Irrationality


came across a very interesting paper regarding human irrationality and self-defeating behaviors.

http://www.cepr.org/meets/wkcn/3/3503/papers/Baumeister.pdf

do be advised that the paper does have a very 'special' formatting quirk - every 10-12 words, two words will be joined. so every line or so, you'll see something likethis. just wanted to give a heads up.

enjoy!

This sort of thing usually vegetates in my subconscious for a while before a good design comes out and then I write it. If that happens, you will be the first to pre-alpha test it


One of the things that I once did was glue together a bunch of eSignal tick data files over months at a time and play those backs. If you download tick data from eSignal for the ES for 10 days I think that the file is about 40Mb in size so if you glue together a 1 month file it comes to about 100Mb so it's not small. (Unfortunately I lost a hard disk with all that on it and didn't have a backup so I've lost those files.)

Anyway, the idea was to allow uninterrupted playback of around 20 to 25 trading days and just focus on taking setup after setup and see if anything came of it. As mentioned above, I couldn't get it to a sophisticated enough level to make it worthwhile.

Now here's an idea that's just flashed into my head:

Why not use the market data to move an object (space alien / target etc.) up and down on the right hand side of the screen. The historical tick data determines the speed and distance that the object moves. You, the player, need to keep your character just ahead of the next move that this object makes... this part still needs to be thought out...

So, the objective of this little game is to train your subconscious into understanding how the alien is going to move. If the market does more in a predicable pattern then after playing the game several times you will become better at guessing the aliens next move.

I think we need to give this serious thought. I think that there's something in here.
quote:
Originally posted by day trading

This sort of thing usually vegetates in my subconscious for a while before a good design comes out and then I write it. If that happens, you will be the first to pre-alpha test it


brotha, you KNOW i will be the first in line

quote:
One of the things that I once did was glue together a bunch of eSignal tick data files over months at a time and play those backs. If you download tick data from eSignal for the ES for 10 days I think that the file is about 40Mb in size so if you glue together a 1 month file it comes to about 100Mb so it's not small. (Unfortunately I lost a hard disk with all that on it and didn't have a backup so I've lost those files.)

what about a self-generating data source? kind of along the lines of NT's simDOM. but more naturally random (you can control the trend too easily in NT). the method and algorithim may be the initial obstacles, but that way you wouldn't have to keep or download HistData on a regular basis. would also keep the overall app size smaller. either have user inputs for things like ATR, ADR, % trend, 60-day H/L, 180-day H/L, ADV, wiggle room, and so forth. they could either base it on a real instrument or organically create a mythical instrument. essentially, write something with all the inputs that essentially allow the user to 'grow' data from inputs they provide.
quote:
Anyway, the idea was to allow uninterrupted playback of around 20 to 25 trading days and just focus on taking setup after setup and see if anything came of it. As mentioned above, I couldn't get it to a sophisticated enough level to make it worthwhile.

i agree. seeing and acting on setup exposure is critical. that's why i theorize the data itself (and therefore the instrument) is of marginal concern. if you sell VAHs, it won't matter how you came upon a VAH or on what instrument - what do you do when you are confronted with one?
quote:
Now here's an idea that's just flashed into my head:

Why not use the market data to move an object (space alien / target etc.) up and down on the right hand side of the screen. The historical tick data determines the speed and distance that the object moves. You, the player, need to keep your character just ahead of the next move that this object makes... this part still needs to be thought out...

any chance of incorporating setup specific parameters? for example, basic MP areas (VAs, IBs, POC), MAs, fib levels/retracements, and so on? maybe have an alarm when the Price Invader hits an area - buy signal alarm, sell signal alarm, profit signal alarm, (annoying) stop signal alarm, and so forth.
quote:
So, the objective of this little game is to train your subconscious into understanding how the alien is going to move. If the market does more in a predicable pattern then after playing the game several times you will become better at guessing the aliens next move.

agreed. i think the key components would be: pattern recognition and execution. just drilling it in over and over and over and over again. so that the actual 'trading' part of traidng, the physical, monkey-brain part of trading is just a mindless action. a reflex. the real work is ideally done before the market opens and after it closes.
quote:
I think we need to give this serious thought. I think that there's something in here.


again, i agree. i'd love to hash out some ideas about this and really look at what we (uh, sadly, that means you) can put together.

along the lines of a product that can 'grow' its own data, while it may take some initial setup by the user, eventually they will have 'modules' or 'worlds'. that way, they don't have to recreate a 'world' every time. if they want to trade a ER2 or DAX like instrument for a while, they can select that 'world' or 'module' that they built and the trading 'game' will grow from those specific inputs. if they want to trade something more like ZB they can create, build, and save that 'world' and choose it at their whim.

also, i wouldn't think the graphics would have to be very intense. charts aren't typically very extreme (though i have seen some charts with a severe case of Indicator Overload). not sure how willing either would be to help (i.e. share for free
) but it may be useful to look at IRT or QT since they seem to have some of the least resource intensive charting software and the most user configurable software around. maybe NT would help out as well, since they already have a pseudo-randomized (is that an oxymoron?) simulated data generator.

ideas, just popping into the noodle, through the fingers, and onto the screen, that's all they are ...

tag, you're it!!
and have dynamically varying 'pit noise' based on price fluctuation + volume fluctuation.

and keep 'profits' off-screen. trade in points. if a trader wants to see his profit, he has to go to a non-trading screen to see it. if ya wanna be a real hard-a** about it, make it where you can't be in a trade and look at your P/L.

One could not use randomly generated data because the outcome of a random stream of prices will, in the long run, be 50/50 in either direction. The whole point in trading patterns is that a pattern produces a non-random semi-predicative effect. i.e. The market will have a higher than 50% probability of moving up/down due to what has already happened. A random price generator would negate this effect and so you are forced to use historical data here.

I'm picturing this game in my mind now. I see a smiley face about 3/4 of the way across the screen (from left to right). To the left of this face is a trail of where it has been which is obviously the chart that we currently see. The smiley face is on the hard right edge.

To the right of the face you control a short horizontal bar with the mouse. If you think that the price is going up you click above it and the bar moves up and any up move from that point forward adds to your score (and subsequent drops subtract etc.).

Your score is a kind of lifeline or energy counter. So you can be long, short or flat. However, there's a catch here. If you're flat then you are still bleeding energy which simulates your fixed costs.

I wonder if this sort of thing would fit inside one of those MMORPG's? Have you ever played one of those games?
you're right, random may not have been the best term. but if the user can 'build' data structures, you wouldn't have truly random data. it would be dynamic though and grown from the input. but i like your idea as well


as for the MMORPG, i don't play them but my wife and son play toontown (play.toontown.com) and my son just did a 10-day trial of World of Warcraft - both are massive multiplayer games. i think if you could do a MMO version that would rock! especially if you could essentially create a simulated market. i mean, you can could really go gonzo with this nugget. ET alone has over 50k members. plus, you would have non-traders subscribe, just because of the type of game it is. keep it rollin ...

ugh, forgot to add this: i got the idea for 'growing' data and setups and essentially growing 'worlds' came to me from seeing demos of Spore. if you haven't heard about it or seen it, it is the upcoming game from the creators of The Sims. not a gamer or a coder, this stuff looks completely revolutionary to me. when you have a bit, look for some of the demo videos of the the game. there are some of the creator demo'ing the game and talking about it as he's playing it. very awesome stuff to me. again, me, the non-coding, non-gamer. the available applications of this sort of approach are intriguing to me. trading, flight, police, fire, military, networking, medicine, engineering, and, of course, gaming.
quote:
Originally posted by omni72

and have dynamically varying 'pit noise' based on price fluctuation + volume fluctuation.

and keep 'profits' off-screen. trade in points. if a trader wants to see his profit, he has to go to a non-trading screen to see it. if ya wanna be a real hard-a** about it, make it where you can't be in a trade and look at your P/L.


Looks like our conversation taking place here in real time is faster than trading.

I think you'd want to give the player all options because you'd want to get it as close to real life as possible and so the player could do in the game what they could do in real life. If a game player develops habits (such as not looking at the profit) that improve his/her score then they are more likely to carry those back to the real world. In the game you learn what works and what doesn't.
quote:
Originally posted by omni72

...i think if you could do a MMO version that would rock! especially if you could essentially create a simulated market...


Well the idea would be that in one of these MMORPG's you could have a plug in module that was an exchange on which you could day trade or swing trade or whatever to gain extra cash to build your armies etc.

If you pulled CME or CBOT into it then they may be willing to supply historical data to the game so that the game replicates what trading was really like in the past. Players would not know what day they were about to trade because it would be randomly pulled out of a data base of thouseands of days and would not have any date attached to it.
quote:
Originally posted by omni72

...my son just did a 10-day trial of World of Warcraft...


I have seen WoW at a distance and surfed their site some. How did your son find it?
quote:
Originally posted by day trading

quote:
Originally posted by omni72

...my son just did a 10-day trial of World of Warcraft...


I have seen WoW at a distance and surfed their site some. How did your son find it?



well, he's been playing TT (solo) since he was 3yrs old (he just turned 6 in May) and he finds all kinds of games. mostly from nick.com, disney.com, cartoonnetwork.com, wildgames.com, and a bunch of ceral sites (nabisco, etc). when he found WoW my wife asked some of her TT friends what they thought of it and many were very enthusiastic (cuz, just like on TT, they are WoW addicts). we dl'd the demo (about 3gig
), signed up for it, and he built a whole stable of characters. i think he had a couple of characters up to level 8 or 9 and several others at between 5-7. i know it goes up to level 60, but i was pretty impressed for the number of level 5+ ghouls he had built in the 10 days.

so, there ya go
He sounds like a genius. Perhaps he should be trading for you...?
Is this Blair Hull from Matlock Capital? www.matcap.com ?

That's a good quote by the way.

I couldn't get to sleep last night and so I rummaged through the library and found an Anthony Robinson book that I have read the first few pages of (Awaken the Giant Within) and have always intended to read but just not got to it.

I flipped through a few chapters and came across a section called the Alpo Diet. Apparently Alpo is a brand of dog food. The story goes on to tell of a couple of women who had failed every diet and finally succeeded when they committed to eating a can of this dog food if they strayed from the diet. Whenever they thought about eating something not on their diet plan or giving up on it they'd pick up this can and read the ingredients which included chunks of horse meat. Apparently the diet worked.

So coming back to what pt_emini quoted Blair Hull saying, we need the discipline and if we do not have the discipline then we need something to trap/force/cajole us into a disciplined mindset. One from which we won't stray.

I know that I've discussed this before but I feel that a paper trading system could be made far more effective if you could associate some sort of punishment with any break of discipline or rules in the paper/practise trading. I've often joked about wiring the trader up and shocking them which is obviously not practical and dangerous.

Perhaps a timeout chair/corner. I you break discipline you commit yourself to this chair and meditate over what you did and why you did it.

Unfortunately there's nothing forcing you into that corner. No big stick waving over your head and no dog food to eat...