Slartucker
This is a sound strategy, but as you noticed in the previous game, you can't always be sure of what your opponent will do. Sometimes there will be a situation where there really is one best move, and any deviation on their part will be worse for any of the moves you are considering. More often, when you are facing a creative warlock, there are five or six moves that don't completely suck, of which maybe two or three are more likely but not guaranteed. And this is why many players find heuristics like "don't stab" or "enter W when charmed" to be extremely helpful -- if you don't foresee every possible move your opponent could make, they keep you protected against the majority of fatal possibilities. What "don't stab" really means is "It's usually useless to stab, and in a few situations it could be useful but there's always large risk involved, so don't do it without carefully weighing the odds." Han Solo would be a poor Warlocks player.
Awall
The problem with always playing the odds is that it makes you predictable, and when you're predictable you get whomped. I briefly worked through a few examples of situations where I felt each player had one play that gave him the best "odds," but all of these situations had enough wiggle room that if either player could predict his opponent's move and counter appropriately, he could come out ahead. If nothing else, I think most situations involving a Counterspell and a Bolt/Fireball/Antispell can swing the game a good ways in one player's favor if he is able to get inside his opponent's head.
Slartucker
Yes, definitely. If there is a statistically "best" option, it's a good choice to default to, but if you are too predictable, you can get taken advantage of there. So playing the odds does mean playing, and not just "solving."
You have to keep in mind, however, that sometimes being taken advantage of still leaves you ahead of your opponent, and so being taken advantage of may in some cases be less of a disadvantage than opening up even a very small chance of something worse happening.
A good example is the PSD juncture against your monster when you have a disruption or counter to defend with. This is pretty common, but I think there are probably 1 or 2 games in the past 18 months in which I've protected myself and not the monster. This makes me extremely predictable and easy to take advantage of. But if my giant, troll, or even ogre gets charmed just once, I could very easily lose the game because of it; whereas me being charmed is usually easy to recover from when a monster is beating on my opponent. It's still much worse than bouncing the charm entirely, so if you look at things in isolation, you'd expect the optimal choice over time would be to vary your protection, maybe favoring the monster. But in context, I can commit to accepting the charm person every single time and almost never lose games over it, so that's almost always what I do. Does it make my monsters less effective? Yes, it certainly does. But do I still have an advantage with them? Yes, and that's what matters, not winning by a mile, just winning.
Awall
I've seen you anticipate me in our previous games. This time through, after I figured out the best play for you and what I could do to beat it, I tended to second guess myself -- "Waitasec, he's going to expect me to do something like this, so he'll make a move that beats the one I just came up with... so I'd go through and run the process again to try to figure out the new move to beat.
The problems were 1) It's much harder for me to predict this second level thing than just to find the obvious play for an opponent; and 2) You tended to do the obvious play anyway, and punish me for second guessing my original play.
Slartucker
Yes. That kind of recursive thinking is very useful to explore. It can make or break close matches. But remember, I'm going to play all the odds. I'm unlikely to tailor my move to what I expect you to do unless that custom tailoring also happens to work well for the other possibilities.
In my strategy piece I wrote "the game is wholly reactive; the minute you forget this, let your ego slip in and say "I want to do X" your opponent has the advantage." But the same thing is also true if you let your ego slip in and say "my opponent is going to do X." You have to react to constellated possibilities and not to predictions.