Wednesday, April 29, 2009

Risk versus Reward

A pretty lame title and a really long time since posting. I should be doing this daily, I gotta kick myself into writing more.

This is about risk and reward. Specifically, it's about whether to take risks or not in competitive gaming. This topic is pretty wild because always I smack myself thinking about some matches in say, StarCraft. In fact, StarCraft is the biggest game where this skill is useful. In some games, taking 'risks' is so well-rewarded (like Guilty Gear) that you pretty much want to expose yourself all the time because the reward for even getting a hit blocked is so good. In other games (like Killer Instinct), doing anything is so bad that you almost never want to do anything risky at all.

This is not a guide to taking risks in real life. Real life is not zero-sum and you pretty much always want to err on the side of safety in RL because the risk of loss is very bad. In games where you lose real investments (like EVE) you also want to generally play low-risk, as it benefits you more in the end. Some people might like playing high risk, but this is overall a bad thing in real life. If you go gambling at a casino (slot machines, craps, etc.), you might win money and lose money, but the odds are skewed against you. If you are playing competitive games for money (even poker and similar games with high chance vs. skill ratios), then this article applies to you, though.

So I watched a ton of SC matches prior to writing this, and I have a lot of poker-playing friends. Most of them are not as good at bluffing as I am (I am a master liar!) but they are far better at playing the odds than me so I am generally not good at poker.

Sometimes in games you'll want to play safe, even when the odds say you could play more dangerously and come out ahead. A good example is when many ages ago, I was playing Guilty Gear 2, which is a real time strategy-fighting hybrid. I had the advantage in terms of control points and I had a big unit advantage. I could have pushed for the win, or I could have defended and built up some mana to make more and stronger units - but at the same time my opponent could get mana to defend. It is a tough choice to call, and in the end it is probably better to attack and press the advantage. However, in this case I chose to defend and in the end, I won that game by holding my advantage and using it in little pushes to weaken his forces rather than devoting myself to a serious attack.

StarCraft is much the same way, but in SC's case because units can be microed, attacking is almost always good (in GG2, you can only micro your 'hero' who is more a fighting game character than a unit) and defending is much harder. You almost always want to be attacking if you can spare the attention to do it, although you want to use attention for base building and so on too. As I mentioned in my article about nothing, though, sometimes it's better to spend as much attention as you can on army building and focus on defense.

How do we choose these times for attacking and defense? In fighting games it is really easy because it is easy to tell what the opponent's options are. In strategy games it is a lot harder. You've got to use scouting and such to tell where your opponent is - if he is defending, you've got to work on expanding or building up forces - you don't want to attack an opponent who is defending. But you aren't going to know that for sure, so you've got to use the best information you can. Use scouting and intuition as best you can.

Sometimes intuition is your best friend, but you've also got to know choice scenarios. A good example is if your opponent is attacking, and you don't have enough forces to defend. This happens a ton in high level SC. Defending is stupid here - you should go right for the throat and attack at a weak spot to try and deal as much damage to him as he would do to you. This is super common in high end SC - a weaker player will take the advantage from a stronger player by waiting for the attack, and attacking hard at a weak spot. In this way, the match stays even, or the weak player takes the lead, since he invested less army.

In the same vein, sometimes if you see the opponent poised to launch a counter attack (his army is placed where yours is weak, rather than where it is strong) you might want to feign an attack, but then pull your guys out and wipe out his army, hopefully while protecting whatever it is that he was attacking.

I should also bring in some poker philosophy too since it is really relevant here as well. Baiting is practically the name of the game in poker, you should almost always pretend you are dumb and weak. It lets you get away with really strong, winning hands because everyone things you suck and will just call on whatever.

If people can't read you very well (expert lying!) then it'll screw with their risk versus reward ratio. I am not so good at this acting like a moron thing, but I am fairly good at hiding whether or not I am bluffing. Bluffing is always risky, while playing on a strong hand is 99% not risky. Therein lies the problem, though - sometimes when you're bluffing you screw up and lose big, so you gotta know when to take that risk.

When the first call goes down, if you are planning on seeing what people have, you need to do something crazy to make people think you've actually got something good. I'm not saying necessarily you raise. If you do, you shouldn't be too gutsy about talking about it - be as soft-spoken as possible.

Screwing with people's scouting is a good way for them to make bad decisions, like folding when you're bluffing. That's the point of bluffing, obviously.

Predictions will do you in, though. At some point, some fool will call your bluff if you keep raising quietly, especially if you are in the lead. So, of course, you need to raise quietly when you are strong sometimes too. Mix it up?

But yeah, this is less about mindgames and more about decision-making. If you think the people at your table have been properly conditioned, then you can start bluffing. Otherwise, it's intuition time.

I might as well talk about conditioning. I talked about it a little in my mind games article. The idea behind conditioning is simultaneously that you learn the opponent's habits and punish them (they're already conditioned) and that you give them new bad habits to exploit.

A good example to this is when I played Fuerte against my friend Mark (in SF4). Mark got owned over and over by crossup splash (he'd never fought against anything like it before) until I mentioned to him how to block crossup splash. I did one more crossup splash, then did a really ambiguous front splash, followed by many more. He was totally confused and had no idea what to do.

If you always do a certain (typically low-risk) move after a certain sequence of moves, you can royally screw with your opponent's head by doing something unsafe that beats the counter to your low risk move. The simplest idea to this is to do a series of moves, then truncate it somewhere and throw instead. You could also mix in some slow, unblockable move. You could also play very safe, then randomly bluff (in poker). These sorts of conditioning methods let you 'predict' your opponent's moves for free.

Conditioning works on almost everyone, even people who know the power of nothing.

In StarCraft though, conditioning is harder. You're generally better off doing 'little' things that end in a big advantage unless your move is some sort of surprise free win attack (like dark templar rush). If your opponent knows you really like things like dark templar rush or barracks rush, you can give him a scouting tax if you decide not to do those things (since they will have to play towards preventing them). This requires a lot of foreknowledge, though, which is probably less available information to you than the amount of scouting and knowledge in a SF match.

So in SC, you really have to know when to make those gambles work. If you put your cash into a DT rush, you really have to have intuition and know the opponent isn't going to run detection early.

Someone will ask me "how do you get better intuition?" My suggestion is to play fighting games, that have lots of decision points.

No comments:

Post a Comment