games where he wins in just a few rounds - so we should have a good chance of beating him in general". It's not so you can look at the stats and instantly know whether a hero/villain/environment combo will result in a win (what would be the point in playing if that were the case?) - it's to give you an idea as to the general difficulty of a game and the effectiveness of each hero in various circumstances. The more games we get into the database, the more it'll even out and the better idea we'll all have of how well each deck works in as many different circumstances as possible.
Or something.
Statistics are never terribly hard proof of anything. They are statistics.
This whole tangent started because I said 'I can look up whats a longer game, Dawn vrs 4 or Dawn vrs 5 if you want me too!'
I suppose I should have included 'On Average' in there. Thats all Statistics are - On average, this is what seems to happen. This is data we can correlate and draw conclusions from, but of course there are always extremes and other things. No one here is saying that you can't end up having a game with 5 heroes against dawn that only takes 10 mins or one that takes 2 hours - life is made up of extremes.
However, the discussion you were having was not about extremes. It was about - on average which is longer, dawn vrs 4 or vs 5? and thats exactly what statistics are for. On average, such and such information is true.
Extreme cases always will exist, and extreme situations will always be there. No one is telling you to discount them. No one is telling you that the statistics means everytime you play this will happen. What the debate originally was about is what statistics will answer however.
Well the bottom line is that I don't approve of making decisions based on heuristics. If you don't actually KNOW the answer 100%, you shouldn't pretend that you have 'relevant data' which somehow impacts the actuality of the situation. If you go with an assumption that something is 80% likely to happen, all that accomplishes is to make you feel incredibly stupid when the 20% likely scenario comes to pass. I'd rather just take it as read that the outcome is unknown, and plan for every possibility.