Over the past few days I have seen very productive discourse covering a wide range of trading related topics. As humble host of this digital sanctuary one in particular has become a recurring theme for me personally as not a day goes by when I don’t encounter some hare brained train-wreck of a chart posing as ‘technical analysis’. As this is a veritable Pandora’s box of filled with confusion, assumptions, misinformation and deeply ingrained superstitions I will merely attempt to establish a number of basic tenets which should be appreciated as indisputable truths and serve as a quick acid test when evaluating any technical tools into your trading activities:
- Candlestick Charts are at best a rough representation of time series propagation. As such they are deceptive and can lead to misinterpretations of what happened in the past. One needs to be conscious of the fact that volatility as well as participation within one candle can vary widely. This spells true in particular for interval based candles which assign a series of price changes into time slots. Whereas an intra-day hourly candle may comprise over 10,000 ticks one recorded during off hours may sometimes only comprise 1000 ticks or less. This seriously calls into question the validity of interval based charts as market participation does not abide by the rhythm of a metronome.
- Indicators or oscillators which are a product of any time series by definition always have been and always will be lagging. Placing entries purely based on a time series based indicator or oscillator is tantamount to driving on the freeway whilst looking in the rear mirror. Oscillators may only represent a state within a range but rarely account for the explosive trending behavior propelling price outside of normal standard deviation thus producing ’embedded signals’.
- Mean Reversion is an illusion. Price is not naturally attracted to some artificial range or moving average. The one reason why price reverses after trending behavior is due to profit taking. During stop runs profit taking can occur without price reversion. The phenomenon we call ‘mean reversion’ is only observed when selling or buying is met by insufficient opposing demand, therefore driving price back in the opposite direction. As such concentrations of participation at various price ranges can have a large impact on price behavior. To expect price to always revert to a certain point due to ‘mean reversion’ is naive at best.
- Support and Resistance Lines are expected to represent inflection points observed by a critical number of market participants. There more touches the higher its visibility and thus the odds that a price range near a line is expected to accumulate buyers or sellers. However price is not intrinsically affected by lines on a chart but rather by the beliefs and actions of a large number of participants in a market. By drawing a line on a chart you are implicitly expecting other participants to anticipate and then observe the very same formation. Thus you are making a bet on consensus as well as the ability of others to perceive the same patterns and to observe similar price intervals on their charts (e.g. 5-minute, 60-minute, daily, weekly, monthly, etc.).
- Price Channels – see support and resistance lines.
- Retracement Levels such as fibonacci fans or pivot points are based on underlying formulas extracting support/resistance ranges based on prior price movement. Once again you are driving forward whilst looking in the back mirror plus the data analyzed is based on a particular time period, e.g. one day, one week, month, etc.. This should not be confused with statistical trading which is an advanced research topic among large financial institutions.
- Moving Averages only matter if they are being observed. It doesn’t matter if you follow a 20-day or a 100-day SMA. Similarly you could be following a 21-day or 101-day SMA – or a 25-day and 120-day SMA. At which point does your moving average lose its credibility/value? IF a particular moving average coincides with a number of turning/touch points in a particular time series then it is either due to consensus among market participants or pure chance. So instead of looking for a particular moving average to predict price movements instead seek one that best represents existing market behavior. The more touches the better – with recent touches most likely representing equilibrium separating buying/selling ranges.
- Chance and Random Events affect price movement. Sometimes an accumulation of random price movements over time produces context which is hence interpreted as a support or resistance zone and thus becomes one. Human beings constantly seek out patterns and will invent some if presented by purely random data. Appreciate the fact that random events may play a large part in how price movements are being interpreted by market participants.
- Buying/Selling ranges are separated by equilibrium zones which I often refer to as Inflection Points. All trading ranges are inherently a product of volatility (i.e. speed/velocity ratio) and participation (i.e. volume/pressure). Within inflection points entries are favorable as price tends to moves slowly until price follows the easiest direction in response to shifting market behavior. As such they could also be referred to as transition zones. Be aware that the odds of resolution shift constantly as time and price propagates. An support zone which is only tepidly being observed can quickly make way to a selling frenzy. In other words – a support zone only should only be considered as one as long as it offers support, the same applies to resistance zones. Some of the best long trades happen near resistance zones and some of the most lucrative short trades occur near support ranges.
- Volatility is comprised by price propagating at a particular speed and velocity. In a sideways market price moves at low velocity in comparison with the ongoing speed as represented by interval based candle ranges. In a trending market price moves at a higher velocity/speed ratio – meaning that we have high speed and a directional vector. Be aware that volatility differs significantly between period and tick charts. If price moves 3 handles within a one hour candle but most of those 3 handles happened in 500 ticks (within only a minute) then should the entire hour be labeled as volatile? The answer to that should be reflected in your trading system.
- Continuous Futures Contracts do not exist. They are helpful in establishing a larger picture but do not represent reality as you are looking at several futures contracts which have been merged to accommodate a larger time period. This complicates technical analysis in the futures markets for obvious reasons.
- There is no such thing as a Forex Market. Rather there exists a number of Forex islands which are all loosely connected. Prices as well as spreads can vary extensively as all forex traders are inherently prisoners of their particular forex island.
- Stocks And Its Underlying Companies are only loosely correlated. When you are buying a stock you are holding a derivative financial product which represents the collective opinion of the observants of the underlying company. Although senior management is always incentivized to increase the company’s stock prices it is important to realize that the two are in fact separate entities. There are great companies with undervalued stock prices and crappy companies with high flying stock prices.
- Assumed Risk has many faces and dimensions. For example futures and forex markets are commonly considered to be risky due to their inherent leverage. However I do not recall ever having witnessed a futures contract or even a currency dropping to zero (the latter usually go the other way when failing – thus admittedly affecting a cross). On the other hand the annals of the financial markets are littered with stocks that have done just that. Thus assumed risk should be clearly defined based on market, personal requirements/limitations, time frame, etc. Highlight various aspects of a market or instrument that may harm you and then prepare accordingly.
- What Everyone Knows Is Not Worth Knowing. It is rather surprising that such a thing as the financial media exist as any information available has exactly zero impact on your trading performance.
- Chaos And Uncertainty are the norm in all financial markets which stands in stark contrast with the expectation of most market participants who instinctively seek out order and predictability. Be cognizant of the fact that all financial markets operate on the notion of harming the largest number of participants.
- Zero Sum Game – In order for minority to produce significant gains a majority has to be on the losing side. Ponder on this irrefutable fact every single day.
- Hard Work is no guarantor of success. To quote Einstein: The definition of insanity is to keep doing the same thing over and over again expecting different results. Just because you read dozens of trading books and attended expensive seminars doesn’t mean you are a successful trader. In fact you are a only a successful consumer of trading literature and services.
- Successful Trading is an acquired skill as opposed to an inherent ability. Learning how to trade and manage one’s capital is more akin to learning how to play the piano as opposed to memorizing a book or a poem. The more you practice the better you get at it.
- A trader’s daily activities should be comprised of 40% risk control and 60% self control. In turn the risk control portion is only half money management and one half market analysis. Yet most traders emphasize market analysis while avoiding self control and de-emphasizing risk control. To become successful traders need to invert their priorities. — Van Tharp
I think I’ll leave it at this for today. Plenty of material for Halloween haters – enjoy!
In recent weeks I have been quite prolific regarding the current state of affairs on the equity front. There really is not much to add and if you have been following my work then you should be well prepared and ready to pull the trigger with confidence once equities decide to pick a direction later this week. Thus instead of regurgitating my long term charts I have decided to use this Labor Day as an opportunity to indulge your recent requests for some perspectives on basic trading related concepts.
As I am a big fan of the ‘jumping in feet first’ method this series will cover how one may develop a complete automated trading system. This will not only allow me to cover various pertinent concepts but more importantly put them into context. Most recently I have continued to refine our Mole entry signals in my spare time and therefore I will use some new discoveries as our starting point. This will be a comprehensive journey which we will undertake together as things are unfolding on my end. As time progresses I am going to walk you all through the various steps involved. Each consecutive part of this series will cover important concepts to be considered at each stage of development. From the inception forming the basis of a system, the definition and tweaking of entry/exit rules, the resulting MAE and MFE, the definition of expectancy and SQN (and why I don’t care about Sharpe ratio), back testing, forward testing, tools, etc. We’ll go through all the motions and once we’re done you will not only be able to develop your own trading system but you will also have developed a deeper understanding of what separates the wheat from the chaff.
The best way to teach is to lead by example – at least that’s what they say. Let’s pretend I just came up with a promising new indicator – let’s call it the Mole, not so humbly named after yours truly. I am convinced that there may be an opportunity in developing it into a full fledged trading system. So what now?
Today we will cover the first phase – system discovery – which of course is the most exciting part. You are still wearing your rose colored glasses and are filled with hope, convinced that you have uncovered something truly remarkable. Of course throughout the remainder of this series we will go about smashing many of such dreams but that’s how we roll here at Evil Speculator.
What you see above is a screen grab of my current 1-min Mole indicator prototype. The Mole indicator you are currently seeing on the live Zero Lite runs against a 5-min E-Mini chart and that’s a commonly favored chart interval for intra-day swing traders. However two or three signals max a day may be insufficient for a black box trading system capable of dealing with the type of tape we have been observing in the past few years. There are also other considerations based on expectancy and the frequency of trades necessary to maximize profits during a six to twelve month testing period – we’ll explore that in more detail in a future installment of this series.
It does not take much imagination to realize that this system will be based on short term reversals. In other words the aim of our Mole black box system would be to trigger near tops and lows, allowing us to scalp a few ticks and then exit. To some of you this may sound self evident – after all everyone wants to sell the top and buy the very low. But in reality many types of other trading approaches exist. FWIW – attempting to define tops and and bottoms is rather ambitious and borders the arrogant. Many have tried and most have failed – at least on a long term consistency basis. The ones that really work you’ll probably never hear about as the originators have little interest in sharing. Of course that does not keep us from trying – consider the Mole my humble contribution to the search for the Holy Grail of scalpers everywhere 😉
As you can see from the current edition the Mole nails the tops and bottoms pretty well. The current phase of development is one of manual trial and error. Basically you produce your indicator and find some way of visualization that gives you the information you are after. For instance – on the bottom you see the various signals that comprise the original Mole indicator. A few months ago an email from a subscriber gave me an idea   which in turn resulted in what you now see on the price panel as blue reversal arrows. And that is step one: Your indicator (or whatever you use for your system – you may be only looking at candles) exhibits some type of repeatable prescience in the context of ensuing price movements. You want to exploit that and thus you are starting to think of a possible system. Assuming you know how to code you plan to turn your indicator into a full fledged trading system.
But wait – not so fast. Before you write one line of code (or pay someone to do it) I recommend you spend a lot of time playing with the settings, changing the chart interval, looking for patterns, etc. The human brain has an amazing capacity for recognizing patterns and for putting them into context. Computers are getting pretty good these days but there are certain aspects of the human brain and imagination that still remain outside the scope of even the fastest number crunchers. So use it – get a ‘feel’ for your system. Because the better you understand what drives your system and how changes affect it the less time you will spend later optimizing it. It is tempting to immediately write code and to define a dozen or more settings you plan to later use for optimization. But believe me when I tell you that the time spent planning ahead and simply observing will save you days if not weeks or months later down the line. As many other things in life I have learned that the hard way.
This is basically what I am doing right now. I am still fiddling with various settings to arrive at something that ‘visually’ appears to provide a valid edge. This phase can take anywhere from days to months, depending on the complexity of your indicator/system, your tenacity, patience, or obsession to find the perfect settings. Which do not exist – that much I can assure you. I suggest an iterative approach in which you spend a few days or weeks and then proceed to the next phase, which is initial implementation.
The next part of this series will cover the concept of edge and in particular some theory on expectancy and system quality number (SQN). Both are basic ways to determine the expected profit (or loss) potential of your new system. And without knowing that you pretty much have nothing to rely on but your human subjectivity. Usually not a good basis for success, for we have met the enemy and it is us! Of course a predicate for defining your edge is the creation of entry and exit rules which we will cover next time as well. Once you have developed a sound understanding of how to measure success and failure, as well defined standard deviations of returns, we will be ready to implement and start testing your new system.
To be continued…
Over the weekend the old Convict sent me a pretty interesting write up discussing an increasing amount of trend day signal failures. If you are a sub then you know that this strikes very close to home as my own system sends out a trend day alert to all of you each morning at 10:15am EST. If you don’t know what a trend day alert is then please point your browser to our handy cheat sheet.
Chart courtesy of Quantifiable Edges.
The left side of this chart shows us the profit curve of a system trading trend days between January 2006 and April 2009. The right side of the chart shows the profit curve of the very same system between May 2009 to the present. We’ve gone from a pretty juicy edge to something I would consider a coin flip. And I think most of us would agree with the author’s conclusion – quite obviously since ca. late spring of 2009 those trend day alerts have been a hit and miss affair. The author opines that something has obviously changed – he doesn’t know what and neither does he care. As a quant his main job is to weed out trading rules which are not applicable anymore in today’s fast shifting market dynamics.
I cannot argue with the data – so does that mean that trend day alerts are passé and that you should ignore them going forward? Well, on that end I do approach trading a little bit differently than some of those quant boys. Let’s look at what makes a trend day in the first place.
More charts and cynical commentary below for anyone donning a secret decoder ring. If you are interested in becoming a Gold member then don’t waste time and sign up here. And if you are a Zero or Geronimo subscriber it includes access to all Gold posts, so you actually get double the bang for your buck.
Here’s the chart that actually sends you guys the daily alerts – we are looking at two consecutive trend days, one on 2/29 and the next one on 3/1. On the second lane below the spoos you see me plot the actual NYSE TICK via a simple line chart. Behind the scenes my indicator grabs the first 45 minutes of the tape and then calculates the tick ratio – meaning how many ticks (of the NYSE TICK) plotted below versus above the zero mark. Pretty straight forward – it’s not brain rocketry.
Rather apparent on the left half of the panel is the fact that the first trend day alert was a complete and utter failure. It’s almost as if the HFTs were banging the tape higher intentionally just to turn the tape slightly before the measurement time at 10:25am EST (the chart is in PST – so add three hours). The second half also shows a bit of monkey business and it admittedly wasn’t a hugely profitable, day but in general March 1st had all the call signs of a proper trend day.
Now here’s where it gets interesting – this is a snapshot of the Zero Lite spanning those same two days. And it shows rather clearly that there’s some trouble brewing early in the first day. On the second there’s a lack of participation – but at any rate the signal line remains above the mark throughout the entire session. Again, not a great day but the Zero Lite was minimally supportive of a possible trend day.
Lesson learned: when trading a TrendDay alert make sure you keep an eye on the Zero Lite!
Now let’s zoom out by quite a bit – here’s the daily Zero which gets updated once a day. I expanded the panel a little to capture the entire timeframe between April 2009 to the present. Actually we have two Zero panels here – the one on top is smoothed and the raw signal is on the bottom. As you can see there are striking differences in participation and momentum when comparing ‘melt up periods’ (i.e. winter/spring 2011 and winter 2012) to ‘shake out periods’. The latter show a lot more signal strength while the former are relatively flat. I think it’s a pretty good expression of quantitative easing at work and if you know the inception/expiration dates of both QE periods then you’ll realize that they match up very well – I actually presented some pertinent evidence in the past.
Now this chart is a bit experimental – my main difficulty is that trend days do not occur at regular intervals, rather they occur intermittently. To simplify things let’s assume a standard gaussian distribution which admittedly is being presumptive to say the least. So forgive me when I had to use a bit of creative liberty when mapping Rob Hanna’s profit curve to my daily Zero, but that didn’t keep me from trying. A bit of fuzzy logic here and there often levels the path to putting things into proper context.
In essence I think this presents a fairly reasonable case however and I would very much enjoy Rob’s input on this. Basically what I did here was to extract ‘performing periods’ and ‘non performing periods’. The PPs are framed in green while the NPPs are framed in red. I can say via personal observation that we are smack middle in a NPP right now and it’s one that started late last year, right before the onset of the Santa Rally.
I also spoke to the old Convict over the weekend and he agreed that last winter/spring felt quite a bit like what we’ve seen since December – it’s that 2011 analog idea again. Anyway, both of those periods may just correlate with the small signals we are seeing on the daily Zero, which after all is a participation/momentum indicator. No participation – more opportunity to bang the tape all over the place – simple.
Bottom Line: I don’t think the trend day signal is dead for good, rather it seems to me that it does not apply very well (i.e. almost blindly) during periods of quantitative easing with the purpose of lifting equities artificially. However, when correlated with the intra-day Zero Lite indicator then early warning signs are fairly easily discernible as evidenced by the example above. In essence – during a bonafide UP trend day you never want to see the signal line dip below the mark for more than 10 minutes or so.
If you are not a Zero sub then you may at least keep an eye on VWAP, which should not be breached at all – if possible dips do not even touch VWAP. It’s the inverse situation on bonafide DOWN trend days – you do not want to see more than one or two quick signal spikes above the mark and similarly any attempts to breach above VWAP should fail.
I don’t have much to add on the long term front – all the charts I posted in the past two weeks remain valid. Make sure you keep an eye on the spoos overnight in consideration of Friday’s inside inside candle on the SPX. The high was 1374.53 and the low of the day was at 1366.42. Fair value is about $1.15, let’s round it up to $1.25 and then adjust it to the spoos. Ergo we get a high of 1373.25 and 1365.25 which are your two inflection points.
And as ‘coincidence’ may have it we have two matching Net-Lines on the hourly spoos chart – this should make it easy to structure your trades. To quote Volar – happy unbiased trading!