@Robert, what you presented is not a fight of equals, and you can't score which trading method is best over the long term based on the criteria you selected.

I've presented this chart before, but here it goes again:

.

http://alphapowertrading.com/images/divers/SDE_long_term.png

The chart analyses how the trend matters even in a stocashtic trading environment. The SDE equation used to represent a price series was: dP = µdt + σdW. It's a reasonable and acceptable representation of price movements. The chart shows the value of µdt (underlying long term trend) compared to σdW (its random component) as contributors to the long term price appreciation.

The chart shows two sigmoids viewed from a long term Δt perspective of some 21.8 years. As Δt → 0, fundamentals, viewed as the underlying long term trend µdt, have little value in helping one trade profitably. The underlying long term trend is minimal, buried in all the ambiant random noise generated by: σdW. While at the other end of the time spectrum, meaning as Δt → 21.8 years, the random component contributes little to the overall outcome.

This comes out as the probabilities of winning over the long term should be attributed more to the underlying fundamentals than technical issues. To do this, you need to let your Δt expand in time, meaning increasing the trading interval Δt. And increasing the trading interval is the same as buying and holding for an extended period of time.

A closer look of the raw data in the grey box in the above chart would look like this.

.

http://alphapowertrading.com/images/divers/SDE_long_term_delta_t.png

The above chart shows how much randomness can be found in price series. And the chart does depict well random shocks that can occur at any time in a price series.

Therefore, a short term trader should not care that much about fundamentals, they contribute very little to price movements at his/her level of play. But on the other hand, he/she should understand that they are playing noise when viewed from a long term perspective.

A long term “investor” does not have to worry, he/she is just bound to win primarily due to their ability to hold their respecitve positions.

"In the end, the correlation between the tea leaf-reading discretionary macro guy and the systematic (technical) guy is very high. So what extra value is the fundamental data?? "

Most good macro guys probably use an element of either trying to pick out existing trends and/or run stop losses. Together those obviously give you a higher correlation to trend following. What the good discretionary guy should be better at is picking out turning points in advance, via gut feel / analysis, or what Soros calls reflexivity.

Similarly the systematic fundamentals should help you have smaller positions on when there has been a bullish trend, but the market is overvalued, as it definitely was in 1999. What my rather superficial analysis didn't include is that it's very hard to use fundamentals to pick out absolute valuation. Was the equity market overvalued in 2007?

Systematic fundamentals are pretty good for relative valuation however, but relative value trading is an entirely different kettle of fish, where trends are less likely anyway.

"Robert, what you presented is not a fight of equals, and you can't score which trading method is best over the long term based on the criteria you selected."

My (lighthearted) focus was more on the difficulty of doing each type of trading, than what is best.

I agree that technical trends and fundamentals work better at different scale, and as I have alluded to just now with different styles, outright vs relative value. For holding outright positions of a few weeks up to a year or so trend following is definitely better.

At longer horizons fundamental relative value models seem to do a bit better than technical mean reversion models (value beats winners/losers at deBondt / Thaler time scales of several years).

At shorter horizons technical mean reversion models do okay (classic stat arb), and in any case since E isn't changing short term relative PE just collapses to P.

Anyway it remains true that for a guy sitting at home by himself technical trading is a lot less work, but you definitely lose out in the variety of things you can do.

@Robert: I enjoyed your article--thanks :)

Like most folks who trade retail for a living I don't have access to a department full of "rocket scientists" (as you say) and neither is my agenda determined by a head office with its finger on the global economic pulse. I must say however, having sold a couple of high tech startups in my youth and as usual in such cases retained as the "technical expert" during the transition phase, I don't miss men in suits telling me what to think :)

I can't remember the last time the representative of a sovereign state asked me to engage in a little price fixing and also like most retail traders whose income depends on betting correctly on the financial markets I consider opinions expressed in the media, when not wrong or unhelpful, to be deliberately misleading.

The upshot of this is, since I have no access to fundamentals, I choose to believe current price pattern contains everything I need to know about the probable future path of price.

Hence my time is spent devising bots to ferret out sentiment pertaining to fundamentals hoping to align setups with the cannon rather than with cannon fodder, especially prior to news since the aftermath of news (these days industrial bots reacting to other industrial bots) tends to be as hospitable to retailers as X-ray pin-down in the initial minutes of nuclear war.

It is definitely much more harder to implement prediction algorithms based on fundamentals, because:it means predicting human interactions, behaviour and emotions.

Each model is some kind of simplification, because in reality there is (almost) unlimited number of factors and each of them has different weight. What is even worse - this weight of each factor is dynamically changing depending on changing environment and context.

Fundamental predictions are much more demanding, because it is combination of deep undestanding of financial markets, economics and psychology. Implement something like that on professional level is much harder than any technical analysis, where mostly mathematics, statistics, and trading indicators are involved.

Best regards

Stefan Simik

"It is definitely much more harder to implement prediction algorithms based on fundamentals, because:it means predicting human interactions, behaviour and emotions.

Fundamental predictions are much more demanding, because it is combination of deep undestanding of financial markets, economics and psychology. "

When we're predicting prices aren't we always predicting human interactions, behaviour and emotions? They never appear in an explicit way in eithier kind of model (except perhaps in some fancy agent based model), but that's what its all about.

In fact I think most technical models rely on pyschological effects more than fundamental. For example trend following works because of cognitive biases summed up by prospect theory. You don't need to believe in any such theory to trade fundamental models.

Definitely it helps to have more understanding of the markets to build fundamental models; but I have also seen a lot of bad technical models built by people who believed that you didn't need to understand the markets to create such models. Too much maths and not enough market knowledge is often a route to disaster.

Rob

I have found that a large number of retail investors just do not feel comfortable (and never will) handing their hard earned money over to someone else to manage for them. Due to this fact, they also look for the path of least resistance to making money...which is definitely technical analysis. I believe a thorough understanding of both is necessary for long term financial market success but your time frame and the amount of capital you have at your disposal play an important role.

The problem is mathematical. Some of the fundamental data is updated quarterly and then delivered late while prices can change in a nanosecond. If you want to use fundamentals, you need to have a long term view, not just expressed in months but in years. This in itself dictates to hold on to positions for a long time, and also dictates your trading limitations.

Whereas technicals can answer any trading whim you might have on any time level. But it does say that you will not make a big profit flipping shares at the millisecond trading interval unless you are ready to trade a huge volume of trades. Nonetheless, you are bound by the following limiting equation.

A(t) = A(0) + m*Q*$AW – n*Q*$AL - $Expenses

where m + n = total number of trades, Q the quantity traded, $AW (average win) and $AL (average loss) per trade. What ever your trading method, what ever the market you play in, it gives the total profit or loss added to your initial capital A(0) at risk.

For any long term perspective, you simply fill in the numbers you already have. Say you have already done 100 trades in real life or 1,000 in a simulation. You know how long it took to make all these trades. And from there you can extrapolate to get a ball park figure for A(t) for any duration. Note that the above equation has the same output as: A(t) = A(0) + Σ(H.*ΔP) which is a more concise view of the same thing.

You want A(t) to grow, you don't have that many options. Your first constraint is A(0), your initial available capital. When you start, A(t) = A(0) and if you lose A(0) you are out of the game. So your second constraint is strict: m*Q*AW – n*Q*AL > - A(0) meaning that your trading activities should at no time reach a level of loss greater than – A(0) for the whole duration of executing your 1,000, 100 or any number of trades. Your third constraint is: how many trades can you make over your average holding period. This is a limiting factor often overlooked. If your trading strategy makes 100 trades in 10 years time, don't expect to rival Mr. Buffett!

What I often observe is that people trade on a linear and sequential basis. It's like taking slices of time of a price series where you are in a position, then liquidate and start at a future date a new position. If you put that type of numbers in the above equation; you will soon realize the your overall portfolio return percentage is decreasing with time. As if your trading strategy was breaking down in time when in reality the method of play is the cause.

A trading system needs to be consistent within its trading environment. If you want to play on a 5 minutes bar fundamentals will be of little use. Chasing momentum instead by any means you can devised could be more fruitful. And what ever method you want to design, backtesting it thoroughly would be kind of a requirement before going live. If your trading strategy could not survive the past, on what basis can you say it will survive its future?

I have been using exclusively technical trading models for years. However my wife is of the "fundamentalist" gender, looking at fundamentals exclusively. So the household is tied just like the "head to head" contest.

I have however discovered over the past year that if I add a bit of fundamental review to my technical approach, it appears to put me in a better stead. Specifically, for investments targeted at the 2-6 month range, it would appear that those in the lower third of the P/E historical range, as well those with a price lower than Intrinsic Value, perform a bit better. They start off with the benefit of good technicals, and get the value-based kicker.

I agree with the replies here which are explaining why technical trading is usually chosen...

However, there are various traders who are basing their investments purely on economic data releases, building a portfolio just on this approach. I also have some experience in such a trading philosophy, which I used in the past years...

In any case, also for the fundamental trading, some technical analysis is inevitable, in order to be coherent with the money management we would apply in the fundamental portfolio.

Concluding, both styles can work for sure, it depends how skilled are we to build an edge for any of them.

Very nice post!

But I disagree with "either technical or fundamental".

A correct approach is "both technical and fundamental" (actually, this implicitly follows from the post).

Some epic examples:

1. Renewable energies in Germany. Due to aggressive government policy the largest energy companies like E.On and RWE are half-dead. Though - just from technical point of view - there were nothing special.

2. Prior to the crisis-2008 the spreads between German and Greek bonds were negligible. But in Greece they produced nothing, retired at the age of 50 and so on...

What I - a small retail investor - do is the following:

1. Look at the charts looking for attractive stocks from the technical point of view.

2. If I find something I read the annual report of these companies and the recent news about them.

3. Last but not least I look where the DJ and the DAX currently goes

More on this in my book: http://www.yetanotherquant.com

Fundamentals is what defines technicals. We can see all those technically exploitable regularities in the market due to the fundamentals, or, rather, their own regularity. In my opinion a technical strategy that is based on the analysis of fundamental factors is the best. Not only "classical" fundamentals mentioned in the article but also (and perhaps even more important) things related to market organization. In this case my statement becomes valid even for subatomic level (hft, arb, etc.)

@Robert, in the long run, the technical side will underperform even the Buy & Hold. In the long run, fundamentals might help you outperform the Buy & Hold. It's a matter of choice. One has excitement, can provide adrenaline rush, while the other is totally boring, and for a long time. I guess there is a price for everything...

@Alex - great point regarding "market organization" - I generally refer to this as market structure. Having a good understanding of how the market is structured - and more importantly the role each market participant plays - has a meaningful impact on constructing successful investment strategies.

My personal biases regarding fundamentals are the lag in data (i.e. using quarterly information to make non-quarterly decisions - underlying assumption is the frequency of decision making) and quality of accounting information. if we drill down even further, you can also include the inability to accurately incorporate corporate structure (i.e. future corporate structure). E.g. If you decompose beta as a function of corporate leverage, you will notice that although two firms may have equal beta's, if and when their corporate structures diverge (future state), the beta's fail as predictors.

The above primarily relates to the short term - over the long term, fundamentals can be a good source of information.

Thoughts?

Antonio, I agree that most often the term "market structure" is used. However for some time I've been trying to suggest another term that would include many other market related things — like infrastructure and associated costs for various market participants, regulations and so on. Finally I've come to "market environment" which I suggest to my students and so far it seems like the term is generally understood.

If by fundamentals we assume economical indicators and similar things then yes, the data is lagged. However it doesn't mean that it's useless since all information of this kind is released on strict schedule. Then we could for example calculate a kind of "expectancy factor" for the particular indicator and trade it.

There's much more to it of course, but in general only the market environment in broad sense defines all patterns of behaviour of various market participants and consequently all known technical patterns/methods. This way we can use technicals as a tool to read the market sentiment or exploit these behavioural patterns. Therefore it's not correct to oppose technical methods to fundamental, as the former is the visualization tool for the latter.