My "all time" stats:
Played 145 times
Max credits 49
Max combo 8
Win 56 (39%)
Draw 35 (24%)
Loss 54 (37%)
Played 94 times
Current credits 45
Max credits 49
Max combo 8
Win 39 (41%)
Draw 25 (27%)
Loss 30 (32%)
When playing I was not exploiting any assumptions about it being Markov learner (I don't even understand how it learns and how to exploit it)
While it can not learn to beat me most of the time, it is playable.
How do you know it learns with markov chain? How exactly? (What states the chain has? Other details?)
Why does camera shakes at every move? To me that's very annoying.
The game needs stats of the total credits I spent including previous rounds, so that I see the total spent/won balance. I am clearly winning on that measure.
Firstly, thanks for playing!
> How do you know it learns with markov chain? How exactly? (What states the chain has? Other details?)
I wrote it! It uses five chains of different lengths, and each of those estimates the next, then a standard way of resolving those to a single signal.
> Why does camera shakes at every move? To me that's very annoying.
Yeah, it is. The first version you actually spun around on the spot too, which was cool but made you sick, even on a phone. Therefore I hacked it to this, but it's not great.
> The game needs stats of the total credits I spent including previous rounds, so that I see the total spent/won balance. I am clearly winning on that measure.
In the menu there is "statistics" which contains more stats than live stats, and that might have more of what you want!
> It uses five chains of different lengths, and each of those estimates the next, then a standard way of resolving those to a single signal.
Can you explain on more details? (if not a trade secret of course)