I liked this idea so much I wrote up a game. I have no idea if this would be fun or not. I have a feeling it would definitely be fun if you got rid of the bank, allowed borrowing with 10% interest from other players, and had players under $10+4*(number of races per year) at the end of the game pay players over that amount.
Basically players are making odds and placing bets on four racehorses; each player knows exactly which dice two of the horses will roll but doesn't know the other two, and no one knows the same two horses.
This isn't a board game, per se, but ah well.
Players: 4
Supplies:
4x d20
4x d12
4x d10
4x d8
20x d6
4x d4
pencils
paper
money (Monopoly money could work)
whiteboard & markers (optional)
Gameplay Overview: Each game consists of 4 years with some number of races in each (more for a longer game). There are four horses racing each year, but horses change from year to year. A race consists of each horse rolling 5 dice of unknown type. Two players peek at the results of each horse and write them down independently; the horses finish in order of highest sum to lowest. Before the race players announce the odds they are giving for each horse. They are guaranteed to get bets from everpresent "onlookers", but other players may also place bets. The player with the most money at the end of the game wins.
Setup: Each player gets a large dice cup, a pencil, paper, a d20, a d12, a d10, a d8, 5 d6s, and a d4. Players all start with 1000 dollars (write this down).
Year Start: Players roll all of their dice at once, hidden beneath the cup. A horse's stats are determined as follows: sort the d6s from lowest to highest. If the d4 is less than or equal to the lowest d6, the d4 is used. If the d8 is less than or equal to the second lowest d6, the d8 is used. If the d10 is less than or equal to the middle d6, the d10 is used. If the d12 is less than or equal to the second highest d6, the d12 is used. And if the d20 is less than or equal to the highest d6, the d20 is used. Secretly record your horse's stats. Then secretly record the stats of the horse to your right.
Race Start: Each player takes the role of a bookie. Secretly determine what odds you will give on this race by assigning each horse a number (integer) from 1 to 99. If a bettor pays the point value of a horse in dollars, he stands to win back 100 dollars. Your total assigned points must be equal to 120. Everyone reveals their odds at once. Set a time limit for this phase! Write odds on the whiteboard (if using).
Bets: First, each player gains 600 dollars from onlookers' bets (5 bettors for each horse). Each year has a different starting player. In order, you must place up to 2 bets with each of the other bookies, in amounts equal to the point values for the horses announced by the other bookies (you stand to win 100 dollars with each bet). Hand over the money and mark your bets.
Race: Hide your unused dice. Roll your 5 used dice under the cup. Record the sum secretly. Then record the sum of the dice on your right. Now everyone reveals their sum (resolve discrepancies by beatings). The horse with the highest sum has won this race. If multiple horses tie, relevant players examine the horses' lowest dies. Whichever has the higher lowest die wins. If this is also a tie, continue with the next lowest, etc. If every die is identical, roll these horses again and repeat (for just those horses that tied for first).
Payout: Pay 500 dollars to the bank; this was won by the 5 onlookers who bet on the winning horse. Now pay each winning bet to other players. If you cannot pay, or if you just want to, you may borrow money from the bank - this must be paid back with 10% interest during the next payout phase, even if you have to borrow to do so.
Game End: After the final race of the fourth year, total your cash. Subtract any money you owe to the bank (plus 10% interest). Whoever has the most cash wins!
This sounds a bit harsh.
Nope. Say you rolled as follows:
5d6: 4,1,2,6,4
d4: 4
d8: 5
d10: 4
d12: 11
d20: 5
Then your horse would use a d6 (4>1), d6 (5>2), d10 (4<=4), d6 (11>4), and d20 (5<=6): 3d6+1d10+1d20.
You get dice these percent of the time:
d4 - 39%
d8 - 31%
d10 - 35%
d12 - 38%
d20 - 27%
Interestingly, I'm not sure how to resolve discrepancies that last longer than "hey, check again" (i.e., lies). There must be a punishment, but only to the offending player, but you can't tell who that is without checking the dice, which invalidates the year, but then you could just keep invalidating the year until you got a horse you liked...