Recommended Books/Websites?

[doctorj]

I want to put in a big plug for Nassim Nicholas Taleb's ‘Fooled by Randomness’. Whilst not strictly a trading book, the topics are very relevant and, as a trader, Nassim has a lot of relevant examples and discussion.

Has anyone else read this or the ‘Black Swan’? I’d be very interested to hear what the derivative guys think about his ideas and the application to pricing risk, particularly in light of sub-prime.

 

[kingie_d]

I recently read Van Tharps "Trading your way to Financial Freedom" 2nd Ed on the advise of someone on this site. I'm still a newbie trader but I've spent lots of time and money reading up on everything I can get my hands on about trading. I have to say that Van Tharps book is one of the best I've read about putting your own system together. It covers pretty much all aspects of trading - not just entries - which I found alot of books do.
Highly recommended to anyone who hasn't read it yet ( although I'm probably the last person here to read it! )

 

[howardbandy]

I want to put in a big plug for Nassim Nicholas Taleb's ‘Fooled by Randomness’. Whilst not strictly a trading book, the topics are very relevant and, as a trader, Nassim has a lot of relevant examples and discussion.

Has anyone else read this or the ‘Black Swan’? I’d be very interested to hear what the derivative guys think about his ideas and the application to pricing risk, particularly in light of sub-prime.

Greetings --

I have read both, and recommend both. Both are highly recommended (by me) for anyone involved in developing trading systems. They both point out that low-probability, high-risk events are the ones that cause big losses in trading accounts. System developers have a hard time preparing for and dealing with low-probability events since they occur so seldom.

I think the world financial markets have a potential risk, perhaps to be triggered by the problems with sub-primes. While many tradable issues appear to be uncorrelated during stable market conditions, everything is correlated during panics. People who need to raise cash sell whatever is liquid.

Thanks,
Howard

 

[howardbandy]

I recently read Van Tharps "Trading your way to Financial Freedom" 2nd Ed on the advise of someone on this site. I'm still a newbie trader but I've spent lots of time and money reading up on everything I can get my hands on about trading. I have to say that Van Tharps book is one of the best I've read about putting your own system together. It covers pretty much all aspects of trading - not just entries - which I found alot of books do.
Highly recommended to anyone who hasn't read it yet ( although I'm probably the last person here to read it! )

Greetings --

I agree that Tharp's book is worth while. The second edition is much better than the first.

However, I have a difference of opinion with him on two major points:

One, the issue of neuro-linguistic programming. Tharp is a psychologist and an expert in this field, and he wants to apply it to trading. He, and other psychologists who work with traders, want to help the trader become comfortable with the system they are trading. I think that is backwards. If the trader begins by defining the features of a trading system that he or she will be comfortable trading, encodes those into an objective function, and designs trading systems that score well using that objective function, then the issue of cognitive dissonance and the need for trading coaches fades.

Two, the issue of the number of trades necessary to have confidence in your system. You can do whatever you want to while developing your system. The data that is used during development is called the in-sample data. After you are satisfied with the system, the tests run over the in-sample data will be good. They are always good -- we do not stop playing with the system until they are good. However, those results from tests of the in-sample data have no value in predicting how the system will perform in the future. No value. None. No matter how many closed trades. Thirty is not enough, 30,000 is not enough. The only way to estimate future performance is to test the system using data that has not been used at all during development of the system. That is called out-of-sample data.

Thanks,
Howard

 

[nizar]

Greetings --

I have read both, and recommend both. Both are highly recommended (by me) for anyone involved in developing trading systems. They both point out that low-probability, high-risk events are the ones that cause big losses in trading accounts. System developers have a hard time preparing for and dealing with low-probability events since they occur so seldom.

I think the world financial markets have a potential risk, perhaps to be triggered by the problems with sub-primes. While many tradable issues appear to be uncorrelated during stable market conditions, everything is correlated during panics. People who need to raise cash sell whatever is liquid.

Thanks,
Howard

Thanks Howard for that overview.
I'll put those books on my list.

 

[julius]

Hi Howard,

What other books do you reccommend, relevent to trading in general ?

With your background, I'm sure you've read just about everything that's out there ...

Cheers, good to know you still read the forums

 

[doctorj]

Greetings --

I have read both, and recommend both. Both are highly recommended (by me) for anyone involved in developing trading systems. They both point out that low-probability, high-risk events are the ones that cause big losses in trading accounts. System developers have a hard time preparing for and dealing with low-probability events since they occur so seldom.

I think the world financial markets have a potential risk, perhaps to be triggered by the problems with sub-primes. While many tradable issues appear to be uncorrelated during stable market conditions, everything is correlated during panics. People who need to raise cash sell whatever is liquid.

Thanks,
Howard

I have to admit, I'm yet to finish them but, so far it hasn't varied too much from interviews I've heard with Taleb (aside from the fact he's much kinder to Journalists in interviews). It's all very interesting, but I wonder at the practical application of it to trading. So far it seems to boil down to two things well worn concepts (hmmm I wonder what he'd say about gross over-simplications), Sharpe Ratio and Risk Management through position sizing and stop losses.

I'm also still formulating my opinion on the matter, but to this point I have three thoughts:
1) How well do the common risk management strategies deal with events that are several (3, 4, 5??) deviations from the mean, particularly as these events can result in highly correlated markets across asset classes? How well can we manage these non-linear risks?

2) If we accept that for some asset classes it isn't a zero sum game, this would result in a positive skew to the distribution of returns probably through both a shift of the distribution to the right and a fatter tail at the right end. This would mean that being on the long side of the market and taking more risk than average would exhibit a more favourable sharpe ratio (on average) than those that manage their risk more aggresively.

3) Given Taleb's beliefs, how the hell does he trade? Does he treat movements as entirely random and just apply prudent risk management?

 

[MS+Tradesim]

Greetings --

I agree that Tharp's book is worth while. The second edition is much better than the first.

However, I have a difference of opinion with him on two major points:

One, the issue of neuro-linguistic programming. Tharp is a psychologist and an expert in this field, and he wants to apply it to trading. He, and other psychologists who work with traders, want to help the trader become comfortable with the system they are trading. I think that is backwards. If the trader begins by defining the features of a trading system that he or she will be comfortable trading, encodes those into an objective function, and designs trading systems that score well using that objective function, then the issue of cognitive dissonance and the need for trading coaches fades.

I can't agree with this. Tharp does not want to make people comfortable with their system prior to development. You are seriously misrepresenting or misunderstanding him here. He emphasises objective system development based on your style and personality. It is when a person has such a system but still has trouble following it that he advocates psychological work to make the person comfortable with their system.

 

[Temjin]

Two, the issue of the number of trades necessary to have confidence in your system. You can do whatever you want to while developing your system. The data that is used during development is called the in-sample data. After you are satisfied with the system, the tests run over the in-sample data will be good. They are always good -- we do not stop playing with the system until they are good. However, those results from tests of the in-sample data have no value in predicting how the system will perform in the future. No value. None. No matter how many closed trades. Thirty is not enough, 30,000 is not enough. The only way to estimate future performance is to test the system using data that has not been used at all during development of the system. That is called out-of-sample data.

Thanks,
Howard

And this one as well. Maybe I missed it in his books, but I personally do not recall him NOT HINTING the importance of out of sample data while testing systems. He did mention at least 30 samples are needed for predictive values, and that is something I do not agree as the more samples (out of sample, clean, untested data), the better it is.

Maybe he really didn't mention out of sample data testing, I could have missed it or mistaken it from other books. However, I was pretty sure I was NOT not taught on the importance on that in his books.

 

[howardbandy]

Greetings all --

First, I'll repeat that I Do Recommend Dr Tharp's book, "Trade Your Way to Financial Freedom," second edition.

Second, I'll stand by my comments about the book's lack of information about selecting the method by which the goodness of any alternative is measured. There is no serious discussion of any metric other than expectancy -- R, in his terms. Expectancy is very important and very poorly understood. And Dr Tharp is performing a great service by introducing it to a wider audience. But expectancy alone makes a very poor metric for choosing among alternative trading systems -- systems chosen for high expectancy often have equally high drawdown. There are Much better metrics by which to evaluate systems and alternative choices in designing them. And there are techniques for creating and using complex objective functions that can include expectancy as one of the components. None of that is discussed. He does suggest choosing an appropriate time frame and picking trend-following or mean-reversion, but those are trading system design choices, not metric choices.

Third, I'll stand by, and emphasize, my comments about the book's lack of any information about the importance of using out-of-sample tests. There is not a single substantive mention of optimization, in-sample, out-of-sample, walk forward, statistical significance or any related terms in the table of contents, the glossary, the index, or on any of the text pages. In fact, he goes out of his way to avoid discussing those topics. By the lack of mention of the importance of basing decisions on out-of-sample tests, his suggestions regarding stop placement, position size, and so forth imply that use of in-sample results is acceptable, is expected, and that is what the text is using.

Out-of-sample testing is probably the Single Most Important concept of trading system development. It is trading's equivalent of the double-blind tests made by pharmaceutical companies before drugs are released.

There is no relation between the performance of a trading over its in-sample data and the likelihood of its performance when trading with real money tomorrow. The in-sample results are Always good. We do not stop fiddling with the system until they are good. The better the in-sample results are, the worse the out-of-sample results are likely to be. There is no way we can tell by examining the in-sample results whether those results were achieved because the algorithm is accurately measuring some profitable feature of the market or is just curve-fit to the data. No way. None!

For example, I have documented trading systems, plural, that each have over one million closed trades in the in-sample period, yet are not profitable out-of-sample.

Making trading decisions based on nothing but in-sample results is extremely hazardous to account balances.

Yes, read Van Tharp. What he does say is worth while. But do not stop with his book. Include some others that describe the processes of objective function design, in-sample / out-of-sample analysis, and walk forward testing.

Thanks for listening,
Howard

 

[nizar]

I have to admit, I'm yet to finish them but, so far it hasn't varied too much from interviews I've heard with Taleb (aside from the fact he's much kinder to Journalists in interviews). It's all very interesting, but I wonder at the practical application of it to trading. So far it seems to boil down to two things well worn concepts (hmmm I wonder what he'd say about gross over-simplications), Sharpe Ratio and Risk Management through position sizing and stop losses.

I'm also still formulating my opinion on the matter, but to this point I have three thoughts:
1) How well do the common risk management strategies deal with events that are several (3, 4, 5??) deviations from the mean, particularly as these events can result in highly correlated markets across asset classes? How well can we manage these non-linear risks?

2) If we accept that for some asset classes it isn't a zero sum game, this would result in a positive skew to the distribution of returns probably through both a shift of the distribution to the right and a fatter tail at the right end. This would mean that being on the long side of the market and taking more risk than average would exhibit a more favourable sharpe ratio (on average) than those that manage their risk more aggresively.

3) Given Taleb's beliefs, how the hell does he trade? Does he treat movements as entirely random and just apply prudent risk management?

Some great questions there Doc, im looking forward to the responses.

 

[rub92me]

I enjoyed reading 'Fooled by Randomness'. I saw Nicholas Taleb has a new book out titled 'Black Swans'. Anyone read this yet and if so, is it worth reading or is it just a re-hash of the same ideas?

 

[It's Snake Pliskin]

I enjoyed reading 'Fooled by Randomness'. I saw Nicholas Taleb has a new book out titled 'Black Swans'. Anyone read this yet and if so, is it worth reading or is it just a re-hash of the same ideas?

It is a whole lot more in depth and has confirmed some of my thoughts.
It isn't an easy read purely for the concepts and terminology etc that he uses. If you are looking for trading answers you may not find them. But I liked the part on prediction.

 

[Wysiwyg]

The list under Pages has sound knowledge and fascination for traders.

https://jesse-livermore.com/

 

[ASFscalp001]

great thread! I am sure I’ll find something interesting and knowledgeable to read from these book recommendations.