Sep

11

 Victor,

Hope everything is well with you.

I've been tempted to sell that Mark Twain gold coin you gave us one Xmas. At current values, it could finance a round trip flight for Janis and I to Italy. But if it keeps up its current pace and housing values continue to crumble, we probably can buy a second house with the one coin in a couple of years.

I had an idea for you to explore. I doubt if it is "Daily Speculations" - worthy, but i wonder if it has some bearing on your trading theory of "ever changing cycles."

It's based on this game.

I call it RPS intelligence. if you can consistently beat the computer at that game, you might be a good candidate for making money in the stock market. The computer detects patterns in your play and bases it's choices on your previous play. Substitute stock market (or S$P futures trading floor) for computer and you have a good game based on your "ever changing cycles" theory. That is, if investors become too reliant on a trading strategy that what works in the past, the computer or market detects their pattern of play and beats them.

If someone has strong RPS intelligence, they might do well as a trader in the market.

Anyway, I hope you are doing well. We've been expanding our book business so fast that we have to hire a couple of people. Having to manage people is not my strong suit, but maybe Janis can keep employees happy.

Best,

Neil

Newton Linchen comments: 

I always enjoyed this game, although I was never good at it. So I took the challenge to play with the computer.

My best score so far is 46 x 36, with 33 ties. This is the 10th time I play in the past few days, and the other 9 times I was "cleaned" by the computer.

I don't know whether this relates to markets or not, but the thing is the computer "learns" your (mine) style and adapts to it. So in order to beat it, even with a small amount of winnings, it's de rigueur that one constantly monitors when the computer starts to win hands it didn't win in the previous throws.

Consider a tie: paper-paper. My first (and dumbest) strategy was to attack: next hand I would give a scissor. It responds well to attacks, in fact, it seems by default to expect that the player will attack. Therefore, I soon discovered that the dumbest strategy was to attack the computer's last hand. If it's hand was a rock, you couldn't win throwing a paper, and so on.

So, in the event of a tie, I changed my strategy to playing again and again the same hand (say, "paper"), and so the computer. It usually goes three ties until it takes the attack. But it would attack (change from "paper" to other) counter-attacking: he would throw rock (expecting you would have attacked with scissors). Therefore, the strategy of waiting for the computer to attack was a win.

But it learned this move too…

And afterwards in the event of a tie, if you played the same hand, it would start to attack using the direct approach: giving you a scissors.

And that was an example of the multiple learning processes the computer went through, beating me in the previous 9 times.

(It was starting to annoy me: was it reading my mind? : )

I tried the random approach, choosing paper, scissors or rock disregarding the last hand. Say, throwing paper-paper-scissors-rock not even taking in account what was the computer's last hand.

It worked. For a while… (It seems the computer was also good in dealing with this).

This last game, when I won in 108 throws, maintaining a +9 score during the entire play, I tried three different approaches:

First, trying to lose to myself: If my last hand was a paper, I would throw rock in the next hand (rock loses for paper). Bingo. It worked. Then again, and again, and again. Example: paper, rock, scissors, paper, rock…

When the computer started to adapt to it, I would play "losing" to it: if the computer's hand was scissors, my next hand was paper. And it also worked.

Also adapting to it, after a while, we would have a tie, or I would lose.

If I lost, I would choose at random the next play.

If we had a tie, I would alternatively: a) play the counter attack: tie = paper-paper, my next hand was rock. b) play the direct attack: chose scissors.

The only way I could manage to stay +9 over the computer during the game was alternating at random these three main strategies: losing-to-myself, losing-to-the-computer, choosing at random after a loss, and playing the two "tie" approach.

After a string of winning hands, I would change the strategy after one loss.

Then win or lose.

If winning, I would press that win keeping the strategy, until one loss. Then change again.

PS: The way I don't think this game relates to speculation is that we are trying to be the market, ever-changing, running away from the smart spec who is trying to figure us out.
 


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search