Figgie: Simulate stock trading with a deck of cards | Hacker News Reader

Figgie: Simulate stock trading with a deck of cards

figgie.com

122 pointsKortaggio24 comments

FTA: "Figgie is a card game that was invented at Jane Street in 2013. It was designed to simulate open-outcry commodities trading."

This is crediting Jane Street in 2013 as if Pit didn't exist since 1904:

"Pit is a fast-paced card game for three to eight players, designed to simulate open outcry bidding for commodities. The game first went on sale in 1904 by the American games company Parker Brothers."

https://en.wikipedia.org/wiki/Pit_%28game%29

Certainly official mechanics vary, but there were always Pit variants for regular decks, and as long ago as the 1970s and 1980s there were rule variants that play like the linked game.

Parker Brothers' Pit cards were ornate in 1914, featuring the Bull and the Bear:

https://americanhistory.si.edu/collections/nmah_323758

I’m not convinced Figgie is similar enough to Pit to make this comparison. Both are the same genre certainly but there are substantial differences in the setups and goals.

Elements unique/different to Figgie - uncertain value of cards (need to find the 12 card suit) - time limit - scoring mechanism

I came here to comment the same info. Pit game is actually pretty cut-throat and this immediately reminded me of it. Replace Wheat with Spades. Same game.

It doesn’t seem that similar to Pit… the key with Pit is that you make trades by # of matching cards, not what they are.

But they are trying to simulate the same thing, and Pit is a fantastic game, even after more than 100 years.

That's a very interesting game mechanics.

It wonder if it can be adapted for playing with paper cards. One variation can be this one: shuffle a 40-card deck, discard 5 face down, it guarantees that there will be an unknown short suit (with 8 to 5 cards), the goal suit will have from 7 to 10 cards. Deal the remaining 35 cards, and start trading. With 4 players one player will be one card short, so they need a compensation like being able to reduce ante by some amount.

The balance of the game with 35 cards and random suit length will probably be different but the trading mechanics should be able to address that. So ante and payout per goal card may have to be changed.

Sort the cards into suit piles. Turn them face down. Shuffle the suit piles until you don't know which is which (maybe challenging). Take off 1, 3, 3, and 5 cards from piles. You'll have the deck as described.

Two player blind shuffle. All players except one turns away. Remaining player shuffles. All players except a different one turns away. Remaining player shuffles. Now no one knows which pile is which.

> All players except one turns away.

A bit trusting!

Assuming at least 3 players, only have 1 player turn away during each round of shuffling.

Shuffling piles is difficult to do in a way that nobody, including the dealer, knows which is which. And it requires more table space. It also requires an additional shuffle of the final deck. If the final shuffle is not good enough you can guess which one was the short suit by looking at the sequence of the dealt cards (the deck was not cut, and the last 2 cards are the same suit? chances are the dealer is bad at shuffling and it's the short suit). Maybe passing a tray with four piles and letting each player to swap two piles without letting other players see which piles were swapped is good enough. But it's still fiddly.

Normal shuffle and discarding face down is easier to explain and faster to do. And it can be done where the table is not available or is not big enough. Starting with a 44-card deck and discarding 5 down to 39 should make it more similar to the original 40-card version, but 39 is not divisible neither by 5 nor by 4, so one player will always be one card short.

Being one card short might not actually be a problem in practice if the game is played across several rounds. It may even add some interest to the game by creating an asymmetry.

I love how detailed this is but the engineering problem has been solved with card shuffle machines. You can buy one on Amazon for $20. Put the cards face down and hit shuffle (repeat a few times).

I think you can assume decent shuffling for the sake of the problem. And with cards dealt around, you don’t see two cards in a row. You’d have to be pretty bad to get usable info from cards at position n and n+4 being the same suit.

edit: People can be pretty bad at shuffling though. Overhand shuffles are awful and info would be available. But I think this game naturally sorts the cards pretty well by the end of gameplay, so even the method proposed before me would be vulnerable to info from bad shuffles.

It's interesting that they say skill is about finding good trades -- my perception from playing a few times is that there's a lot of subtlety in how many of what you offer, and how the information that you have those cards changes others views about which suit is the goal suit. A successful strategy I've pursued playing with other friends who have never played before is just selling all my cards for >=$5 and making a small, consistent profit.

The images show cards with just the suits on them. Is that how the game is usually played?

It seems that playing with a normal deck (suits and ranks) would change the gameplay considerably. For example if you see the A-2-3-4-5-6 of one suit, trade them away, and later see the 7-8-9-T-J, you can be certain it's the 12- card suit. Or more realistically you can do a probabilistic version of the same reasoning by seeing some cards and asking other players questions like "do you have the 8 of diamonds".

No ranks, just suites. The idea is that stocks are fungible. Stocks don't have ranks, so why should a card representing them have ranks.

Also the numbers would leak far too much information on the goal suite.

So at Jane Street they made custom decks for this, they didn't start with a regular deck?

That's correct. I assume it was initially played with a standard deck but they now have custom decks.

That's a valid observation, but if getting a 12-pack of identical decks is not a problem, it's enough to build 13 custom decks (all Aces, all 2s, etc).

Edit: if the players can announce only what suit they are buying and selling but do not show cards to the others, it may end up difficult to deduce what cards are missing, because only the buyer and the seller know the value of the traded card.

Is this playable with physical cards in practice? It seems that there is a lot of handling of money, more than cards, which would be impossible to do with any speed.

These seems close to the game Pit.

How do you figure out which suit is the goal suit?

One way to be sure is you could accumulate 12 hearts and then be certain the goal suit is diamonds.

In practice you'll have to guess with much more limited information.

The use of knowledge in society - Hayek

https://www.econlib.org/library/Essays/hykKnw.html

I’ve played a few rounds and the group is often surprisingly good at finding it. Players start with their hands and set prices accordingly. If the other players agree, the price difference gets more extreme quickly.

Reply on news.ycombinator.com