Elliott Wave Theory – Critique from a Scientific perspective aka my ramblings
June 17, 2009
– Comments (64) |
RELATED TICKERS: EL
, IO
Prologue ( You can miss this para if you are ONLY interested in EWT)
----------------------------------------------------------------------------
First, I really needn’t do this. However, basically I have an important meeting – and I over-prepared and suddenly have some time to kill, and I love abstract discussions – which I think this one might turn out to be – so you have been forewarned.
Additionally, for someone who has been on the Fool for about 20 months – I have to say I am deeply saddened by the polarization in CAPS and how it has recently devolved into the practice of some virulent attacking/mockery etc. So much so – that I made a promise to myself to becoming a silent spectator as far as the Fool is concerned ( For the TMF staff like Jake and Chris – they know that maybe hard for me to do – or maybe they’ll heave a sigh of relief from the absence of the email barrage!) . However as we know – thar is gold to find in the Fool – you need to look and you WILL FIND it.
Of course a lot of you do not know me at all – so some pointers. ( This is required – otherwise why would you read – a so-called scientific critique of a pseudo-scientific and proclaimed “Cultish” technique)
WHAT I AM NOT
I am not a good trader
I am not an expert in markets
I am not an EWT or for that matter a TA expert
I am not a “cultist” ( I was one of the first to warn Goodvibe of this potential pitfall – and as I said it saddens me and in spite of all his and other’s efforts – things are taken in a different light!)
WHAT I AM
-- I am an investor and a “Strategy-lover”
-- I think this really encapsulates my presence in CAPS and outlook vis-à-vis the markets
-- I am a fairly technical and analytical guy ( you’ll realize that much to your chagrin if you continue reading from here)
-- At one point in time ( read last summer) I think I used to have possibly the highest number of recd Pitches ( you know they exist in CAPS, right – other than the satirical?) on Bank stocks ( You might have noticed my name crop up in that adieu blog by Everydayinvestor – with all due respect – do not take me as an expert there either). They were all FA driven portfolio valuations. You possibly could still find them – otherwise I did a few blogs also, they are still there.
-- I am deeply honored to count within the Fool – people like Mary953, FloridaBuilder, EverydayInvestor, TDRH, Tastylunch, Binve, GoodVibe, DonnerV as my cyber-friends/acquaintances , as well as the TMF development staff– all thru TMF.
-- I am a proud member of the Stinkyfeet crowd ( board – is a disrespect to someone as colorful as madcowmonkey , founder esquire) and the GoodVibe community. The first is a Fundamental (GARP) group , while the latter is mostly TA. I do not know whether you realize that GV welcomes FA – there’s no reason not to.
Background
+++++++++++++++++++++++++++++++
Still there?
So first questions first – are you a believer in TA(Technical Analysis)/FA ( Fundamental Analysis – or in EMH ( i.e. Efficient Markets Hypothesis)? Or Maybe both? ( How can that be – you ask ? Well I can give an answer – but lets keep that for later)
Also note that I put FA in the other bucket also – apart from EMH , ie with any analysis at an individual stock level – you are going against EMH saying that you have incremental information that is exogenous to the current stock price.
Of course you need to be at least interested in TA for this blog to be of value to you.Before I begin a little pre-amble on some of my beliefs.
-- Markets are efficient only on Longer-term basis and only to mundane, apparently readily linked information in the short-term. It’s the intermediate where I think you find aberrations either way. I.e. keep Analyzing , there’s no harm. Just know what you are doing
---With any sort of analysis – you need to understand at the end you are essentially trying to fit a “model-based structure” irrespective of whether it was explicitly mentioned or not.
--THE most critical item with these – then become what ASSUMPTIONS are made in such a – “model/hypothesis” – and you becoming cognizant of those
--Models work – till the point of time – THEY DON”T. And its always because of the underlying assumptions
Critique of Fundamental Analysis (based models)+++++++++++++++++++++++++++++++++++++++++
First of all I love Ben Graham. As an investor it’s the first book you need to read. It helps you understand in basics how businesses operate – ie to generate profits for its shareholders – and what the key drivers are – again on a general basis.
The main point you need to understand vis-à-vis most FA based metrics are that they are “post-facto” analysis – they are looking at what already happened. And then using it to “project” ie forecast ( hence model) what WILL HAPPEN – which is ex-ante.
Now you are getting angry with me , especially if you are a “FAist” (that’s of course not a cult – because of large numbers) - I mean you do your due diligence , nice FCF/DCF etc ( please don’t even get me started with the discounting factor) – and I have the gall to point out a flaw in this.
Please be patient – I am critiquing – not discrediting or decimating. See when you do all these analysis – what you end up using is a stream of information gleaned from the balance sheet and stored in a database somewhere – and then you are using that prior stream of “Time Varying” information (called Stochastic Time Series , where I come from) to project the future – and lets say for some reason you use the latest 10Q datapoint to come out with your “Value”. That my friend – is a classic EMH assumption ( not to the stock price – but to the time variant input/driver to your nice FA model) – or called “Random Walk”/ “ Brownian Motion” etc – ie the last data point in time – is the best gauge for the future – again where I come from called an AR(1) [ Autoregressive Time Series model of order 1 - I am lazy – and this is turning out to be much longer than I projected – so am not able to be as meticulous as Hans ( porte) and do linking – please use lovely Google if interested). Oh I am sorry ! You are more sophisticated/rigorous - like Morningstar – which doesn’t rate anything without 3 year history – well you just moved your time varying parameter from 1 to whatever period you CHOSE ( why may I presume? ) Unfortunately Autoregressive models are by design Trend following – ie they will never catch an inflection point ex-ante. Its those nice graphs on sales pitches which shows Sales projections running up – at a level if extrapolated would mean you are running out of global population real fast!
Net net – you are making an assumption, some big ones – and they are typically unidirectional – just be aware.
Before I move on – let me also mention one metric from Graham – which is really holy grail and is No-Brainer investing rule. Trading below cash-at-hand. Well if you find companies like that – ie Stock price is below the liquid cash ( or cash equivalents) – then you have practically no risk – and all upside to that level , right?. Well consider this case – what if you have a BIG stakeholder in the company in a foreign country , who did the same analysis and takes a stake based on that same assumption. Company is no great shakes as far as business model ( Kodak , anyone?) – so that Cash is the only reason for investment holding. Local currency appreciates by 50% against the greenback – suddenly that below cash in USD – is above par in Local. By introducing Cross-border exchange risk – I just put in unknown volatility risk in the transaction. Your fellow shareholder may be forced to liquidate – while you sit happy with the hope to parity. Can happen – but stock will tank on forced liquidation. And below cash did not work out for the overseas investor. This can work for you also – see a recent post by Porte – where he contends that US markets are dirt-cheap for European investor – due to this dynamic.If you didn’t have this in your model – you made an INVARIANCE assumption – which was not true.
EWP as a Time Series/Neural Network/Systems Dynamics Combo
====================================================
You are completely freaked out, right! I am the guy who’s throwing out “2-dollar” scientific words – just to make this sound legitimate. Just track some tête-à-tête I had with Hans ( Porte) – who is a physicist , on some technical stuff – and maybe your nerves would be calmed a little.
Elliott Wave – as GV likes to point out to me – is a “Principle” and not Theory. Essentially it is a conjecture – typical parlance in scientific world – meaning someone proclaims its existence – till someone can refute it with documented evidence.
Let me tell you - it is almost impossible , post-facto to disprove EWP. Because the graph-structure is so flexible – it will fit any open-ended non-intersecting graph. ( Now you have smile on your face – you non-believer – because you might notice a trace of sarcasm in that last statement!). See this is where EW – is similar to Time Variant N-Nets – it can fit any function in the problem space its dealing with – ie Stockcharts.
The way to make good N-net models and make them avoid “overfit” issues ( ie where you did a perfect job of matching history – but when you start projecting forward – you are dead wrong! Kinda like fitting to a stock chart series of N-daq from 1997-2000 perfectly and projecting 10 years forward – Dow 40K?) – is to control its “complexity” parameters – ie Keep It Simple, Stupid! ( You possibly read the article blogged by EverydayInvestor on Mebane Faber, I personally tested a variation of the 50/200 MA system with some parameters I built, and yesterday came across one developed by fellow fool salvadorveiga –which looked similar in terms of signals generated – which shows simple TA actually works! Conservative, but works) EW tries to do this with its “Rules” – but its still not enough from my point of view – to completely automate this into a model driven structure. Thus making this “individualistic” – and hence cultish to a certain extent. We may come back to this point later – but be aware that we know of one automated EW platform in the commercial domain – and also anecdotally in some Wall St proprietary ones.
System Dynamics ( Look it up – MIT Sloan school of thought. Anathema of Economists all over)
---------------------------------------------------------------------------
There is a simple basic thing you need to understand about EWP ( Without going into esoteric stuff like Golden Ratio etc) - it says motive ( ie trend) waves are in 5 parts. 3 in the main direction alternating with 2 in the opposite. With the middle ( ie the 3rd) being the strongest
So if you understand Investor “psychology” – it is stating that in a systemic sense – the bullish sentiment (Buyers) “loop/cycle/dynamic” and the bearish sentiment dynamic ( Sellers) – work constantly against each other. In a dominant cycle – the outcome of this system – will subdivide into 5 parts ( sub-cycles or sub-systems) – and each loop will have a variable effect in each of those. ( If you want to understand Sys Dyn - go ahead – its also fairly conjectural in the sense that you need to hypothesize about the sub-systems and loops and linkages – and then essentially use data to fit a series of partial differential equations to estimate the level of the individual effect)
One way to estimate the magnitude of these outcome levels is to use Fibonacci Ratios and Extensions ( You could use your own – in the stock market TA does work to the extent , that you need a critical mass of people to follow certain levels for them to be respected – this is a self-reinforcing loop in SD terms. Everyday had a comment of these being random – they could very well be arbitrary – however most recent evidence points to some decent fits. )
Thus depending on the quantity/magnitude of the “extension/retrace” you are getting – is essentially a read on the relative effect the two forces are having on the market trend. So in a motive , you have Wave I – where the new trend gets established Then doubts start surfacing – some will bail , others hang on , while new believers jump in – Wave II. . Then there comes a strong phase Wave III – clearly established winner now. Then comes “trading profits” – Wave IV – depending on severity and extraneous factors – this typically will be mild ( Flat corrections, Running Triangle etc) – followed by some desperate , ( possibly lower volume) – run up , Wave V.
EW believes this cycle repeats itself in the micro timeframe also. I think one reason this is becoming increasingly relevant is due to “trigger-happy” trading these days ( Ultralong had a nice blog on this – reading it was like deja-vu for me , because they echoed my thoughts). Little longer term – it is very difficult to refute this structure.
Now from Motive ie Trends – you go to corrections. This is where it becomes messy. Simplistic thinking – Correctives are 3 Waves.SD explanation– is as follows – you have a Dynamic Loop switchover on Wave 4th – essentially what maybe you thought was a motive and started as a 4th count – becomes dominant as switches over. These should be sometimes self-evident on Wave 3 weakness. Thus you essentially have an A-B-C alternating structure in these.
The MOST CRITICAL point – Trading/Forecasting
============================================
Did I sell you on EW – or have you dozed off by now ( Possibly the later!) ? As I said due to the “non-deterministic” ie if you applied all the EW principles – you will rarely come to “unique” answer – as to what the current wave structure is. There will always be an alternative ( Market will go UP or DOWN LOL!) – however using some of the rules and Ratios – you have a possible framework and can establish Risk/Reward.
That’s it.
Let me tell you – I have not made a ton of money using EW. There are some reasons – most are due to my lack of time commitment, and micro-management and the other KEY important metric in EW – due to the nature of the Correctives being messy – trying to go against the trend can really burn you.
Its a timing system - if you believe Timing works - there's no rhyme or reason EW won't. If not , well - that OK too.
Conclusion
===============
---The thing you need to understand about a Trading system ( like say the 50/200 SMA one – mine generated only 54% wins) – it needs just to beat the random (50/50) by a smidge to be profitable.
----I have read and linked ( you need to look for it in GV’s community) – of documented trading evidence on an automated EW platform efficacy – with other TA based methods.
---With a good structured trading template ( used also for the other methods) EW beats others hands-down.
---The trick with EW is to trade the MOTIVES. Correctives are wealth takers – except of Larger Scale.
---No one claimed to be able to fore-tell the market. It’s a trading system.
---Most TA based ( Support-Resistance etc) are Horizontal ( on a graph) methods – with exceptions of MAs, Channeling etc ( which are time series ones again) – which are verticalEW simply combines the Time Series aspect with the Horizontal thru Fib Ratios.
Whew ! That was WAAY LONGER!!!!!!!!!!!
RSS Headlines
Fool UK