No more BS - What have you learned? | Page 29 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

What have you learned? 231 replies

What I have learned trading & How I developed my system 6 replies

It's been 3 years what have I learned? 82 replies

What Have I Learned Since 2006?? 1 reply

Attachments: No more BS - What have you learned?
Exit Attachments

No more BS - What have you learned?

  • #561
  • Edited 4:01am Sep 23, 2010 1:49am | Edited 4:01am
If you carefully read what I wrote, I said we didn't have the computing power at this point to defeat the top Go players. I didn't say never.

This is a minor point, but it wasn't the database that beat Kasparov in the second match. It was brute force number crunching and the developers modified the program during the course of the match which was why the results have been disputed (due to human intervention). As I said earlier, Chess is a tactical game, and these were tactical computations, so has nothing to do with the point I was making about trading.

I didn't say that the actual strategy employed to win at Go should be applied to trading. That would be silly.

The fact that most Westerners have never heard of Go does not relegate it to obscurity. Go is held in very high esteem in China, Korea, and Japan and quite a large amount of funding and time have been dedicated to developing computer programs to try beat the Grand Masters.

Also, I didn't say I held a "mystical" fascination for the game.

Please try not to base your points on statements I didn't make or coloring me as someone holding mystical beliefs.

Cheers

Quoting Sauron
[/color]20 years ago no computer has been able to beat Kasparov at chess. They have more power now but don't forget that it was not the power who beat him but the computer used the database with the previous games of him. So they devised a clever EA which looked at the old patterns and used them profitable.

Go has nothing special, maybe you like it more and find something mystical about it which can't be put in a formula or, even a formula could be devise, we don't have the computation power right now to beat the best players. We...
  • #562
  • Sep 23, 2010 2:41am Sep 23, 2010 2:41am
Quoting mikkom
Being programmer does not mean that you can think analytically. Actually programming is the easy part.

About PA, all the things you might call PA can be separated and tested separately for non-random effects after separation. You don't have to have a combination of random rules that you call PA to test if one element of it has a non-random effect. You just have to be clear to yourself what you actually are looking for.
I'm a good programmer so I think very analytically and tried some programing some really complex systems.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #563
  • Sep 23, 2010 2:48am Sep 23, 2010 2:48am
I think the chess was still more down to raw computational power they did use a mainframe which was the quickest ever to be fair and used the chess game to help sell it basically.

I've got a mate ( bad loser ) who played GO for years, I beat him on my first game the strategy is very basic compared to chess.

I actually wrote a chess game that kicked ass when I was 16 for kicks on a bbc in 6502, good times.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #564
  • Edited 3:48am Sep 23, 2010 2:54am | Edited 3:48am
Yeah, I think it was 200 million computations a second. ...I wish my computer was that fast.

With Go, the level of strategy is dependant on the opponent.
Quoting Turveyd
I think the chess was still more down to raw computational power they did use a mainframe which was the quickest ever to be fair and used the chess game to help sell it basically.

I've got a mate ( bad loser ) who played GO for years, I beat him on my first game the strategy is very basic compared to chess.

I actually wrote a chess game that kicked ass when I was 16 for kicks on a bbc in 6502, good times.
  • #565
  • Sep 23, 2010 4:59am Sep 23, 2010 4:59am
Quoting Fudomyo
Yeah, I think it was 200 million computations a second. ...I wish my computer was that fast.

With Go, the level of strategy is dependant on the opponent.

My opponent was all about patterns big huge ones, keeping it simple meant I totally kicked his ass, he had 4 centres or something and I had 40+.

I haven't played chess in years, don't like it on the internet though, got to be face to face and hard to find the time for that
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #566
  • Sep 23, 2010 10:21am Sep 23, 2010 10:21am
Quoting mikkom
Basically yes. Not necessarily executed but it must be testable by mechanical method.

I'm not saying that some people can't make a lots of money only by "feeling the market" but what I claim is that every one of their methods can be systemized as long as the person who is defining/creating the system knows what the right questions to ask are.
I agree with you. Even the most complex discretionary method, or 'right brain' pattern recognition process, can potentially be broken down into a set of rules. Human thought processes are finite, and there is a rationale behind right-brained, 'strategic' thought. By asking the right questions, every variation and exception could eventually be covered. But the rule set, though finite, would likely be massive, and hence defining and coding it would represent a herculean - and possibly unworkable - task.

A further consideration is the fact that markets might 'evolve', causing the discretionary trader to progressively modify his approach; or as he 'learns' more about how markets operate. Hence the rule set would need to be modified accordingly, a continuous and never-ending process.
  • #567
  • Edited 11:20am Sep 23, 2010 10:37am | Edited 11:20am
Quoting Turveyd
I think the chess was still more down to raw computational power they did use a mainframe which was the quickest ever to be fair and used the chess game to help sell it basically.
No, I think Kasparov defeated it the first go round. The IBM team had to increase the knowledge base of DeepBlue (using a chess GM for input, Joel Benjamin, I think) before they could defeat Kasparov the second time round. DeepBlue was deficient in background knowledge; the PC processing power alone couldn't beat Kasparov.

Goes to Fudomoyo's point about background knowledge, context. Many of the best traders have this skill from screen time, and it's almost impossible to code in an EA. Too complex a weighing of variables for a machine. Think this point is made in Schwager's Wizards book interviews: many of the traders are making decisions that they are not even aware of (and can't describe to a coder) because it's implicit background knowledge/skill of screen time. They will say they are adhering to their system, but are in fact very often applying loads of discretion that make the so-called system out-perform. Same person with the same system would fail, yet the great trader/originator of the sys. will swear it's a wonderful sys. And they're right because they know how/when to use it. Same as driving a car: a ferrari doesn't make the driver. The driver is the person/entity making the profits and who gets from point A to point B--not the car. The context or framework (outside-in) is more important than the system (inside-out).

Programmers or engineers generally don't make good traders.

Per pipmutt, the good traders are always weighing intangibles against each other (qua "trading" them in their mind, BEFORE even using their system). On the impossibility point, yes, it's the same thing as Godel's incompleteness theorem about math.

The algos in stocks are making their money off microsecond gaming of the bid-ask spread, repeated millions of times on thousands of stocks--NOT off trends or any TA at all. No TA being used in high-frequency computerized stock trading, but millisecond gaming of bid-ask spread between computers. Adds up to a whole pile of money off nothing . Same as brokers in FX, but just much, much faster, more systematic. Stock algos play the broker game. Nothing like EA's/TA/trends.
  • #568
  • Sep 23, 2010 10:44am Sep 23, 2010 10:44am
Quoting hanover

coding it would represent a herculean - and possibly unworkable - task.
Herculean, I would say impossible!

I suppose it's a remote possibility for a purely mechanical strategy but taking into account fundamentals, unexpected news and data, geopolitical events, central bank interventions, risk aversion.....and a host of other influencing factors, it's a non-starter as far as I can see. How can the unexpected be programmed in?
  • #569
  • Sep 23, 2010 11:05am Sep 23, 2010 11:05am
Quoting pipmutt
Herculean, I would say impossible!

I suppose it's a remote possibility for a purely mechanical strategy but taking into account fundamentals, unexpected news and data, geopolitical events, central bank interventions, risk aversion.....and a host of other influencing factors, it's a non-starter as far as I can see. How can the unexpected be programmed in?

You could ignore those issues cause if your in 50% of the time those will go in your favour, the system should let them run and kill the losers quickly so over all you should be up.

I just took a trade because I saw a buyer scooping up EJ on ever pull back, it'll be very hard for a EA to see this and despite having direction market rules based on MA's, also very hard to code that into a EA.

Have a look around at the EA section, has anyone got a profitable EA ?? Nope nobody, yes EA's win competitions but the same EA tested after might fail miserably the winner is generally the person that got lucky it seems.

25years ago in my hey days I'm sure I'd of cracked it with my C64
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #570
  • Sep 23, 2010 11:07am Sep 23, 2010 11:07am
the only algo I've ever heard of that worked was not using any form of TA, charts, patterns or math, and furthermore, they advised the sooner you get away from stuff the better off you are.

They were also running on futures and stocks and not FX.

The algo made money but nothing crazy.
  • #571
  • Sep 23, 2010 11:13am Sep 23, 2010 11:13am
Quoting Monroe
the only algo I've ever heard of that worked was not using any form of TA, charts, patterns or math, and furthermore, they advised the sooner you get away from stuff the better off you are.

They were also running on futures and stocks and not FX.

The algo made money but nothing crazy.

Stocks are more suited for longer term trading, they trend better and don't have out news released 3+ times per day which totally changes the trend and makes forex effectively random longer term.

Shame 8years ish back, the SEC stopped 3 round trips ( in and out same day ) in a 5day period or something, you did that with a margin account and account frozen and margin removed ( 2:1 ) and no ability to short!! Unless you've got 25K+. This totally ruined my trading
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #572
  • Edited 11:29am Sep 23, 2010 11:19am | Edited 11:29am
Yes, I think with intraday trading you need to maintain 25K for stocks (doing multiple trades thoughout the day)

There are some very volatile stocks with large volume and plenty are great for trading intraday, plus you have level 2 and time and sales, huge advantage I would assume. Plus costs are much cheaper.

only thing is the tax disadvantage, but i'm not sure how that weighs out...
  • #573
  • Sep 23, 2010 11:25am Sep 23, 2010 11:25am
Quoting Monroe
Yes, I think with intraday trading you need to maintain 25K for stocks (doing multiple trades thoughout the day)

There are some very volatile stocks and plenty are great for trading intraday, plus you have level 2 and time and sales, huge advantage I would assume.

Yep you got 4:1 margin as long as your back below the 2:1 for the close to.

I'm about 24.5K short of this sadly LOL I think I hit 18K.

Never found Level 2 useful at all, I used to multiple day swing, but with 10 stocks in play at a time, you only needed 3 stocks to hit there SL same day ( I used to trade with a tight SL ) in a row and game over

I moved to high risk options to get around this, made stacks when I was with the market then the market turned and wiped me out over night, If I was on the other direction I'd of made 100K inside of the week doh!!

I was supposed to have 1/3rd of my positions, counter trend incase of a direction turn, broke my rules and DIED the death.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #574
  • Sep 23, 2010 11:31am Sep 23, 2010 11:31am
level 2 and time and sales are like the main thing man, you can get a lot from this. Anyone I have ever seen trading stocks intraday and making money, this was what they were looking at majority of the time, and the only thing they were looking at while they were placing orders. Or else just watching top movers and volumes to find the stocks.

I've heard of some crazy options guys out there making cash, but options are way more costly I believe.
  • #575
  • Sep 23, 2010 11:35am Sep 23, 2010 11:35am
Quoting Turveyd
Have a look around at the EA section, has anyone got a profitable EA ?? Nope nobody, yes EA's win competitions but the same EA tested after might fail miserably the winner is generally the person that got lucky it seems.
It's actually quite simple to design and build a profitable automated trading system, but it's not something that's particularly suited to collaborative development, and why would anyone want to share it with people incapable of working things out for themselves ? The whole idea is that those who can get to exploit those that cant.
  • #576
  • Sep 23, 2010 11:36am Sep 23, 2010 11:36am
Quoting Monroe
level 2 and time and sales are like the main thing man, you can get a lot from this.

I've seen some crazy options guys out there, but options are way more costly I believe.

Dunno, just end up watching it mesmerized and not doing anything.

Options are good in a trending market, in a sideways market your losing value cause your losing time, I've took a 100% loss on many but at the same time 1000% profits are quite normal, most of these are $100 all in bets, so lose $100 or make $900.

Definately high risk though!!
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #577
  • Sep 23, 2010 11:40am Sep 23, 2010 11:40am
Quoting Turveyd
Dunno, just end up watching it mesmerized and not doing anything.

Options are good in a trending market, in a sideways market your losing value cause your losing time, I've took a 100% loss on many but at the same time 1000% profits are quite normal, most of these are $100 all in bets, so lose $100 or make $900.

Definately high risk though!!
not too shabby...
  • #578
  • Sep 23, 2010 11:50am Sep 23, 2010 11:50am
Quoting Monroe
not too shabby...

I bought some puts and gained 1600% or something equally as stupid cause some SEC news fear in under 45mins, dropped from $20 to $8 I think it was.


If I hadn't of blown it, I'd be a millionaire already damn market reversing on me
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #579
  • Sep 23, 2010 11:52am Sep 23, 2010 11:52am
Quoting Turveyd

You could ignore those issues cause if your in 50% of the time those will go in your favour, the system should let them run and kill the losers quickly so over all you should be up.
Sounds like a mechanical strategy to me and you're never going to get comparable results, TA indicators are notoriously lagging.

I'm sure there are thousands of simple automated strategies plodding away producing mediocre returns, I guess it depends on what you want out of the market and how much time and effort you're prepared to invest.
  • #580
  • Sep 23, 2010 12:11pm Sep 23, 2010 12:11pm
Quoting Monroe
not too shabby...
hehe, it's not, is it.
Enter Signature
1 trader viewing now