The Occasional Jam No. 28

No. 28 strategy

Getting Better at Pot Luck

6 min read

Pot Luck is five games of chance, and in three of them no pick beats another. Your points are made in the two where the odds pay, plus one coin call that can change your chances at a scoring tier. The other three are for enjoying.

It assumes you know the rules. If you don't, the how-to has them, with an example animation. Some figures below are read straight from the game's scoring code. The rest come from simulated players run through the game's own engine on many thousands of made-up days, never a real one, and those say so.

What the score rewards

Of the day's 300, the three games of pure chance, which come first, are worth 120 between them, and higher or lower and dice rolls are worth 180. Most of the pot sits in the two games where knowing the odds pays.

Of 120: what the three games of chance average
52.5
For everyone, whatever they pick. Read from the scoring.
Of 80: higher or lower, called by the odds
About 61
Simulated. Guessing averages 40.
Of 100: dice rolls, kept by the odds
About 60
Exact, over fair dice. Stopping on the first roll averages 41.

Add those up and a day played by the odds averages about 175 of 300, and a day of pure guessing about 135. Even so, guessing came out ahead on almost one day in five in the simulation, so on any one day the odds can still lose to luck.

What each scoring tier asks

The points each scoring tier takes move with the day's cards and dice, and the scale at the top of the board shows that day's numbers. On four days in five, Jam takes between about 140 and 170 points, Cat between about 160 and 195, and Wow between about 175 and 215.

How often each scoring tier was reached, simulated

                   Jam  Cat  Wow
By the odds        82%  49%  18%
Guessing           28%  13%   5%
By the odds means every card call and every keep of the dice made by the odds, and staying on a right first coin. The three games of chance are played the only way there is.
Guessing reaches Jam on a little more than one day in four. Playing the odds reaches it on about four in five, and Wow on fewer than one in five.

A day played by the odds still misses Jam about one time in five, and Wow takes luck on top of good play. The wishbone alone moves those chances a lot. On the half of days when it snaps the other way, playing the odds still reaches Jam about 71% of the time, but Wow only about one time in twenty.

The three games of chance

The wishbone, heads or tails, and rock paper scissors are pure chance. Which half keeps the top, how each coin lands, and the day's three throws are all drawn before you play, with even odds. No pull, call, or throw beats another, so a favorite throw does exactly as well as mixing them up.

So enjoy them for what they are. Pull the side your hand goes to, call heads because you like heads, and throw what you'd throw at a friend. Between them they average 52.5 points for everyone, and your Track record keeps how often your wishbone kept the top, which drifts toward half over a long run of days.

Stay or double

Heads or tails holds the one choice in the three. After a right first call, staying at 20 and going double or nothing for 40 both average 20 points, which is why the how-to calls them even. A scoring tier isn't an average, though. Doubling makes the pot swing wider, which lowers your chances at Jam and raises them at Wow.

Stay or double, simulated

                   Jam  Cat  Wow
Always stay        82%  49%  18%
Always double      77%  47%  22%
Double when behind 79%  50%  22%
Behind means the wishbone snapped the other way.
The same coins in every row, and everything else played by the odds, so only the choice differs.

Stay if you're playing for Jam, then, and double if you're after Wow. Doubling only when behind did a little of both: it reached Wow nearly as often as always doubling and Cat a little more often than either, and gave up less of Jam than always doubling. Whichever you pick, the coin is still a coin.

Higher or lower: read the card

Higher or lower is where the odds pay most plainly. The next card is never the rank of the one showing, so an ace or a king is a sure call: everything left is above an ace and below a king. From there the odds fall off toward the middle of the deck.

How often the odds call is right on the first call, by the card showing. Read from the game's own count of the deck.
How often the odds call is right on the first call, by the card showing. Read from the game's own count of the deck.
ItemValue (%)
Ace or king100%
2 or queen92%
3 or jack83%
4 or 1075%
5 or 967%
6 or 858%
750%

Call higher on a 6 or under and lower on an 8 or over, every time. Even a 6 starts 28 higher to 20 lower, and by the last call only three other cards are out of the deck, never enough to close that gap. The one card worth a second look is a 7, which starts even, 24 cards each way.

The board counts the cards left in the deck that are higher and lower, and the how-to's tip is to call the side with more. On a 7, though, the two counts are never more than three apart, so either call is close to a coin flip. Played this way, a little over three calls in four come right, for about 61 of the 80 points.

Dice rolls: what to keep

Short of five alike, no sum reaches 60, so a lil run, at 60, already beats the best sum.

Made-up first rolls, and what the odds keep

3 3 3 5 5
A full house, 75. Stop: every keep from here averages less.
2 3 4 5 5
A lil run, 60. Keep 2 3 4 5 and roll the spare 5. A 1 or a 6 makes a big run, and the lil run can't be lost. About 74 on average.
6 6 5 5 4
The sum, 52. Keep both pairs and roll the 4, for a full house on a 5 or a 6: about 63 on average, against about 57 keeping 4 5 6 for a run.
1 1 3 4 6
The sum, 30. Keep just the 3 and the 4, and roll the other three. The middle of a run is the best start, at about 58.
Each with two rolls left, as the game's own odds play keeps them. The averages count both rolls.

A die you keep never changes, so keeping the dice that make your pattern and rolling the rest can't lose the pattern. Stopping early pays only on a full house, a big run, or five alike. With anything else and a roll left, rolling again pays more on average, even though it can lower the sum.

For points, stop on a full house or better. Anything less, keep what makes your pattern and roll the rest.

Played this way, the dice average about 60 of their 100, against about 41 for stopping on the first roll. A hand ends on a run about 43% of the time, a full house about 18%, five of a kind about once in a hundred, and on the sum the rest of the time.

The last game: know what you need

By the time the dice come up, four games are in, and the scale at the top of the board shows your pot and how many points each scoring tier takes that day. Subtract your pot from the one you're after, and you know what the dice need to pay. Now and then that changes the right keep: sometimes it says stop early, and sometimes it says roll on.

Two made-up last games, played for the scoring tier

Need 50 for Wow · roll 6 6 5 5 4
The sum, 52, is already Wow. For points you'd roll the 4 for a full house, but a 1 or a 2 could leave you short. Stop: there's no scoring tier above Wow to roll for.
Need 60 for Wow · roll 6 6 6 6 2
For points, keep the four 6s and roll for a fifth: about 68 on average. But no sum reaches 60, so only a fifth 6 gets you there. Rolling all five again gets there more often, about 47% of the time against 35%.
The keep for the most points and the keep for the scoring tier aren't always the same.

In the simulation, playing the dice exactly for the scoring tier, rather than for the most points, reached Wow on about 22% of days against 18%, and Cat a little more often. It cost about two points a day, and reached Jam a little less often. Together with doubling when behind, Wow came up on about one day in four.

For what it's worth, I usually double on the coins, but not always. It has definitely cost me meaningful points on some days.

And one note about the scoring tiers. The cuts for Jam, Cat, and Wow were set by hand, before there was real play to tune them against. If one feels wrong, that's my number and not your play.

Sources

  1. Pot Luck's scoring code The points each game pays, the dice patterns, the scoring tiers' shares, and the first-call odds by card are read from it when this page is built, so they move if it does.
  2. An exact search over the dice Every keep weighed over every fair roll, as the game's own odds play does, for the dice's best average, the worked keeps, and how often a hand ends on each pattern.
  3. Simulated players, run through the game's engine Made-up days drawn at random by the engine's own simulation deal, never for a date, each judged against its own random plays as the game judges a real one. Every player was given the same pulls, calls, and throws, so the rows differ only by the choice compared. The comparison with guessing is the game's own harness.