Backtesting based on fundamental data

[KnowThePast]

There are pre-canned systems out there which fill fundamental data and time series data and then allow you to manipulate it within reason for basics. These will cost upwards of USD 50k per annum. And, KTP, that's just part of the reason why you are a genius in the flesh.

Thanks again RY, you are embarrassing me now

You've accomplished great things and your thread is loaded with wisdom and contemplation from yourself and others. What a catalyst you've been. I am just so pleased to find you because you work in an evidence-based framework with hard data. Plus you've built a useful univariate analysis tool from scratch. Not too shabby at all. Bloody awesome in fact.

1. Did you work in a funds management environment as a developer?

2. What's your latest area of investigation?

1. I've worked for a software company that wrote funds management software that we then sold to the funds, but haven't worked in a fund itself.

2. Lately I've been playing around with filters for underperforming. Basically, strategies that allow me to pick companies to short. I haven't implemented any of it in my investing yet, but I am considering it more and more.

Another thing that I think has really not been covered much by formal research is backtesting of portfolio management strategies. Most backtest results that are published are done by buying a group of lowest/highest percentile of a specific criteria, then rebalancing every year. This is perfectly fine if one wants to find out whether a specific fundamental filter performs above average. In general. But for practical purposes, I also very much want to know what kind of an effect these kind of things may have:

- have a sell filter, as well as a buy filter. For instance, buy @ P/B < 0.7. Sell @ P/B > 2.0. Sell criteria could even be on a different criteria than buy.
- having that sell filter than allows me not to re-balance my portfolio every year, but to see how funds would flow in and out in a more "natural" flow.
- cash rate on unused funds.
- averaging up/down.
- selling losers after a defined period.
- position sizing, max portolio size, max holding size, etc.
- evaluate on a monthly/weekly/daily basis
- limiting number of trades per month/year/etc

The way a portfolio is managed may have a huge impact on a winning (or losing) strategy. But I haven't yet read anything where this was properly back tested. Most of literature on this kind of portfolio management deal with theretical risk, not performance.

KnowThePast, very impressed with your software, some SERIOUS work has gone into that.

I've coded a really basic backtester in the past using C#/SQL backend but nothing that smooth (threading made my head spin) or anywhere near as advanced.

Thank you for the kind words jet, and for starting this topic!

Believe it or not, my software is done with C#/SQL as well. Previous analytical software that I created, I used C++ for the engine and we created our own file format for the data. Worked really, really fast. My latest software I initially created for a different purpose, so it is not optimized as much. But it handles perfectly well what I need it for, so there was no need to change it.

It's pretty much out of the league of the average punter. Maybe you can group up and share the costs and combine research? If you work this out, it might make sense relative to the cost of time that you would spend developing and maintaining your own systems.

I've considered making a website that hooks into my engine. People could subscribe to for a monthly fee and get access to it. However, the cost of sourcing the data and running the site makes it a little too risky of a proposition for my situation.

3. I used to use this type of stuff as decision aids before I retired, my team built one of those multi-million dollar developments which took two years to achieve - oh, the pain - and it was state of the art at the time. But I'm not a developer. Now I have my own baby version.

RY, you seem incredibly knowledgeable on this. Could you please share you experience in this as well? I showed you mine, you show me yours kind of thing

 

[KnowThePast]

Greetings --

Visit this web page:
http://www.blueowlpress.com/WordPress/links/#books
Just a few entries down from the top of this section is a link to a paper --
"Bandy, Howard, Use of Fundamental Data in Active Investing, pdf file, 2009."
Click the link, and you will open a 30 page pdf file with some of my thoughts on using fundamental data.

The executive summary is that it is not useful because:
1. Granularity of reports is too sparse.
2. Delay between corporate or government action and reporting is too great.
3. Revisions are too numerous.
4. The agenda of the reporter is unknowable.

In addition, there are a few comments about data in the introductory chapter of my forthcoming book. Visit this web page:
http://www.quantitativetechnicalanalysis.com/book.html
Click the link to Introduction, and you will open a 13 page pdf document.

One of the conclusions of the analysis of risk -- not shown in that introductory chapter, but discussed in the full book and in my presentations at the ATAA in Melbourne May 2014 -- is that the data series being traded inherently establishes the upper limit of profit potential, even before a trading system is applied to it. Rapid response to changes in the data, along with short holding periods, are required to extract profit without encountering unacceptably high risk. These trades are typically much shorter than the period between reports of fundamental data. Fundamental data could not be used to generate signals for them.

Best regards,
Howard

Hi Howard,

It always amazes me that public internet forums regularly get such quality contributions. Thank you for your work, I've read it with great interest.

I can now start arguing with you

While I fully agree that your reasons make perfect sense, does the data agree with them? After all, if one was to make trades based on fundamental data, the main thing should be that it works, not that it makes sense. Of course, making sense of it is very important. A very real danger of back testing is that you find something that happened to work during that time period for that group of companies, rather than something that has a real chance of working ever again.

Let's pick something safe. Ben Graham, 80 years ago, wrote about a strategy of buying companies under their asset value with a margin of safety. Preferably under working capital, but let's stick to Net Assets. Numerous studies since then, have shown that this strategy has produced a better than index result in almost any 5 year period in every developed market in the world. Not significantly better, but after 80 years and millions of trades, it cannot simply be dismissed.

Furthermore, it has been shown that the lower P/B was, the better the performance. The higher, the worse. Also, no matter which measure of risk one preferred to measure, lower P/B strategies did not measure as higher risk.

While I do not use this strategy myself, I've ran it in my back test on an Australian market for the last 10 years and got precisely the same result as all researchers did for the last 80 years.

The executive summary is that it is not useful because:
1. Granularity of reports is too sparse.
2. Delay between corporate or government action and reporting is too great.
3. Revisions are too numerous.
4. The agenda of the reporter is unknowable.

I agree, all the data is extremely suspect for these reasons, and probably more. But nevertheless, accurate predictions on a group of companies can be made from them. A very high error rate for analysis of an individual company, yes. But as a group, different story.

My personal favourite explanation lies in behavioural economics. People are too pessimistic about worst prospects and too optimistic about best ones. Which would explain the lower than rational prices being paid for, an average, for cheapest stocks.

But one can't also ignore the predictive power of fundamentals on business performance itself, which has nothing to do with anyone's psychology. For instance, Altman Z score has now, for decades, been pretty accurate in predicting chance of bankruptcy.

One of the conclusions of the analysis of risk -- not shown in that introductory chapter, but discussed in the full book and in my presentations at the ATAA in Melbourne May 2014 -- is that the data series being traded inherently establishes the upper limit of profit potential, even before a trading system is applied to it. Rapid response to changes in the data, along with short holding periods, are required to extract profit without encountering unacceptably high risk. These trades are typically much shorter than the period between reports of fundamental data. Fundamental data could not be used to generate signals for them.

FWIW, I found that value strategies, on average, have an average holding period of 3-5 years for best performance. Which seems consistent with prior research as well. What kind of periods did you test with your data?

I've always wondered why it was 3 years+. Perhaps it's the fact that most analysts concentrate on 2 year forecasts, or perhaps it has to do with a length of a typical business/credit cycle.

I highly recommend Tweedy, Browne paper on this topic, which can be found here:
http://www.tweedy.com/resources/library_docs/papers/WhatHasWorkedFundVersionWeb.pdf

In addition to describing their investment approach, the describe the results of other similar studies in different markets.

 

[howardbandy]

Hi KTP --

Yes, projections about groups of companies have lower variance than projections about individual companies. At the limit, just buy the benchmark index. If you have faith that the market will behave as you hope.

The risk is drawdown. Of experiencing a drawdown greater than my risk tolerance. Of having funds ties up in positions that have large losses for long periods of time. Of being forced to liquidate a position at a serious loss.

Even America's most famous value-oriented investors had drawdowns of 50% or more in 2009. They held their positions, having little alternative. Their positions were too large to liquidate, and they were seen as being patriotic for having faith in a recovery. Perhaps them holding their positions did contribute to moderating a steeper decline. But individual investors / traders do not have those limitations or those responsibilities. Rather, my responsibility is to keep my family safe and financially sound.

This recent recovery was as swift as it was entirely, in my opinion, because of government funding, which is distorting interest rates, distorting alternative uses of money, and forcing investment in stocks and real estate. The government funding will end and / or its effect will lessen. I expect a revisitation of 2009 market lows, or lower, but without the followon swift recovery. In my opinion, forecasts of holdings of several years are at risk of government action -- or equivalently -- of world action in reaction to government action or inaction. I would not be surprised if it takes two generations -- an entire adult lifetime for many people -- for equity indexes to recover to today's levels following the next crisis. It did take that long following 1930.

Rather, I prefer to work with fairly short holding periods, looking for high probability opportunities, retreating to cash regularly, reevaluating often, calculating confidence rather than relying on faith.

I wish you continued success.

Best regards,
Howard

 

[DeepState]

I've always wondered why it was 3 years+. Perhaps it's the fact that most analysts concentrate on 2 year forecasts, or perhaps it has to do with a length of a typical business/credit cycle.

MOM reverts after two years to three years instead of extrapolates in the cross section.

Markets display excess volatility. Earnings growth and P/E contraction/expansion are highly correlated.

There is no overwhelmingly great theory behind why the reversion occurs in that period. Perhaps the most likely simple theory is where investors initially under-react to news, see the movement, hop on the trend, then overreact and then realise it happened. But this theory does not suggest a time frame. It is an observation that holds.

Behavioural science experiments undertaken in an agent based setting produced these results in gross terms. So it's rationale may be found in the exploration of agent based dynamics. More realistically, it has also got to do with the time horizons and peer relative nature of big blocks if the market. Insto makes up around 60-70%, mostly by fundies who can't take heat beyond 3 yrs because their client base uses that sort of time period for assessment because they would look silly for holding a money loser over that period and make up reasons to support the action. The alignment is all off kilter, but that's the world we live in.

I believe it is partly due to business/cyclic credit rational that you have put forward.

But...there are deeper reasons relating to within-market concerns that make the cross-section behave as it does.

Auto-correlation of daily results of ASX do not pass statistical tests for auto-correlation at even the 10% level for data sample of the last 10 years. Hence MOM at aggregate level does not seem present, although I have found cross sectional mom across markets many years ago for a naive signal. This had modest predictive power.

 

[DeepState]

RY, you seem incredibly knowledgeable on this. Could you please share you experience in this as well? I showed you mine, you show me yours kind of thing

Sorry for the delayed reply,

Back in the day, we had to build our own database because commercial systems miss a bunch of things and have problems with primary keys. Stock codes would be re-used etc. We also had to classify stocks into the correct sectors too.

Fundamental data for 3-way accounts was obtained from a range of data vendors in historical terms. These had to be compared and corrected. Estimates data was obtained from the brokers who were producing those estimates. Proprietary data was sourced from wherever it needed to be sourced from.

The backtesting engine was built in MatLab. The database was SQL and visible through PowerBuilder as well as via SQL query. C# was in there as well. MatLab was also used to form portfolios by manipulating all this data into a trade list. Optimisers in MatLab, though awesome, were not awesome enough for these purpose. Hence optimisation was linked to Lindo Pro which provided commercial grade stuff useful for backtesting and real portfolio construction.

All proposed trades were run through a compliance check via extraction of holdings via PowerBuilder from HiPortfolio files. These were compared against client restrictions. The HiPortfolio files were also source of truth for portfolio construction with information from back office relating to overnight cash received or intra day cash receipt. We also loaded information parcels where a client wanted tax aware portfolio management which were coded for tax rate and accounting convention applicable. These were considered in the trade off between forecasted returns and tax costs amongst other things.

Trade feeds via DFS IRESS allowed direct communication of trades and bookout without manual entry and intra day management of the trading activity of the day.

Then we sipped Pina Coladas and had long lunches. Or, sometimes, took a dive off the Barrier Reef or Bermuda.

 

[craft]

Then we sipped Pina Coladas and had long lunches. Or, sometimes, took a dive off the Barrier Reef or Bermuda.

So you’re now retired and could do those things all day every day except they only retain their true appeal when done in moderation and yet the market obsession can soak up endless time without losing its appeal. So retired from employment but not the market is my guess – question then is how are you going to approach things being a private investor/trader?

Do what you used to do but with less resources and without management fees from other people’s money, that seems an uphill road.

Try and find a statistical niche that is not being employed by the big end of town – from your posts this seems where you’re headed.

Use what you know about the big end against them.

What are the objectives – obvious a competitive spirit there so I guess some sort of outperformance would be the goal. But how far do you want to reach given limiting the downside limits the upside. Are you looking for risk adjusted outperformance or you prepared to put it all on the line to achieve the best absolute return you can?

What’s the strategy and objectives – what are your competitive advantages now to compete against what you used to do when employed? Retail traders/investors obviously don’t have the resources you describe in your working life – I’m interested in how an ex-insider intends to combat their obvious resource and scale advantage now he’s on the other side.

Off topic – maybe if you want to answer (don’t feel you need too) do so in another thread. It would be interesting to also get others thoughts on what they think their competitive advantages are over institutional/hedge etc managed money.

 

[DeepState]

So you’re now retired and could do those things all day every day except they only retain their true appeal when done in moderation and yet the market obsession can soak up endless time without losing its appeal. So retired from employment but not the market is my guess – question then is how are you going to approach things being a private investor/trader?

Do what you used to do but with less resources and without management fees from other people’s money, that seems an uphill road.

Try and find a statistical niche that is not being employed by the big end of town – from your posts this seems where you’re headed.

Use what you know about the big end against them.

What are the objectives – obvious a competitive spirit there so I guess some sort of outperformance would be the goal. But how far do you want to reach given limiting the downside limits the upside. Are you looking for risk adjusted outperformance or you prepared to put it all on the line to achieve the best absolute return you can?

What’s the strategy and objectives – what are your competitive advantages now to compete against what you used to do when employed? Retail traders/investors obviously don’t have the resources you describe in your working life – I’m interested in how an ex-insider intends to combat their obvious resource and scale advantage now he’s on the other side.

Off topic – maybe if you want to answer (don’t feel you need too) do so in another thread. It would be interesting to also get others thoughts on what they think their competitive advantages are over institutional/hedge etc managed money.

I don't think I've actually had a Pina Colada for five years. That world was theoretical. It's not even aspirational. That's how interested I actually am in all that in real life....

The main advantages retail has over insto are, in my view:
+ Much reduced market impact;
+ Longer time horizon;
+ No need for concern over peer risk;
+ No need for concern about career risk; and
+ No frictions between when you find a good idea and when you do it.

The main disadvantages:
+ You don't have monster, commercial grade rocking systems;
+ You don't have as much access to high grade people whose individual insights you can piece together into something that sparks your mind to find the next thing. This is the part I miss the most. It's harder to build on something and you have to be more self-reliant. Given the variance in expertise that may be needed to produce a good investment outcome, this can be constraining; and
+ It's alright now, but I loved working in a team that was unified in objective and free flowing. Those were great days I shall always treasure.

You actually don't need commercial grade stuff to make money in retail. Because market impact consumes 30-50% of raw alpha in equities for even medium sized investors and the pay-off for alpha search is diminishing, a smart investor like you and some others here are actually good enough to pull it off.

My objectives are different. My floor is that, short of war or confiscation of assets...and even then, we will be alright for the next 60 years. I figure I'll be dust by then and want to make sure that my wife is fine beyond it. So there is a liability structure which tries to achieve that as best as possible. It's not sexy, you can see long term yields in nominal and real in local and foreign markets. Some credit risk is acceptable. Some equities (local and foreign), but not a dominating component (approx. 30%), are in the mix. Given these dominate absolute risk for the overall portfolio even at that level, they are actively hedged according to a secret sauce process to ensure that I can mentally handle things like GFC when they occur along the journey. That's bed rock. That's premium related long term investing, developed on scenarios and long term valuation, profiting from risk premiums, or otherwise serving as a hedge to economic outliers (ie commodities), which have and I believe will generate above CPI style returns over the very long term.

Then there are surplus assets which are available for risk. These include equity l/s, commodity l/s and currency l/s. Pure alpha on market neutral. This is much tougher because equity l/s comes with such a financing hurdle that a big chunk goes to the financier. The others are done more like futures, but the edge you get is smaller. I'll pass on methodology in here.

Then, be real about tax....so you have to do various things to sort that out, obviously.

Everything is slow turnover. No day trading. All super-patient. My statistical edge lies within the way in which stocks, commodities and FX edges are found. Each of these is grounded in fundamentals of some sort. They have some technical elements, but nothing like the sorts of strategies I am learning about here. All highly diversified.

 

[McLovin]

Off topic – maybe if you want to answer (don’t feel you need too) do so in another thread. It would be interesting to also get others thoughts on what they think their competitive advantages are over institutional/hedge etc managed money.

I'd say it's a couple of things.

First off, I don't have someone breathing down my neck when I underperform for the week/month/quarter/year. I think that is a huge problem in the mindset of the institutional investor. The corollary of that is I'm able to take positions that have far more volatility because I have a loooong time horizon. This also comes back to temperament. The market shakes out those who don't have the stomach or the conviction behind their position. Which in itself is an advantage. My grandfather use to always tell me that the truth almost always lies somewhere in the middle.

Secondly, I think a fair few fund managers are hopeless business people. They're really just accountants (there is nothing wrong with being an accountant!). The make overly complex models when all you need to do is sit in a quiet room and think. The recent downgrades at CCL make that point really well. There were people on this forum talking about the price discounting at Coles and Woolies, but the fund managers just took the company line, then got burnt.

Finally, size. Having a few million to invest is a far simpler task than having a few billion. My universe is multiples of a large fund's. The sweet spot for me seems to be companies around $80m to $1b. Any higher and there's too many smarter minds working on it, any smaller and it's difficult to sort the noise from real competitive advantage.

Anyway. That's my.

 

[banco]

You actually don't need commercial grade stuff to make money in retail. Because market impact consumes 30-50% of raw alpha in equities for even medium sized investors and the pay-off for alpha search is diminishing, a smart investor like you and some others here are actually good enough to pull it off.

.

What do you mean by this?

 

[DeepState]

What do you mean by this?

Hi Banco

It means that when you are managing a lot of money, it's really hard because you move markets every time you trade. When you do this, you get set at unfavourable prices. This is called market impact. Anything beyond crossing the spread is called market impact as you move through the spread searching for liquidity. For completeness, the spread bit is called 'Half-spread' because the ideal point is right in the middle of the touch to touch prices that are sitting in front priority on the bid and offer. Thus, if you are a buyer and you hit the sell price, you have crossed half the spread.

This market impact is much bigger for an insto of even medium size for anything which is trend following or has an element of this in it. This takes away from your ability to produce an edge after t-cost (market impact, half-spread and commission).

It doesn't work this way exactly, but alpha production has diminishing rates of return for effort. It's much easier to find your first $1 of alpha than you $300 millionth. After a while, try as you might, it's hard to squeeze out another dollar. Meanwhile as you get larger, all of this moves against you as market impact increases and your ability to produce alpha doesn't keep up with either of these and becomes a smaller proportion of your growing asset pool. This makes it very hard to manage money for large asset pools.

In contrast, retail may not quite find it as easy to squeeze out the first $1 or $10,000 or $100k...but it doesn't have to. If you have edge, no two people in this forum that I have read does things too similarly and if they did, it would be small relative to available liquidity. Further, you would have next to no market impact for the most part.

Hence, when you are retail size, even if you aren't as geared up as insto, you can do better because you have less frictions relative to your ability to generate returns.

This is not to say that you can just rock up as retail and expect to make money, of course. You still need an edge and that is no where near as easy as the average trader, live in the market at any given time, seems to imagine it to be. Bending distributions does not create an edge, for example.

Cheers

 

[DeepState]

I'd say it's a couple of things....

...They make overly complex models

An example of the lack of predictive ability despite the 'complex models'. This is broker consensus EPS estimates for MSCI Australia from Thompson Reuters. It equally weights the estimates of, dunno, maybe 15-20 brokers. These are then aggregated into as MSCI Aust EPS using what is called a 'single stock' method. Treating the whole market as if it were a single stock.

In absolute terms, I don't know why they bother other than they get asked to produce something. People who have looked into it find virtually no forecasting power exists. Same goes for market economists and strategists. In fact, if anything, they are contrarian signals. So flipping a coin is probably as good a method for stock forecasting than heavy analysis for the most part. And retail cap flip coins as well as insto. It's not quite the whole story, of course. However here's a graph for your consumption:

 

[DeepState]

I'd say it's a couple of things.

1. First off, I don't have someone breathing down my neck when I underperform for the week/month/quarter/year.

2. My universe is multiples of a large fund's. The sweet spot for me seems to be companies around $80m to $1b.

1. Not married are you....

2. Please check out the below. This is the aggregate performance of small cap insto managers participating in the Mercer surveys. They are killing it too. Who do you reckon is the patsy at the table? I'm not saying you are. Just wondering who do you think is? This result has always puzzled me.

 

[McLovin]

1. Not married are you....

2. Please check out the below. This is the aggregate performance of small cap insto managers participating in the Mercer surveys. They are killing it too. Who do you reckon is the patsy at the table? I'm not saying you are. Just wondering who do you think is? This result has always puzzled me.

View attachment 57842

Not married, and the girlfriend is quite undemanding. At least when it comes to financial matters.

As to two, I'd argue that the small ords does a pretty poor job of measuring genuine small caps. TPG is an almost $5b company and is included in XSO, FBU is a NZ$6.5b company and included in XSO. Meanwhile, it doesn't capture anything under about $150m. I think you can see where I'm going with this. So it's really weighted to what are more mid caps, or large caps in the Australian sense.

 

[DeepState]

1. Not married, and the girlfriend is quite undemanding. At least when it comes to financial matters.

2. As to two, I'd argue that the small ords does a pretty poor job of measuring genuine small caps. TPG is an almost $5b company and is included in XSO, FBU is a NZ$6.5b company and included in XSO. Meanwhile, it doesn't capture anything under about $150m. I think you can see where I'm going with this. So it's really weighted to what are more mid caps, or large caps in the Australian sense.

1. Total sweet spot. Congratulations.

2. Just checked it out. ASX Smalls contains everything from ASX 101-300. Correct on TPG total cap (now $5.3bn per below) but it only has partial float. The 31st company, Independence Group (IGO-AU) falls into your $1bn zone and the smallest stock has a market cap of ~$50m. Extract from S&P Index composition report:



So this stuff is in your indicated zone. However the top caps would dominate the index - agreed. Still, unless we are saying that the first 30 underperforms the rest systematically, the relative performance stats are reasonably valid as an indication of profit extraction. At least - I think so.

Also, micro caps active managed funds have similarly ridiculous outperformance in aggregate. But I don't have a chart than I can show you. I just know the results from the top funds and they are way above any notion of an index in the micro range.

I'm not expressing doubt about your prowess or the wisdom of your strategy. I was just looking at it and curious about it. I'm just wondering who you think you are taking money off (if this is an alpha play) or are you positioning for long term growth in this sector generally which is a game that all can play and get along in? The answer can validly be anything. But you are clearly thoughtful and well considered. Are you in a dog fight with other, less-skilled, retail? Are you in a dog fight with foreign? etc.

Part of the reason I ask is that funds management is supposed to be zero sum less expenses. You know, the usual Vanguard argument. But Australia seems to defy it for professional funds across the spectrum from large through to micro caps. It's puzzled people for a long time. I still don't think we really know why. But it has been like this for decades. Perhaps you could guess - no-one else knows. Who are you and the professional community taking alpha from? In large cap, it's foreign. Retail does alright. But foreign in the small and micro doesn't seem right.

Anyway, penny for your thoughts.

 

[maffu]

Thanks for all the replies, looks like it will be more difficult than I first thought.

I've been impressed with Meb Faber's work and it had given me a few ideas to play around with.

KnowThePast, very impressed with your software, some SERIOUS work has gone into that.

I've coded a really basic backtester in the past using C#/SQL backend but nothing that smooth (threading made my head spin) or anywhere near as advanced.

The best free Aussie data I found was morningstar http://financials.morningstar.com/ratios/r.html?t=XASX:CBA&region=AUS which wouldn't be hard to scrape, but it only goes back 10 years.

The US computstat looks very impressive (and pricey). I read somewhere on a forum that if you joined an investing club you could get access, which I'll look into when I get some spare time.

Morningstar (DatAnalysis Premium) is very good for Australian firms. It has all the financial statement data, but it is not as good with market data. Scraping it is very easy.
Thompson Reuters Datastream is excellent for market data around the world.
Compustat/CRSP are the best databases for US firms.

I think they are all way to expensive for a retail investor, I use them through my university.

All the fundamental screens you mentioned in your initial post have all been used previously. Have a look on Google Scholar for results of those kind of experiments. The general result is the bottom 10% of P/E firms will outperform the top 10% of P/E firms, the result is stronger when using M/B than P/E. No one has a perfect explanation as to why, it may be a risk factor.

 

[DeepState]

1. Morningstar (DatAnalysis Premium) is very good for Australian firms. It has all the financial statement data, but it is not as good with market data. Scraping it is very easy.
Thompson Reuters Datastream is excellent for market data around the world.
Compustat/CRSP are the best databases for US firms.

2. All the fundamental screens you mentioned in your initial post have all been used previously. Have a look on Google Scholar for results of those kind of experiments. The general result is the bottom 10% of P/E firms will outperform the top 10% of P/E firms, the result is stronger when using M/B than P/E. No one has a perfect explanation as to why, it may be a risk factor.

Hi Maffu and Jet 328

1. Watch out for survivorship. Because you can only dump data for firms that are alive now, any database build from webscrape will be inherently biased to survivors. This will skew your results in material ways.

For example, look up CGJ-AU (Coles Myer) which was acquired by Wesfarmers in 2007. The Morningstar portal can't access it.

2. Success of Value has two key explanations: Risk (Fama) and overreaction (Lakonishok). Evidence exists for both. Neither PE or PB are perfect...how could they be? But there is a premium to be earned for both. Low n-tile Earnings yield (PE inverted) can include loss-makers (if included in the measure) or companies that are marginally profitable. You can't tell if they are ramping up or declining to oblivion. Higher n-tiles also suffer this, but it is less of an issue. Dispersion within lower n-tiles is also larger as if to highlight the uncertainty associated with such firms and the difficulty of translating this measure for such companies. Some investors exclude the lower n-tiles when determining P/E based measures for investment as a result of this. The lowest n-tile of P/E (inverted) often exceeds the returns generated by higher n-tiles. PB suffers less from these issues. Given the n-tiles show monotonic return improvement for PB as we proceed down the valuation spectrum...with the same results showing through for other value metrics apart from these, it is less likely that the premium shown for lower n-tile PE is a return to risk, but is just simply noise which should be set aside for valuation purposes because the measure ceases to have meaning in that universe. Doing so improves the PE valuation measure's return.

Strange thing to me is that you'd think a PE or PB measure can't be compared between a gold spec and WOW-AU, say. This implies that grouping should be made on a like for like basis. Retailers only compared to other retailers etc. But Fama and Lakonishok do it across the whole market in actual funds management activity and both are successful. I guess the factors work across industries as well as within for reasons as previously stated.

 

[craft]

I'd say it's a couple of things.

First off, I don't have someone breathing down my neck when I underperform for the week/month/quarter/year. I think that is a huge problem in the mindset of the institutional investor. The corollary of that is I'm able to take positions that have far more volatility because I have a loooong time horizon. This also comes back to temperament. The market shakes out those who don't have the stomach or the conviction behind their position. Which in itself is an advantage. My grandfather use to always tell me that the truth almost always lies somewhere in the middle.

Secondly, I think a fair few fund managers are hopeless business people. They're really just accountants (there is nothing wrong with being an accountant!). The make overly complex models when all you need to do is sit in a quiet room and think. The recent downgrades at CCL make that point really well. There were people on this forum talking about the price discounting at Coles and Woolies, but the fund managers just took the company line, then got burnt.

Finally, size. Having a few million to invest is a far simpler task than having a few billion. My universe is multiples of a large fund's. The sweet spot for me seems to be companies around $80m to $1b. Any higher and there's too many smarter minds working on it, any smaller and it's difficult to sort the noise from real competitive advantage.

Anyway. That's my.

Thanks McLovin and RY for your thoughts on this.

I think my advantage and it’s not just an advantage against institutional management but also against most of what I observe in other retail participants is that I can disregard price because I’m valuation focused (defined as buying future discretionary cash flows for less than they are worth) whilst nearly everybody else is to some extent price focused and that puts me on a road less travelled. No need to be the brightest and fastest, the fruit on that road hangs low and most of the time in abundance.

Even good value driven fund managers can’t ignore price – watching Peter Hall struggle with what needs to be done to retain funds under management is very telling.

And statistical/system based valuation as discussed on this thread isn’t really a competing valuation approach as such because the measures can never be what it needs to be and that is price to ‘value’. Value is subjective and based on good judgement. Statics can tell you in which way people generally err i.e. the low percentile P/E’s out performing high P/E’s etc but that’s not valuation – that’s market prices assigned a value designation – it doesn’t capture the essence of value just a glimpse of how things on the whole are predominantly mis-valued.

For as long as valuation stays a skill/art based on judgement, requires effort and lacks herd reinforcement of your conclusion in the form of instant price validation, I suspect it will remain a road less travelled.

 

[DeepState]

I think my advantage and it’s not just an advantage against institutional management but also against most of what I observe in other retail participants is that I can disregard price because I’m valuation focused (defined as buying future discretionary cash flows for less than they are worth) whilst nearly everybody else is to some extent price focused and that puts me on a road less travelled. No need to be the brightest and fastest, the fruit on that road hangs low and most of the time in abundance.

And here is an indication of your profit opportunity based on rationally discounted dividends (I guess it's a decent proxy for FCF). (From Shiller, "American Economic Journal", 1981):



It's real. Go well....

 

[sinner]

Who do you reckon is the patsy at the table? I'm not saying you are. Just wondering who do you think is? This result has always puzzled me.

That's because you obviously haven't read Seth Klarmans "Margin of Safety" wherein he explains all of this at the beginning of the book. Some excerpts

re smallcaps

Most of the major money management firms consider only large-capitalization securities for investment. These institutions cannot justify analyzing small and medium-sized companies in which only modest amounts could ever be invested. To illustrate this point, consider a manager at a very large institution who oversees a $1 billion portfolio. To achieve rea*sonable but not excessive diversification, the manager may have a policy of investing $50 million in each of twenty different stocks. To avoid owning illiquid positions, investments might be limited to no more than 5 percent of the outstanding shares of any one company. In combination these rules imply owning shares of companies with a minimum market capitalization of $1 billion each (5 percent of $1 billion is $50 million).

At the beginning of 1991 there were only 559 companies with market capitalizations this large, a fairly small universe.

I refer to this type of limitation on institutional investors' behavior as a self-imposed constraint. This one is not, however, a completely arbitrary rule adopted by managers; the size of the portfolio dictates such a restriction. Unfortunately for the clients of large money managers, like the one in this example, thousands of companies are automatically excluded from investment consideration regardless of individual merit.

But more importantly on the general "who is the patsy at the table that craft takes money off when he's winning":

An important stock market development in the past several years has been the rush by institutional investors into indexing. Indeed this trend may be a major factor in the significant diver*gence between the performances of large-capitalization and small-capitalization stocks between 1983 and 1990.

Indexing is the practice of buying all the components of a market index, such as the Standard & Poor's 500 Index, in pro*portion to the weightings of the index and then passively hold*ing them. An index fund manager does not look to buy or sell even at attractive prices. Even more unusual, index fund man*agers may never have read the financial statements of the com*panies in which they invest and may not even know what businesses these companies are in.

Klarman posits that the largest money in the market - instos - are indexing. This is literally antithetical to the idea of a stockmarket, i.e. capital formation for the companies with the highest ROC, where instead you invest in a company regardless of its ROC, or any fundamental factor whatsoever. In the index case whoever has the largest market cap (shares on issue * current price) is invested in most heavily. The more people who index, the less efficient the market becomes - stocks outside the index are underpriced and undertraded relative to those in it. You are often allocating more money to companies that destroy value than those who create it.

Not just that but there are plenty of other irrational, similar, reasons covered at the beginning of the book that explains where the largest market inefficiencies that a value investor can capture come from:

* Securities in the index are often playthings for speculators and wannabe arbitrageurs, resulting in significant overtrading relative to their fundamental value and this overtrading (while giving the appearance of liquidity) is often the source of return drag. Low liquidity stocks outperform high liquidity stocks across all market cap quantiles. Low liquidity value outperforms low liquidity growth, low liquidity smallcap value outperforms low liquidity large growth, and so on.
* Funds don't necessarily choose when they sell - long only funds often prefer to retain earnings than sell a security looking for another representing better value and capital formation properties so they will still be holding stocks at "the top" when Klarman is in cash. Leveraged funds might sell an entire holding on a drawdown that's 1 tick beyond their pain point. and so on....
I suggest reading the book.

 

[McLovin]

Ahh thanks Sinner.

In a half-arsed way I was trying to make that point by highlighting the nature of the index; that 50% of the small ords is also the XJO. So there would be plenty of index huggers in the top part of the small ords. While small cap fund managers tend to be stock pickers not index followers from my anecdotal observation, at least in Australia.

Even good value driven fund managers can’t ignore price – watching Peter Hall struggle with what needs to be done to retain funds under management is very telling.

The way he's having to deal with his SRX holding is a very interesting case study in funds management.