What's the deal with this whole Game, Set and Match thing in Tennis ?
I mean if I want to know who is the best at hitting balls why not just tally the total number of point.Research and LLMs give some mildly convincing answers, like :
- Dividing the match into games and sets gives players chances to reset and regroup
- A player can lose a few points but still win the match
- Sets allow for comebacks, which add drama
So what really makes the usage of Sets different?
One thing stands out: Sets seem to make some points more impactful than others.
But it made me wonder... Is that even true? And if so how much?
So I made this ! (WIP)
A Simple Yet Tedious Game
- Each player has an energy pool used to make moves.
- In each round, the player chooses how much energy to use.
- After each round, both players recover 1 energy.
- In each round, one of the players is randomly awarded a point.
- The first player to reach 3 points wins the set.
How the winner of a point is determined:
- Let E = energy used by the player.
- The dummy always uses 1 energy.
- A random number R is chosen between 0 and (E + 1).
- If R < E, the player wins the point.
- Otherwise, the dummy wins the point.
This means the probability that the player wins the point is E / (E + 1).
- You play against a dummy opponent.
- The dummy always uses 1 energy per round.
- The game doesn't "end" — it's just a toy game for trying out different strategies
Feel free to keep playing but be warned: even with the optimal strategy, your edge over the dummy is so small that playing manually could take hundreds of round before gaining a meaningfull lead. You might want to head over to the next tab sonner rather than later.
Try it!
Let a bot do the work
Tell the bot what to do and see if you can find a good strategy off the top of your head
Can you consistently beat the dummy?
Let's tinker!
Use the botID to simultaneously evaluate a lot of different strategies at once.
Get a feel for what could be the best strategy.
Forget the tinkering
What are we? Macro Data Refiners? Let's automate the tinkering.
At this point, it's not much of a game, so here is a solution. Feel free to discard it and find your own.
Or even better, maybe you can find a better one?
[ [ 1, 1, 1 ], [ 1, 1, 1 ], [ 1, 1, 1 ] ]
What if there are more points in a set ?
energy: winrate:We may need a better way to find the best strategy when N gets bigger...
What if there are more points in a set ?
We may need a better way to find the best strategy when N gets bigger...