Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypotheses

[sinner]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Hot off the press

Authors: Fan, Opsal, Yu
Title: Equity Anomalies and Idiosyncratic Risk Around the World
Link: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2611047

In this study, we examine how idiosyncratic risk is correlated with a wide array of anomalies, including asset growth, book-to-market, investment-to-assets, momentum, net stock issues, size, and total accruals, in international equity markets. We use zero-cost trading strategy and multifactor models to show that these anomalies produce significant abnormal returns. The abnormal returns vary dramatically among different countries and between developed and emerging countries. We provide strong evidence to support the limits of arbitrage theory across countries by documenting a positive correlation between idiosyncratic risk and abnormal return. It suggests that the existence of these well-known anomalies is due to idiosyncratic risk. In addition, we find that idiosyncratic risk has less impact on abnormal return in developed countries than emerging countries. Our results support the mispricing explanation of the existence of various anomalies across global markets.

 

[tech/a]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

I think craft is right, you just dont read what other people write, then you reflect back something entirely different, and argue against that. Its very odd!

I am going to follow his example in the other thread and leave you to it, good luck with what ever it is you do!

I'm sure Sinners happy with that

On topic, I think evidence based paper that show methods dont work are more common than the reverse, so might be slim pickings. Nearly everything I have seen is not of rigorous academic standard, its just dressed up to look like it!

Not looking for right OR wrong.

Some great stuff so far
I'll have a chance to digest over the weekend.
Appreciate any and all replies.

 

[luutzu]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

I think craft is right, you just dont read what other people write, then you reflect back something entirely different, and argue against that. Its very odd!

I am going to follow his example in the other thread and leave you to it, good luck with what ever it is you do!

On topic, I think evidence based paper that show methods dont work are more common than the reverse, so might be slim pickings. Nearly everything I have seen is not of rigorous academic standard, its just dressed up to look like it!

Maybe I read into what people are saying. That I know what they're implying.

Anyway, by the sound of it you guys think that DCF is simply about plugging in some rates and variables that's been forecasted and neatly worked out by some genius somewhere. All it need is just to add the earning reported and a couple scenario.

And man... all I have done is simply repeat and rephrase what Graham wrote and presumably taught and use himself. But ey, 30 years later he changed his mind... and the 30 years before that all his students were just lucky.

And no, I do not imply that I am his student or a genius or that I will be anywhere as successful as some of them have been. There is such as thing in investing as opportunity.

 

[howardbandy]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Greetings --

Regarding:
On topic, I think evidence based paper that show methods dont work are more common than the reverse, so might be slim pickings. Nearly everything I have seen is not of rigorous academic standard, its just dressed up to look like it!

----------------

If rigorous analysis of a technique shows something does not work, you are comfortable with the assumptions of the study, you can and do replicate the analysis the paper reports, then it probably does not work. Changing the analysis to less rigorous will not make it more likely that the technique works.

Slim pickings is accurate. This is a difficult problem. As Daniel Kahneman writes so well in "Thinking, Fast and Slow," we are all experts at fooling ourselves. We see what we hope to see, ignoring facts and rational conclusions as necessary.

Best regards,
Howard

 

[tech/a]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

We are all experts at fooling ourselves

How true that is in many areas.

 

[Value Collector]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth


I think you would get alot out of this book Tech, as I said in the other thread, F/A is not just about formulas and ratios, you really need to have a good head for business, not just a proper understanding of the individual businesses you are looking at, but you need to understand general business and investment fundamentals.

This book is a compilation of Buffetts letters and Essays from over the years, all broken down and organised into cateregories, I think if you have a read of this you will get an idea of the way his brain works, and you will see his success is not down to a valuation formula or ratios, but rather a deep passion for business, a deep understanding of accounting, and a solid foundation of security analysis.

Anyone that tries to teach you the "Buffett formula" is leading you up the garden path.

The book is called.

The essays of Warren Buffett, lessons for investors and managers.

 

[tech/a]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Thanks for the recommendation
I like the format.
Will grab it.

 

[craft]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

I’m not sure how Buffett or most of F/A in general fits into an evidence based thread. Assuming we are looking for evidence to a scientific standard.

Certainly Buffett & Co and there are long term successful T/A exponents as well provide enough evidence to nullify the efficient market hypothesis. But there is no evidence I have ever seen of what does work every time.

Most of the papers approach things from the perspective of showing the lowest deciles beat the highest deciles or such & such approach beats buy& hold – none of it is really aimed at testing hypotheses by looking for nullifying evidence. If you approached it this way every concept I know of would be nullified because none of it is underpinned by an ironclad law.

What the academic literature shows is “generalities” of things that tend to work on a probability basis. Some papers go to the level of confidence levels to try and ensure the observed generality is not a random occurrence but even a 95% confidence that an observation is not random is far different from collecting evidence to nullify a hypothesis.

The better generalities have appeared again & again over many papers, many markets and many timeframes, but how stable they will be into the future as people try to harvest them will be the big question. My suspicion is only the hardest (least cost effective or psychologically difficult) to harvest will survive.

Deciphering quantitative evidence in F/A is harder than T/A because you have the language of accounting defining a lot of the terms whereas price is pretty straight forward. So even if you get a confidence level that a F/A generality exists you have to be mindful of the underlying definition of the variable.

Here is a good summary of some generalities that have been persistent. (But none of it is sufficiently evidence based in my opinion – so not sure it fits this thread).

http://www.tweedy.com/resources/library_docs/papers/WhatHasWorkedFundOct14Web.pdf

My main interest around the concept of evidence based is how people cope once they see holes in the beliefs that used to fortify their involvement in the market. Do they look for new beliefs or can they still embrace the market in the face of their uncertainty. Not a lot seem to make the step into coping with true uncertainty, they keep looking for another belief to trust in or refortify their existing beliefs in the face of nullifying evidence.

 

[tech/a]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

I’m not sure how Buffett or most of F/A in general fits into an evidence based thread. Assuming we are looking for evidence to a scientific standard.

Thought it would be a good read--didn't and don't expect it to have any relevance to the thread.

Certainly Buffett & Co and there are long term successful T/A exponents as well provide enough evidence to nullify the efficient market hypothesis. But there is no evidence I have ever seen of what does work every time

.

I've not seen evidence myself in a great deal of both methods even that which works more than 50% of the time---lots of belief!

According to the academics I know --- all is not necessarily lost.
Personally I feel to tools necessary to answer even some of the simplest generalities aren't readily available or available to the masses like myself who isn't a Howard Bandy!---Howards made a great start.

I don't think like Howard thinks.
I don't know what to group together to actually have a half a chance of defining an answer to many of my trading ideas/questions---all I have are suspicions---from experience in many cases.

I don't have the knowledge or the tools at my disposal to come close to a determination.
I don't know what those tools are and even if I did it would be like hopping in a Jumbo Jet and pointing me to the run way.----want to fly with me??

Talking to watching and discussing things with my in house quant just shows me how cavernous that divide is ----- to know what these guys know would be awesome. They actually-------"Know what you/they need to know!!!"


I think academia can show us ---DIRECTION.
I really don't think its up to academia to provide results.
One day the likes of Howard will bring academia to the likes of myself so I CAN answer my questions and prove or dispel my theories, and guide me toward that which I truly need to know---EVALUATION

Yet over 20 yrs trading I'm net Profitable.

A great chunk came by remarkable luck catching a 7 yrs bull market

But I've come to the conclusion I'm trend biased---if I get it (trend) Ill profit.
luck will truly be with me.

It must trend often with good range.

I need something that once it trends it trends strongly and is not likely to look back--Long and Short.
I want one instrument----Personally I find it more than difficult to replicate 2/3/4 in the same way.
I need a high win rate and a strong Reward to Risk.

This in trading I've been able to "Know"---evidenced from experience and result.

The DAX---for me-- ticks the boxes.

 

[DeepState]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

My main interest around the concept of evidence based is how people cope once they see holes in the beliefs that used to fortify their involvement in the market. Do they look for new beliefs or can they still embrace the market in the face of their uncertainty. Not a lot seem to make the step into coping with true uncertainty, they keep looking for another belief to trust in or refortify their existing beliefs in the face of nullifying evidence.

What do you do with your own investment research? Aren't your investment conclusions based on a set of beliefs, supported by data and a framework of thought that combines them? If these are never fully certain and subject to ongoing development, what do you do when the facts as you knew them change with the arrival of unexpected information or your understanding of how you combine them changes? Generally speaking, the appropriate step to take is to act in accordance with what you believe to be true now, which may include leaving a wide(r) margin for error. You'd be familiar with what Keynes had to say about what to do when 'facts' change (in which case they were never really facts, but beliefs).

This is the same deal with the investment models which are created, tested against data and modified in light of evidence. When data contradicts a belief entirely, the belief must be discarded. If it is somewhat inconsistent, then you apply judgment, or reach for things like Maximum Likelihood estimation, Bayesian Adjustments, thresholds for rejection vs another hypothesis, or simply guess about what the right thing to do with your understanding of the world should now be.

As to coping with true, pervasive, uncertainty from a research perspective, you really can't. No margin for error is large enough to allow for it unless you know the bounds within which it is uncertain. Nothing can be proved conclusively or not if the idea under investigation is dominated by true uncertainty. Sure, you can get observational outcomes and some of these may produce stuff which meets some stats tests and have '*** (Significant at P = 0.01)' next to them. However, in a fully uncertain situation, you can't put too much weight on these outcomes. The more correct thing to say is "don't really know what this observation means" to any observation drawn from a fully uncertain situation. That would make for seriously boring copy...or otherwise be an excellent argument for indexing, in some ways.

As to confirmation bias, or otherwise making your career on an 'anomaly' (Shiller) or explaining it away via a risk exposure (FAMA), why should that behaviour be so surprising? You get Nobel Prizes and tenure for doing that. It also fills about 80% of the thread volume here. Where would ASF be without it?

 

[craft]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

What do you do with your own investment research? Aren't your investment conclusions based on a set of beliefs, supported by data and a framework of thought that combines them? If these are never fully certain and subject to ongoing development, what do you do when the facts as you knew them change with the arrival of unexpected information or your understanding of how you combine them changes? Generally speaking, the appropriate step to take is to act in accordance with what you believe to be true now, which may include leaving a wide(r) margin for error. You'd be familiar with what Keynes had to say about what to do when 'facts' change (in which case they were never really facts, but beliefs).

This is the same deal with the investment models which are created, tested against data and modified in light of evidence. When data contradicts a belief entirely, the belief must be discarded. If it is somewhat inconsistent, then you apply judgment, or reach for things like Maximum Likelihood estimation, Bayesian Adjustments, thresholds for rejection vs another hypothesis, or simply guess about what the right thing to do with your understanding of the world should now be.

As to coping with true, pervasive, uncertainty from a research perspective, you really can't. No margin for error is large enough to allow for it unless you know the bounds within which it is uncertain. Nothing can be proved conclusively or not if the idea under investigation is dominated by true uncertainty. Sure, you can get observational outcomes and some of these may produce stuff which meets some stats tests and have '*** (Significant at P = 0.01)' next to them. However, in a fully uncertain situation, you can't put too much weight on these outcomes. The more correct thing to say is "don't really know what this observation means" to any observation drawn from a fully uncertain situation. That would make for seriously boring copy...or otherwise be an excellent argument for indexing, in some ways.

As to confirmation bias, or otherwise making your career on an 'anomaly' (Shiller) or explaining it away via a risk exposure (FAMA), why should that behaviour be so surprising? You get Nobel Prizes and tenure for doing that. It also fills about 80% of the thread volume here. Where would ASF be without it?

I’m not sure if you want me to answer those first two questions because the rest of the post basically answers it.

Embracing uncertainty gives you the mindset and flexibility to work on improving your guesses and fixing your bad ones. Charlie Munger allegedly responded about Berkshire success. Because our guesses are better than yours. Says it all about the right mindset IMO. Guesses can be educated or experience backed but guesses they are.

 

[luutzu]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

I’m not sure how Buffett or most of F/A in general fits into an evidence based thread. Assuming we are looking for evidence to a scientific standard.

Let me get this straight...

The world's most successful investor - Buffett - just told you that 8 investors he personally know, many of whom he had worked with, all of whom he know for a fact learn from the same two teachers and apply, with very little modification, the same set of principles... and all of these nine all run different fund, all pick different stocks (with very few overlap, and no coordination among them)... and all of them managed to produce annualised, compounded returns of 20% or over for at least 20 years up to when the lecture was given...

And that is not "scientific" proof enough? Those successes are not evidenced enough to meet your strict criteria of scientific inquiry?

Seriously?

OK... how about the simple idea of looking for a $1 value and only ever pay 50 cents for it?

If you set out to find a buck but will only buy it for half that, there need to be convincing evidence whether or not that will make you profit?

Whatever happen to just accepting the word of the guy with $70 billion or something?

 

[wayneL]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Let me get this straight...

The world's most successful investor - Buffett - just told you that 8 investors he personally know, many of whom he had worked with, all of whom he know for a fact learn from the same two teachers and apply, with very little modification, the same set of principles... and all of these nine all run different fund, all pick different stocks (with very few overlap, and no coordination among them)... and all of them managed to produce annualised, compounded returns of 20% or over for at least 20 years up to when the lecture was given...

And that is not "scientific" proof enough? Those successes are not evidenced enough to meet your strict criteria of scientific inquiry?

Seriously?

OK... how about the simple idea of looking for a $1 value and only ever pay 50 cents for it?

If you set out to find a buck but will only buy it for half that, there need to be convincing evidence whether or not that will make you profit?

Whatever happen to just accepting the word of the guy with $70 billion or something?

Notwithstanding the sense of "value" investing, how Buffet and his ilk does things bear no relation to an individual private investor.

Suggested reading.... Freakonomics and Fooled by Randomness.

 

[luutzu]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Maybe it's true... investing is very very, very very hard.

Here's a word from our sponsor:


Hello.
At Master of the Universe Mutual we know how hard it is for you to invest money. That's why we only select the best and the brightest - It is who we reject that makes us the best.

We have Subject-Matter Experts from all fields - Nobel Laureates in economics, in finance, in history, in political science. We have MBAs, Master of this and all that. We got graduates from Oxford and Harvard and Yale... and those "MBA" from UNSW... they can bring us tea and coffee and dust our Bloomberg and FactSet before we even look at them.

Look around you - here are our latest Bloomy - all 12 screens of them per terminal - we've personally added a few extra carbon footprint to Mother Nature with those. Here's our datacentre with state of the art fibre connected to London, Paris, New York and directly injected into our analysts brainstem - why? Because information is power, and the more data you got, the more information you have - obviously!

I know what you're saying... It's all very impressive, but what about the results?

Well, here they are. Take a look.

Yes we know, it suck egg. Underperforming a simple do-nothing Index pretty much every year; Losing hundreds of billions of people's life savings into one toxic asset and financial scam and corporate bankruptcy.. one after another. It suck so bad even a monkey throwing darts have done better. But are you a monkey? Don't think so.

So ask yourself... if we're so smart, so rich, and have all these resources... and we still can't do our job properly. What on Earth make you think you could? Are you smarter than us? If you're so smart why aren't you rich ey? ey!

So give us your savings, let us play with it. You know you want to. It's one way to not blame yourself when you lose it all on trying to be a Warren Buffett or a business person studying annual reports like it contain anything useful we don't already know.

---

So, who do I give my money to and how will they take my two percent?

 

[History Repeats]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Expected Returns by Antti Ilmanen. And hundreds of reference paper included inside. The author one of the best in its field and a practitioner not academic. It will be a long read, require understanding of asset pricing theory but no maths involved in the book .

 

[luutzu]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Notwithstanding the sense of "value" investing, how Buffet and his ilk does things bear no relation to an individual private investor.

Suggested reading.... Freakonomics and Fooled by Randomness.

No relation to individual investor?

Why is that? Because they are Gods and we are mere mortals?

Don't take what I'm saying to mean that anyone can be Buffett; or anyone who read Graham will be Buffett or Munger or his pals.

A bit of intellect and common sense aside, there's the question of hard work, interests and luck involve in successful investing.

I'm not fooling myself or anyone and say that investing is easy. It takes a lot of work to understand a business, takes a lot of effort and motivation when you have little capital and no time, but it's absurd to think that another human being could do something and the rest could not because we're not made for it.

 

[wayneL]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

......fooled by randomness.

 

[luutzu]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

......fooled by randomness.

Pick a company with a business you could understand.

Read its reports and account.

Read into its industry and competitors.

See if the company is going to go broke. See how it has been doing. See what are its assets and liabilities. See how its returns has been.

Estimate its position in relation to the industry. Have an idea of how its future would work out - will it still survive, will it go broke...

Then estimate its approximate value.

Buy below those estimates.

---


Not easy. Not effortless. Not always possible to find business you know and like and sell at a good price.

But not exactly Rocket Science either.

 

[craft]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Pick a company with a business you could understand.

Read its reports and account.

Read into its industry and competitors.

See if the company is going to go broke. See how it has been doing. See what are its assets and liabilities. See how its returns has been.

Estimate its position in relation to the industry. Have an idea of how its future would work out - will it still survive, will it go broke...

Then estimate its approximate value.

Buy below those estimates.

---


Not easy. Not effortless. Not always possible to find business you know and like and sell at a good price.

But not exactly Rocket Science either.

That's a a theory - so is buy low sell high. Wahoo for profitable theories

But how goes the application? Made your millions yet?

 

[craft]

Re: Looking for papers on Evidence Based Results for T/A and F/A methods/ideas/hypoth

Expected Returns by Antti Ilmanen. And hundreds of reference paper included inside. The author one of the best in its field and a practitioner not academic. It will be a long read, require understanding of asset pricing theory but no maths involved in the book .

There's a free summary of this book available (180pages)

http://www.cfapubs.org/doi/pdf/10.2470/rf.v2012.n1.1