The Birth of a Quant Trader | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Quant basket trader? I need your help! 92 replies

Quant trading & QuantLib 6 replies

SAS for quant analysis 0 replies

Quant Corner 12 replies

questions from a future quant 39 replies

The Birth of a Quant Trader

  • #21
  • Mar 21, 2011 3:42pm Mar 21, 2011 3:42pm
What you are saying is that the key here isnt so much the analysis itself (be it TA, FA, watever..) but the actual trade management.

Ok I agree with you with regards to 'consistency is in how we trade'.
But that is one of my points as well.

If consistency comes by having a disciplined approach, then this in itself
is comprised of 2 elements in my newish experience here:

1) A clear trade management plan.

2) Consistency and discipline in applying the trade management plan.

Discipline = following a plan = consistent results

Also the trade plan itself should be suited to the conditions of the market and also the time exposure in the market.

i.e. daily chart is obviously better for mid to longer term trade, meanwhile 1m/5m etc is more suited to shorted 'in and out' trades.

With regards to 'automated systems built with wrong data' what im trying to say is that I believe many auto systems are built by using random/wrong indicators that are not fully understood by the end user.

In fact, you can see it everywhere on this forum (not wanting to offend anyone here):

The typical I will use RSI, CCI, MACD and a MA if 123 happens I buy if 456 happens I sell. Just pure random crap IMHO.

Isnt this in itself, an automated process? A mechanical one? It just hasnt 'yet' been automated...but no matter how you look at it, it IS a auto system.

As for 'most being built on conventional tools' I DONT see anything wrong with that, IF and only IF the end user knows exactly WHY such tools are being used for the system itself (like I wrote just before this...).

Many conventional tools can pretty much show similar data, all of them derived from price itself to quantify certain aspects like you say.

I also agree many are misused by default.

And that is exactly where im going with all this!

If indies are used to 'quantify' certain aspects of price, then why do I need a 'phd' in order to develop my OWN indies that quantify specific data that I want?

I think indies are a blessing... IF used correctly.

Just like a speedometer in a car. you dont need it, but it does avoid speeding tickets...

Thanks for your pdf i'll read into it.

And thanks for your interest here!
  • #22
  • Mar 21, 2011 3:55pm Mar 21, 2011 3:55pm
I also agree that once you have developed a "winning trading system", trade management is the last and most difficult hurdle to get past. You can be a Quant Trader, Swing Trader, Day Trader, Astrology Trader or whatever. The art of minimizing losses and maximizing wins is what separates the pros from the amateurs.
  • #23
  • Mar 21, 2011 4:00pm Mar 21, 2011 4:00pm
Quoting Pipalicious
I also wonder,

If we take psychology, mindset, lack of confidence, greed/fear out of the equation...

And only have conventional systems that people trade now publicly such as this forum.

Why do most people continue to struggle?

Why do most people keep failing no matter what?

Is it because the systems themselves are floored? Over exploited?

Could it be with the subjectivity involved in TA?
People continue to struggle not because of lack of "objective" systems, but because of lack of self-awareness. They keep on failing because they are not aware of the cognitive biases that keep them from seeing and thinking clearly. In the area of speculation this has been thoroughly explored by Nobel laureates Kahneman and Tversky.

Having struggled for a while you find it hard to believe that trading can and should be simple, and you may come to believe that what would be required is a more objective, quantitative approach. But MAs, RSI, Stochastic, etc, are perfectly objective. The problem is psychological.

Quote
What do you suggest whole heartedly? What do you recomend I do?

What would you do if (knowing everything you know now) you could start...

Get to know yourself really well, in particular when it comes to the cognitive biases that keep you struggling and prevent you from experiencing trading as simple.

Get to know the market really well, and then ask yourself whether more quantitative modelling is what you need. If you were admitted to the Royal Ballet School and struggled in your first year, would you think that you needed to develop a quantitative model that describes ballet in order to make it as a dancer?
  • #24
  • Mar 21, 2011 4:20pm Mar 21, 2011 4:20pm
Pipalicious,

I wish you the very best in your journey down the quant road, but do indept research before jumping in. Take a look at a book called "The Quants" & you will find there have been some crash & burns along the way. Not trying to null your spirits, but I do try to live in realville.

Do keep everyone informed on your progress.

____________________________
  • #25
  • Edited 5:15pm Mar 21, 2011 5:05pm | Edited 5:15pm
Quoting Pipalicious
If indies are used to 'quantify' certain aspects of price, then why do I need a 'phd' in order to develop my OWN indies that quantify specific data that I want?

I think indies are a blessing... IF used correctly.
Quants don't typically use "retail indicators".

Here's how a "quant strength" indicator looks like. They take a standard EMA and convert it to something which they say works much better:

http://dl.dropbox.com/u/190212/ff/quant_ema_1.png

http://dl.dropbox.com/u/190212/ff/quant_ema_2.png

http://dl.dropbox.com/u/190212/ff/quant_ema_3.png

http://dl.dropbox.com/u/190212/ff/quant_ema_4.png

http://dl.dropbox.com/u/190212/ff/quant_ema_5.png

http://dl.dropbox.com/u/190212/ff/quant_ema_6.png

To quote the gist of it: The property of this kernel is a plateau in the function before it asymptotically declines to zero and it is robust against extreme variations leaving the average by assigning exponentially decaying weights. Contrary to simple exponential average, which is very sensitive to the most recent history, it assigns relatively uniform weights to such new information. Therefore, a robust kernel preserves only the desirable robustness properties of the simple average and exponential average kernels but ignores their highly noisy unrobust properties.

Also, quants don't use candles. Really. I have over 200 quant papers and there is not a single candlestick chart in all of them. They mostly use "line charts", but a denser version (not the simplified one found in MT4).

Typical quant chart (of volatility in this case): http://dl.dropbox.com/u/190212/ff/vol_1.png
  • #26
  • Mar 21, 2011 5:13pm Mar 21, 2011 5:13pm
Quoting Adal
Quants don't typically use "retail indicators".

Here's how a "quant strength" indicator looks like. They take a standard EMA and convert it to something which they say works much better:



Also,...
...first time ever that I use this emoticon...
  • #27
  • Mar 21, 2011 5:19pm Mar 21, 2011 5:19pm
what a load of bolloggs
rat
  • #28
  • Mar 21, 2011 5:20pm Mar 21, 2011 5:20pm
Quoting Adal
To quote the gist of it: [i]The property of this kernel is a plateau in the function before it asymptotically declines to zero and it is robust against extreme variations leaving the average by assigning exponentially decaying weights. Contrary to simple exponential average, which is very sensitive to the most recent history, it assigns relatively uniform weights to such new information. Therefore, [color=Red][b]a robust kernel preserves only the desirable robustness properties of the simple average and exponential average...
This is getting funnier by the minute...
  • #29
  • Mar 21, 2011 5:28pm Mar 21, 2011 5:28pm
Quoting medici
This is getting funnier by the minute...
Of course you are laughing at stuff you don't understand.

Do you know who is one of the authors of this paper? Richard Olsen, creator of OANDA, one of the biggest automated market makers in the world.

Do you know that he, and three others co-authors of the paper I quoted from wrote one of the first book about High Frequency Trading? 10 years ago, when you probably didn't even know what MetaTrader is.

Book: http://www.amazon.com/Introduction-H.../dp/0122796713

Paper I quoted from: http://citeseerx.ist.psu.edu/viewdoc...=rep1&type=pdf

Keep laughing.... you just show how ignorant you are
  • #30
  • Mar 21, 2011 5:36pm Mar 21, 2011 5:36pm
Quoting Adal
Of course you are laughing at stuff you don't understand.

Do you know who is one of the authors of this paper? Richard Olsen, creator of OANDA, one of the biggest automated market makers in the world.

Do you know that he, and three others co-authors of the paper I quoted from wrote one of the first book about High Frequency Trading? 10 years ago, when you probably didn't even know what MetaTrader is.

Book: http://www.amazon.com/Introduction-H.../dp/0122796713

Paper I quoted from: http://citeseerx.ist.psu.edu/viewdoc...=rep1&type=pdf

Keep...
I know Richard and I know the book, and it's a piece of marketing. But you are ignorant enough to take it seriously 10 years later

Also, I have 20+ years in this business, and the piece that you quoted like a priest in church anybody with proper quant skills would skim in three seconds and wonder who's being duped this time
  • #31
  • Mar 21, 2011 5:42pm Mar 21, 2011 5:42pm
Quoting medici
I know Richard and I know the book, and it's a piece of marketing. But you are ignorant enough to take it seriously 10 years later

Also, I have 20+ years in this business, and the piece that you quoted like a priest in church anybody with proper quant skills would skim in three seconds and wonder who's being duped this time
Please, don't let us be duped. After 20+ years in this business you obviously are not here on this forum wasting your time trying to learn something. You are here to educate others. So let us know the dangers in that paper oh master of knowledge.

Waiting for a new spin now from you with some technobabble... A new reframing could also do, for example you could say that your are not here to learn or to teach, but to observe the sheeple and plot against it. I'm sure you can find another way to divert the discussion, you are a master at that.

Oh, and by the way, why not show yourself and prove your 20+ years track record. You know, Richard is not afraid of putting his name on this paper. Be a man, and don't attack him from behind.

Oh, wait again, you are not attacking him, it's just me the stupid one. Richard knows what his doing.
  • #32
  • Mar 21, 2011 5:52pm Mar 21, 2011 5:52pm
Quoting Adal
Please, don't let us be duped. After 20+ years in this business you obviously are not here on this forum wasting your time trying to learn something. You are here to educate others. So let us know the dangers in that paper oh master of knowledge.

Waiting for a new spin now from you...
Thank you for providing continued entertainment

Contrary to what you may think, I am here to learn, but for some reason I can't find much...But I have met the occasional seasoned trader who have enriched my everyday life.

Also, out of gratitude to those who once taught me, I feel an urge to "give back", and some seem to appreciate that...here's the most recent quote from Thursday: " It's certainly helpful for people like me, who have never had an actual mentor and are trying to learn this business. It should be almost impossible to do, but thanks to the occasional helpful person such as yourself with years of experience and a willingness to share I've been making it work as a profession so far..."

Until you have a long file of testimonials to your ability to help make the step from amateur to pro, I would suggest you continue to perform in the entertainment segment.
  • #33
  • Mar 21, 2011 5:58pm Mar 21, 2011 5:58pm
Quoting medici
Thank you for providing continued entertainment

Contrary to what you may think, I am here to learn, but for some reason I can't find much...But I have met the occasional seasoned trader who have enriched my everyday life.

Also, out of gratitude to those who once taught me, I feel an urge to "give back", and some seem to appreciate that...here's the most recent quote from Thursday: [i]" It's certainly helpful for people like me, who have never had an actual mentor and are trying to learn this business....
Just as expected, you re-framed again. You are the nimble humble person still learning, me the evil attacker not giving back.

Of course you will not stand by your previous affirmation that the paper is laughing material, now that you learned who wrote it, one of the most successful quants in the world.

Yes, it's year 2000 material, and yes, it doesn't contain any secret sauce, but the EMA described there is light-years ahead of the typical indicators found in MetaTrader. 10 years later, the world still has not caught on. Now imagine what kind of stuff Richard created in the last 10 years and which he is not ready to publish yet.
  • #34
  • Mar 21, 2011 6:02pm Mar 21, 2011 6:02pm
Quoting Adal

Of course you will not stand by your previous affirmation that the paper is laughing material, now that you learned who wrote it, one of the most successful quants in the world.
Didn't say that the paper was laughing material, only that it was as basic as 2+2=4...which you quoted as a sermon And having read the book, I knew the source, which made it even more laughable
  • #35
  • Mar 21, 2011 6:08pm Mar 21, 2011 6:08pm
Quoting Adal
Yes, it's year 2000 material, and yes, it doesn't contain any secret sauce, but the EMA described there is light-years ahead of the typical indicators found in MetaTrader. 10 years later, the world still has not caught on. Now imagine what kind of stuff Richard created in the last 10 years and which he is not ready to publish yet.
Didn't read this initally, but oh, what entertainment

You don't seem to know the least about these things, and it may not be found in MetaTrader, but if you send your wishes to Santa Claus, Rovaniemi, Finland, who knows, maybe Santa will be good to you?
  • #36
  • Edited 6:22pm Mar 21, 2011 6:10pm | Edited 6:22pm
Quoting medici
Didn't say that the paper was laughing material, only that it was as basic as 2+2=4...which you quoted as a sermon And having read the book, I knew the source, which made it even more laughable
More spinning. I will let our watchers judge what your initial opinion about the paper was.

Quote
You don't seem to know the least about these things, and it may not be found in MetaTrader, but if you send your wishes to Santa Claus, Rovaniemi, Finland, who knows, maybe Santa will be good to you?
I can program my own stuff, thank you:

http://dl.dropbox.com/u/190212/ff/veloce_charting.png

http://dl.dropbox.com/u/190212/ff/veloce_r_session.png

Don't forget to notice that I used candles here, after I said quants don't use them.

http://dl.dropbox.com/u/190212/ff/tick.png

http://dl.dropbox.com/u/190212/public/oanda_box.png


I have watched your other threads, and I know there is no contradicting you, because you have more free time to engage in these discussions than pretty much anyone else.

So here, you win this one from me.

Congratulations. You won the Internets.
  • #37
  • Mar 21, 2011 6:30pm Mar 21, 2011 6:30pm
I think you don't need to have Ph.D. in math or statistics to build a successful trading system. Just read about some simple tools from artificial intelligence that designed to work with discrete datastructures. I would suggest to read about

1. Naive Bayesian estimators - very simple tool that uses conditional probabilities to estimate probability of event.

2. Decision Trees - tool that allow semi-automatically build complex strategies like "If A observed to that, if B observed do this". Also very intiutive and reasonably simple tool.

3. Linear regression - tool to find linear dependency between events and outcome.

If you don't know anything about math, then it will take about 1 month to get pretty good working understanding of Naive Bayesian estimator and Decision trees (1 month for each), and probably couple of weeks for linear regression.

Stay away from complex fancy tools like Neural Networks and Kernel Methods. First, they are designed to work with continuous data and second they require significantly higher level of understanding of math and Artificial Intelligence. Remember in order to achieve success you don't necessary need fancy tools you need good preprocessing of raw data (meaningful feature extraction). Artificial Intelligence methods is not a magic bullet if you misuse them you will get nonsense results. As they say, junk in - junk out.

Next, you will need to learn a programming language. I would recommend Java. It is free, there are a lot of free tutorials and sample code online, and this is mainstream language that might be useful for you.

Regarding suggested reading from the quant library. I would recommend NOT to read anything at all. All that you can find online for free is utter junk created by clowns. You might skip through few papers just to make an impression what they are doing; mostly it is a good example how not to think. Screenshots of paper posted above is a good example of junky woodoo math, read enough of such papers, start to think this way and you will lose years of your time sorting out nonsense. Remember when quant get accepted in a firm, he signs non-disclosure agreement, it is simply impossible to find online example of real reasearch that is done in a quant shop. You will not find anything even of smallest value in the papers you can find online. Instead, get yourself good introductory book on Artificial Intelligence or Machine Learning, it will teach you basic skills necessary to build a successful application.

Last but now the least, you will need good historical data to train your system. 1-minute bars won't make it, you will need tick-by-tick data. Oanda gives 7 years of tick-by-tick data on 5 currencies if you have balance more than 1K in account with them.
  • #38
  • Mar 21, 2011 6:38pm Mar 21, 2011 6:38pm
Quoting triger88990
it was worth staying awake this night

just don't jump at my head, will you (to bad I can't use more emoticons)
Glad you liked it I had so much fun I don't pay for cable TV, and why would I with free entertainment like this And apologies to Adal if anything was taken personally. Wasn't meant like that. I hope Pipalicious can find some meaning in it...
  • #39
  • Mar 21, 2011 6:44pm Mar 21, 2011 6:44pm
Quoting medici
Glad you liked it I had so much fun I don't pay for cable TV, and why would I with free entertainment like this And apologies to Adal if anything was taken personally. Wasn't meant like that. I hope Pipalicious can find some meaning in it...
No need to apologize, we were just having a friendly spar, right

I remembered some posts with some reference books for algorithmic trading for Pipalicious:

http://quantivity.wordpress.com/2010...thmic-trading/
http://quantivity.wordpress.com/2010...rading-part-2/
http://quantivity.wordpress.com/2010...rading-part-3/

You can also look at his blog-roll, maybe you find something that suits you there.

Cheers.
  • #40
  • Mar 21, 2011 7:34pm Mar 21, 2011 7:34pm
Quoting cash_machine
you will need good historical data to train your system. 1-minute bars won't make it, you will need tick-by-tick data. Oanda gives 7 years of tick-by-tick data on 5 currencies if you have balance more than 1K in account with them.
Good suggestions overall, cash_machine, thanks.
Personally, I don't think someone just starting should get anywhere close to tick data. If you don't do ultra-high frequency trading, M1 is the shortest TF you'd use - and usually, you'd use higher TFs. Most retail traders don't have the required latency (and spreads) to succeed in HFT.
1 trader viewing now