Contents The Trend IS Your Friend Trend Following Types of Trend Following We Are Wrong Until Proven Correct Precedents and Confirmations Modeling the Future The Determination of Optimal Stops With Donchian Channels Long Term Dow Industrials Gold Trend Research Trend Following Gold Short Term What is Gold Gold Model Leadership Change Safe-Haven Gold Models McClellan Oscillators Usefulness of the VXZ to VXX Spread Ratio Monthly Close Moving Average ETF System Death-Cross Indicator Secular Bear Market Developing and Practicing Patience
The Trend IS Your Friend Timing the markets for buy and sell points is often very difficult. But if one has a method to determine the trend, mistakes in timing are usually easily overcome by the movement of the prevailing trend because a trend in motion is more likely to continue than to reverse. Determining the trend may seem simple when one sees a beautiful trend channel such as that in gold. However, the reliability of trend lines and channels by themselves is often low. They are often violated repeatedly, only to see a reversal of the retracement in the direction of the trend. Also, if one waits for enough significant highs and lows to draw a meaningful regression trend channel or trend line, it is often late in the game of the trend in whatever time frame one is attempting to trade. A trend is produced by a progression of new highs and higher lows for an uptrend, and the reverse occurs for a downtrend. The slope of a trend is determined by the swings of the highs and lows on either side of the central regression line. The most predictable movement in all markets is the return to the mean following a high or a low. But it is by using the highs and lows that a trend is entered, followed and exited. Most investors and traders feel that it is important to buy low and sell high. This is not the way to success in the markets. Famous trader Bernard Baruch once said: “don’t try to buy the bottom or sell the top; it can’t be done except by liars”. Attempts to buy the low are most frequently failures leading to large losses because one is not trading with your friend, the trend, but against it. To buy low or sell high is reliant upon hope and successful prediction of the future. Both are the most common causes of lost hopes, dreams and capital from trading unsuccessfully.
Trend Following Trend following is a method of trading and investing in which the trend is determined by various technical analysis modalities such as moving averages, trend lines, regression channels or high-low channels. When one is a trend follower, the method is systematic and not discretionary. That is, it is based upon a specific signal which has been determined to be successful upon historical data. The trader does not try to buy low and sell high, but waits until a trend is established before making an entry. An exit is made when the trend is signaled to have ended. Price determines trend. Price is occurs over time and movements in price are related to volume. Derivative indicators are calculated from the basic price data: open, high, low, and close. These four data points are used to create bars or candles with which to graph price changes. Richard Donchian did the initial research on high-low channels and is considered the father of trend following. He was an economics trained independent market analyst and trader who developed his rule-based method of technical trading in the late 1940’s. He first used moving averages to establish the trend. He discussed waiting for a buy entry on a countertrend until the 4 day high was broken or the daily high was broken after a 3 day countertrend move. His idea of using highs and lows to establish entries and exits has been refined by many and is called Donchian channels or high-low channels. The method of trading based upon highs and lows was later proven to be successful by Richard Dennis, even when used by novice traders. These traders were later called the Turtles, and most became very successful traders and money managers. The most important aspects of trading and risk management which are required for trend following are: • Determining the trend • Setting the trigger to buy or sell • Money management and position sizing
Types of Trend Following The trend is determined by a progression of higher highs and higher lows for an uptrend. Those that use moving averages, regression channels, or trend lines need to wait through a substantial portion of a trend before it is determined to be a trend. It takes multiple points along the highs and the lows to establish a visible trend within a regression channel. Moving averages have significant delays inherent in their methods of calculation. Trend followers using high-low channels have a distinct advantage: there is no mathematical delay in the determination of the signal level. When an opitimal new highest high is hit, a buy entry is made. Long exits, short entries and short exits are signaled without delay as well. How is the optimal signal without delay determined? There are two high low methods: number of bars since the highest high or lowest low and the percent from the highest high, lowest low or alternatively, the close of a bar. In the number of bars method, used by Donchian and Dennis, a trigger signal is generated from an optimal number of bars’ cycle of highest high or lowest low. The key is the number of bars that optimally will result in a winning trade if a new high or low over x and y number of bars is achieved. In the system below of the monthly Dow Industrials, a buy occurs when a new high of 1 tick above the highest high of the past 3 months or completed 3 bars. A sell occurs when a new low is printed, 1 tick below the lowest low of the past 48 completed bars, or months.
We are Wrong Until Proven Correct Famous but anonymous CBOT trader, Phantom of the Pits talks in his on-line book about how we must allow the market to verify correct positions. He says, “we are wrong until proven correct.� We are proven correct by the breaking of a precedent. [Precedent: An act in the past which may be used as an example to help decide the outcome of similar instances in the future]. 3 http://en.wiktionary.org/wiki/precedent Donchian Hi-Lo channels provide the visual precedent of a downtrend in progress. We are wrong to try and enter against a trend and the channel keeps us from entering long- until the precedent is broken and entering long is proven correct. This is the most important feature of Donchian HiLo trading: when we are wrong about the continuance of a downtrend, the market has verified the correctness of entering the position long. Below is an example of a precedent set by a Donchian channel, using the percent method, set at 6% above the lowest low. The channel sets the precedent by indicating that each time prices approached the channel [the action], a similar outcome [the reaction] was produced by rebounding down from the channel line. When the breakout occurs, the precedent is broken, we are proven wrong that the downtrend would continue, and a surprise trade is engaged because it has been proven correct. When we think we are smarter than the market and take an incorrect trade against an unbroken precedent or fail to react quickly to a broken precedent - a Donchian channel break- we become losers. We are winners when we realize we are in a losers’ game and expect to be wrong, Phantom said. 4 http://www.webtrading.com/phantom/chapter5.htm
By optimizing the data of the vehicle to be traded, we are establishing the precedent from which the trading oscillates. By following a new precedent which has been proven correct by the breaking of an old precedent, we are trend-following. We are wrong to act against the new precedent as it is continually proven correct by reactions away from the channel. When the trend fails to prove itself correct and penetrates the channel, we exit short and enter long, since the latter is now proven correct. Precedents are continually proven correct not only by the repelling of prices away from the optimized channel, but also by serial confirmations from the breaking of precedents of progressively longer duration. This is done by optimizing multiple time frames of the index, stock or futures data.
Confirmations of this Dow rally are: 1] 6747- 4000 tick YM- x bar hi Donchian buy [ACHIEVED] 2] 6882 -YM 6% HiLo which has kept the index short from 9088 [ACHIEVED] 3] 7186 - 6.5% HiLo on daily [short from 1/14/09 @ 8375] 4] 7615 - 13% HiLo on daily [short from 1/17/08 at 12300] - IT signal 5] 9089 - Monthly HiLo xbarHi LT signal
Precedents and Confirmations A study of Dow retracement precedent history and confirmation levels can be extremely helpful in proving a position to be correct. One can use optimized confirmation levels for trend-following position increases and have a much greater likelihood of success than one can achieve with bottom-picking. This is true, because the confirmation levels are determined by optimized high-low systems of various time frames that determine the best risk/reward entry and exit levels. These systems use two separate Donchian [HiLo] methods: an optimal percent above the lows [or close] in a downtrend for a buy [percent retracement or trailing stop level], or one can use the highest high within a time range [cycle] of an optimal number of bars for a buy. Waiting to buy until the likelihood of a new uptrend has increased sufficiently to make the risk reward of a trade optimal is much safer than trying to pick bottoms and “average down�. Confirmation levels to watch that would indicate a long position is correct and which would cause a prudent investor to reduce and drop hedges, then get progressively more long are listed below as of 3/7/09. The market is what proves a position correct. No one can predict the future from charts very well for major trend changes just as they occur, but if one increases long exposure as a trade is proven correct CONTINUALLY, you ally yourself with the trend. Levels to watch on the Dow Industrials are: 1] 6882 -basis YG [6.0% HiLo buy on 4000 tick YG- no false buys during the entire sell-off from 9088] 2] 6979 3/4/09 high [4000 tick Donchian HiLo #ofbars best system for YG {dow indu futures} optimized very short term buy] 3] 7186 [6.5% HiLo buy [daily chart best percent hilo system] 4] 7615 [13% HiLo buy [below highest daily close]- best daily system: no false entries since 12,197 sell on 1/17/08] 5] 9089 [best long term system, using #ofbars; buying the 3 bar high on the monthly chart is an 87% successful trade since 1922] Those levels should be important non-Fibonacci-derived prices to watch for confirmation or non-confirmation of position and sizing decisions. A discussion of history and precedent is something totally different from picking a bottom based upon indicators. Bottom picking is dangerous! Making position and size changes relative to the market confirming you are correct by PRICE CONFIRMATIONS will reduce risk and increase reward in trading and investing dramatically. Knowing what the historic precedents are within the markets in which you trade and invest is prudent. Not knowing or caring about important price history precedents is risky and foolish, respectively. On 3/7/09, the Dow challenged a 77 yr precedent: the 50% retracement level from the highest monthly close. This is what I call the Maginot Line, from the name for the formidable French defensive positions established before WWI. The Maginot Line is the optimized percent retracement level from the highest monthly close of the Dow Industrials. It is optimized so as not to have any stop outs since 1930. That turns out to be a 50% retracement from the highest close [13930 in Oct 07]. A retracement more than this amount, we had not experienced for 77 years. The level represents a dark abyss, into which one cannot tell if a fall would be 10 feet or a thousand. One might think of this as if we are peering into this abyss after a very nasty fall already. We are injured and below is uncharted territory which one shudders to get near and most certainly not to fall into. But this is a Maginot Line, since it is not necessarily a support level, supported by major market players, and could probably be easily overrun if strongly attacked. The Maginot Line is not a support level but it is a precedent level. It is named with the given title because it CAN be broken, but if it does break, it will be the first time since 1930. A line drawn 50% below the highest monthly close produces this percent based Donchian HiLo channel. Using this system, one would have bought the Dow in August of 1932 at 72, and would never have been stopped out for 77 years. The closest the index came to being stopped out was April of 1942, when it reversed 1 percent above the line. The only time in modern Dow history that the line has been penetrated intraday was April of 1930. It subsequently declined an additional 73% over 15 months. That is the only precedent in the last 88 years of the 50% HiLo line being broken.
The first chart is the big picture: The Maginot Line is in magenta below the price candles.
The second chart is 1929 to 1944.
The chart below of the past 10 yrs shows the recent severe sell-off and approach to the line of historic precedent.
Precedents: • • • • • • •
The Dow has not violated the 50% correction line [magenta] since 1930. The optimal investment cycles for the Dow Industrials are to buy a 3 month high, and sell a 48 month low Trades using these numbers of bars for new highs and new lows respectively were 86% successful Average time until a new buy signal [3 month high] was 5 months Having reached a new high in the Dow this yr, our present market is resembling 1973-1975 the closest The best corollary for this market is the huge diamond shaped sideways market of 1965-1981. 3 sell and buy signals were given during that time, with the average percent that the ultimate low occurred below the sell signal being 12%, but with a maximum of 19% and a minimum of 1%.
Modeling the Future Systematic trading is based upon a model of the future from which rational choices are made before trading begins. This is done by choosing the optimal settings for trading by applying the model to past data. Subsets are tested independently to assure that the model is robust and not curve fitted with the data. When one applies discipline to a method which produces a risk assessment of the future by using a systematic method, trading can be done with increased confidence that a successful outcome is likely. A model of predicting the future in trading can be highly successful with smaller than 50-50 odds of success. A trend following model can do this because it waits to enter until the likelihood of a successful trade has increased sufficiently to create odds in one’s favor and produces much larger wins than losses. A trend following model will make an assessment of future risk based upon the present trend. “The basic idea of rational choice theory is that patterns of behavior in societies reflect the choices made by individuals as they try to maximize their benefits and minimize their costs. In other words, people make decisions about how they should act by comparing the costs and benefits of different courses of action.”(1) The analysis of trend-following with a high-low channel breakout system provides potential outcomes of trading decisions so that costs and benefits can be assessed prior to trading. By systematizing the rational choice from the combinations of 4 trading decisions [long entry, long exit, short entry, short exit], upon the highest high and lowest low of the data, each over a period of X number of bars, one can see the risk and reward of trading very clearly. Once one is trading, there are no split-second decisions to be made. It is as simple as following the “Yellow Brick Road”. But simple does not mean it is easy. The difficulty is in relinquishing rational choice while trading, waiting patiently, and allowing the trend following model to determine the correct entry point. One can easily see what the correct entry point will be by applying hi-lo channels to the charts.
The Determination of Optimal Stops With Donchian Channels All stocks, indices, and futures will optimize at different percentages from the high as the optimal setting. Although the 8% rule is not bad, it can be markedly improved upon by optimizing each vehicle you trade. The 8% rule in the case of AAPL would cause a lot more trading that most VF’ers would like, I suspect. Being able to objectify your buy and sell plan for each investment is the advantage of using an optimized Donchian system. By just using the eyeball technique, you can find the optimal percent with any software that will place a Donchian channel. In doing that, I found that 25% would give a buy and hold with only two exits, 2/9/06 and 1/22/08 [148]. Notice that the reentry in 7/06 gave up only 2 points for the safety and comfort of not allowing more than 25% to be given back. That percent is a LOT however, and by doing this type of analysis, you can see immediately if you should be even trading/ investing the vehicle. Note that the stop at which one could say the bull market is over is 116, because that 41% stop has never been touched during the entire advance. Of course it can reverse, but the odds favor a full bear market in APPL if 116 is touched, even on an intraday basis. A similar Donchian line on the SP cash chart produced a Maginot Line of 1366. Thus I wrote to the TA group of the importance of that level a week before it was breeched. This is not rocket science and can be done by anyone. It gives you a great deal of confidence to know the historic levels and optimal levels of import for each stock or index you trade.
Finding the Donchian Channel for AUY is just a matter of optimizing the data. AUY gave a daily buy signal on 11/26: the first after the secondary peak at 17. Today it is close to confirming the correctness of this near the recent high of 5.84. The buy occurred on a 24 bar high. That is shown by the green line above the downtrend. Once the buy is initiated, then one watches the red line below the prices, which is a 28 bar low. Since the recent buy is the first since May, it is highly significant. Long trades have been 71% successful and have averaged $25k profit trading 10k shares and lasting an average 96 trading days.
Here is the channel to tune into for Donchian on DRYS. Buy an 18 bar high, sell a 12 bar low. Sell short an 18 bar low and buy to cover a 4 bar high. Using this method, the system sold short successfully 6 of 7 times on the way down DRYS’ waterfall and did not go long once. The buy signal is now 1 tick above the 18 bar high of 11.12. The reason I mentioned the short of DRYS is that it was only a half point below the Donchian channel high buy signal and when a channel has such strong precedent of holding for so long and such a huge drop [6 months, from 90.90 to 3.04], it is very unlikely that it could move through such resistance on the first attempt. At the time of the recommendation, it was 10.44 and had failed to go to a new daily high. A retracement back to 6.37 would not be unexpected [.62 Fibonacci] but don’t get greedy and keep a HARD STOP at 11.13. When it moves through that level, it will move smartly. A cover short stop and buy long at that point is my plan. Targets to take 1/3 out each could be 8.08, 7.23 and 6.37.
The system performance report and equity curve on the daily DRYS system show excellent results trading 1000 shares over a 2 yr period. This is just an example of how the optimal Donchian channel can do the trend following for you.
A standard method by many investors is to pick a random stop level for long exits, such as 8%. The results of trading AAPL on a Donchian channel percent method, 8% below the highest close for a sell, are generated from trades over 7 yrs with winning trades lasting about 2 months and there are 40 trades. Maximal peak to valley drawdown was 73k on a 100k initial investment, so it would take 173 k to trade this system without any significant risk [other than a “black swan� event] requiring re-capitalization. Of critical importance is the maximal drawdown as measured by the ratio of net profit to maximal peak to valley drawdown.
By optimizing for the best performing percent stop, results can be dramatically improved. Below is an optimized Donchian system that uses the percent method for stops, but is 11.6% below the highest high close. This also has a similar 74k drawdown peak to valley, but the net profit is 40K$ higher with fewer trades and higher net profit to drawdown ratio. There were 24 trades with this setting and winning trades lasted 3-4 months. The ratio of win amount to loss amount is very important and this also shows a dramatic improvement with optimizing.
Gold Trend Research Gold futures [GC] received the first buy signal last week, since 3/11/05 when GC was at 446. The signal occurred on overcoming 835.5 on Mon. That was one tick above the prior 10 wk high of 5/22/08. The system sets the primary trend for gold using Donchian HiLo methods on weekly bars, with a 10 wk highest high buy signal, and a 16 wk lowest low sell signal. This occurred as the HUI/GC ratio approached the low level of the Aug 07 panic low, showing HUI to be undervalued relative to GC. The ratio is shown by the gold line overlying HUI in the 2nd subgraph.
The way I use the primary trend system is as an authorization to trade. Now that the weekly system has gone long, I have begun to enter longs in gold stocks again as well as gold futures. By waiting for an authorization in the desired direction of the underlying commodity, one can avoid a lot of false signals trading the related stocks. HUI gave a weekly buy signal in the week of 9/21/07 and did not give a sell during the down action of GC since the all time highs above 1040. Long term gold stock holders have thus remained with the trend of the stocks but have had to suffer through a 25% correction as HUI has been weaker than GC [see red line in bottom subgraph].
Now that I am authorized by the weekly GC trend, and know that the weekly HUI is also long, I look at the daily charts of GC, HUI and individual stocks for pulling the trigger to enter. Both the HUI and GC daily HiLo systems have been long since 5/8 and 5/16 respectively. One can then look at the relationship of GC and DXY on the daily to see what the future might hold. The McClellans [top subgraph] show positive divergences of GC relative to DXY, suggesting a continued weakening of DXY and strengthening of GC.
One can use any two time frames in this manner on both the commodity and the related stocks, using the longer time frame as the trend indicator and authorization to trade, and then choosing a method for triggering entry on the shorter time frame. Since gold stocks have corrected further from the trend buy signal early last week, it would seem wise to enter a partial position now and then really pull the trigger hard when the uptrend is reconfirmed in your chosen stocks. HUI’s long term channel support line is nearby and using a failure of this support at 400 as one’s mental stop would provide protection from a large loss while the potential for a large gain on this trade is increased by the presence of the long term trend signals from the weekly charts.
Gold appears to be ready for a significant breakdown. There is a clear 5 wave advance from the 684.60 print on 10/24. Now that the short term up regression channel has been broken to the downside, the alternate count of an extending w5 into next weeks confluence of Delta high points appears to be very unlikely. The GC Med Hi appears to have come early, which often presages a downturn. The GC range has contracted into a coil with a daily range now only 21 from a recent high of 43 on 10/10.
The failure today to challenge the 931.30 high on 1/30 appears to validate the wave count and would be confirmed by a break of the optimized Donchian channel trend-following system for GC at 890.30.
This projected breakdown of GC appears to produce a significant likelihood of HUI ITD #1 on 2/12 to invert into a low. Although a Dollar ITD Lo and GC ITD th th Hi are expected the 11 and 12 , these may be seen early.
th
th
HUI appears to have already had a 5 wave advance and a very weak 5 wave was portended by a failed 5 wave of 3. Wave 4 support is at 240. Delta theory says that the HUI Med Hi for 2/25 will produce a high greater than 318.76 as soon as 2/11, but the wave structure does not appear to support this. The 2/11 earliest date is a 1 SD range date, and is thus part of the bell shaped curve of a SD range 68% of the time. A 2 SD range to include 95% of the Delta turning points would include the 1/26 high of HUI, I believe. The negatively diverging McClellan oscillators of HUI-GC [6 Fibonacci ratio related length sets in top two subgraphs] are supportive of a very early arrival of the HUI Med Hi. th
Thus, this environment appears very dangerous for gold and HUI longs, at least until the GC Med Lo of Mar 4 , challenging 800 for GC and 240 for HUI.
Trend Following Gold Short Term
The best way of establishing entry and exit points from former Hi and Lo levels is a Donchian trend following entry and exit system. One can set up an hourly bar chart of 24 hr GC trading. The optimal number of hourly bars is the 65 bar highest high to buy and the 85 bar lowest low to sell when back testing is done over the past yr of electronic 24 hr trading of GC,.
An easy way to see where the signals are without TradeStation is by using a calipers to easily see the highest high in the past 65 bars. At this time, a re-entry buy is at 798.40.
Here is the Performance Report of trading just 1 GC contract. The acct size is set at 100k, but it would only take a few thousand to trade one contract. The system had 18 trades over the yr [long only], which were 57% successful and having a very good 5.21 win to loss ratio in dollar terms. Also, the profit was 540% of the max drawdown, which is incredible. So this could be a decent model to follow in terms of seeing what the short term signals of GC are.
What is Gold?
Gold is money, capital, and a commodity. This can be shown by how models track GC [gold futures] in price action. Below is a daily chart of gold, in the 3rd sub-graph. The lines overlying the GC price bars are the 3 models:
•Green= currency or money model. •Gold dashes= capital model. •Magenta dots= commodity model.
1.Gold is money. This is the currency model, shown with GC [gold futures]. It is derived as follows: GC = EURJPY / USDCHF or since the currency crosses are ratios themselves: GC = EUR*CHF / JPY*USD These currency crosses comprise the 4 largest banking systems in the world and the most important currencies. Just becsuse one does not use gold to go to the corner store to get gas and milk does not mean it is not money. It is interchangeable with the world’s most important monies and thus the central banks are afraid of the scenario of gold prices skyrocketing. Note how the currency model lags behind the gold move of yesterday and today: strong evidence of a severe short squeeze of the massive hedge fund and ETF investments in short gold [as shown by the extreme short position of large speculators in the COT data]. 2. Gold is Capital. The rate dollar model has a more cyclic nature of tracking gold well. This model is calculated using the major capital investment in the world, the long 30 yr US Treasury Bond and its futures, US. GC = 1 / US*DXY The extreme low level of this model is indicative of the extreme financial pressures on the capital system of the world. 3.Gold is a commodity. Since the undisputed leader of commodities is crude oil, CL is used in this model. GC = CL / DXY This model tracks more cyclically with gold as well, depending upon the CL microeconomics and the world crude oil situation. I use it rather than a commodity index because it leads the commodity indices and trades around the clock. Each model has its uses, but the best trades occur for gold when the models are in agreement. The models work down to the microscopic level of trading, as this one minute chart shows. Thus, gold trades related to all of its three hats: a currency, capital, and a commodity. One can believe or disbelieve various explanations of why gold is this or that, but when you see how gold tracks with these models, a picture tells a thousand words. GOLD IS MONEY, CAPITAL AND A COMMODITY.
The gold models can be used for short term trading to increase or decrease position size in the direction of the HiLo trend system. The graph may look like "too many lines" upon first glance. But I like to think that this criticism is similar to that made of Mozart in the movie "Amadeus" when the duke said "too many notes" in disgust after hearing one of Mozart's incredible improvisations. All of the lines are useful and more easily interpreted by the eye and brain when overlying one another rather than if all of this was put onto 6 monitors. Subgraphs: 1] CL with McClellan pair #s 2, 3 and 5 of EURJPY-USDCHF [green, purple and gold with pair #3 a 0.78 ratio shorter pair for crossovers] 2] EURJPY green line with McClellan pair #4 [magenta with a .78 ratio shorter pair crossover indicator] YM Dow Indu futures cyan bars Gold models tracking with YM indicate safe-haven conditions apply A] Reverse Risk Model red line is CL/US [crude oil / US T Bond futures] = growth vs safe-haven sentiment; declines with increasing risk B] Currency Model =EURJPY/USDCHF= green dot-dash C] Capital Model = 1/US*DXY =blue dots 3] GC Safe-haven models tracking with gold a] Reverse Currency Model=USDCHF/EURJPY= dark green line b] Risk Model =[US/CL] = red dash; rises with increasing risk c] Carry Trade Model = 1/EURJPY*YM =cyan dash [which is JPY/EUR*YM= Yen down with Euro and Dow up] 4] US 5] DXY candles and USDCHF purple line
Gold Model Leadership Change Recent months have been dominated by a currency influenced market and not so surprisingly, the currency model of gold [EURJPY/USDCHF]. Since the beginning of the year, the leadership has changed to the Capital Model of gold valuation [1/US*DXY]. The capital model becomes prominent in gold pricing when concerns heighten about potential inflation or Treasury Bond weakness [US= long bond futures]. This relates to gold’s tendency to be sought and bought in times of uncertainty and great change and particularly when capital preservation in the US Long Bond becomes more uncertain. rd
rd
In this chart, note how the currency model [green line, 3 sub-graph] and the commodity model [CL/DXY, magenta dots, 3 sub-graph] have rd rd continued to weaken into 2009 while the capital model [gold dashes, 3 subgraph] has strengthened dramatically with GC [gold futures, 3 subgraph] during the last two weeks of January.
This leadership change has propelled gold to a breakout of its down channel off of its all time high just before the 2008 credit crisis and panic developed. One can see this clearly in the spread of GC/US in the bottom sub-graph [blue short dash lines] as the McClellan oscillators of the pair continue to strengthen.
This strength of GC has occurred even while the dollar has had a resurgence of power. The latter’s action has muffled GC in dollar terms while the rest of the world has seen new all-time highs in gold.
[from Ron Rosen’s Market Timing Letter]
As GC appears ready to retest the important 940 level [prior downtrend rally resistance], the dollar tries to return to its retest highs [86.81]. A failure to move beyond that level should usher in a new wave of dollar weakness and accelerate the strength of GC in dollar terms.
Gold “Safe-Haven� Models In early December, 2008 after the banking crisis caused a severe sell-off in world stock markets, the capital and commodity models essentially rd decoupled from gold [yellow dash, magenta dots- 3 subgraph]. By mid January, 2009, the currency model had also separated from following gold rd [green line-3 subgraph with gold]. From that point, gold took on a steep rally without the support of its models.
By the end of January, 2009, the de-coupling of gold from its usual models was complete and the placement of two safe-haven models tracked gold well. rd
US [long bond]/SP [magenta dash,] and 1/EURJPY*SP [green dot-dash] are shown in the 3 sub-graph with gold.
A third safe-haven model is also useful by inverting the currency model to USDCHF/EURJPY [cyan dash].
Gold had become a safe-haven choice because of the continued sell-off in stocks. Thus, SP is placed in the denominator [when SP declines, GC th goes up]. Although the US long bond [4 sub-graph] had lost some of its safe-haven status, it was tracking gold quite well, and thus is placed in the numerator of GC~ US/SP. EURJPY is actually EUR/JPY, so the second model places JPY [Japanese Yen] in the numerator, and this conincides with its safe-haven status as carry-traders move funds back to Japan to pay off loans as stocks sell off. rd
The 3 model uses the same concept with CHF [Swiss Franc-also a huge carry-trade money source] and JPY as safe-havens in the numerator, since USDCHF/EURJPY= CHF*JPY/USD*EUR. The models tracking of GC can be used for very short term trading and discretionary re-sizing of positions on Donchian trend following systems, as this 2 minute chart shows. One looks at the highs and lows of the models relative to GC. For example, when a new high in gold is not confirmed by all 3 models, it is very likely a great situation for adding to one’s short Donchian position.
McClellan Oscillators Using McClellan Osciallators of key market internal data can give us a very nice heads up prior to a mojor rally. What are McClellan oscillators and how are they used? McClellan oscillators are used as a means of measuring market strength. They compare two moving averages each for two data sets to measure hidden relative strength or weakness. Historically, the McClellan Oscillator has been used as a breadth indicator, measuring Advances minus Declines. However, unknown to most if not all market participants is how it can be applied to any two data sets that are related or use any two data sets that relate to a 3rd data set for analysis of the 3rd . Thus, for example, one can use McClellans to compare Tick-Trin as a predictor of the S&P. Investopedia says: “The McClellan Oscillator is a market breadth indicator that is based on the difference between the number of advancing and declining issues on the NYSE. It is primarily used for short and intermediate term trading. Usually, a small number of stocks making large gains characterizes a weakening bull market. This gives the perception that the overall market is healthy, but in reality it isn't, as rising prices are being driven by a small number of stocks. Conversely, when a bear market is still declining, but a smaller amount of stocks are declining, an end to the bear market may be near.” In the latter example, the McClellan Oscillator shows this internal strengthening of the market as prices are weakening but diverging from price: that is, a positive divergence is when prices are moving to lower lows but the McClellan Oscillator is not decreasing and has higher lows progressively. A negative divergence occurs in a similar manner with rising prices. The classic McClellan calculation is done in the following manner: “To calculate the McClellan Oscillator, subtract a 39 day EMA (of advancing issues - declining issues) from a 19 day EMA (of advancing issues - declining issues). Simplified, it looks as follows: (19 Day EMA of Advances - Declines) - (39 Day EMA of Advances - Declines).” (2) However, the McClellan Oscillator has a much wider usefulness than its traditional application, and this is the most important discovery I have made in my research to improve discretionary trading success in the McClellan’s application in predicting turns of markets on ANY time frame. It is important to note that intermarket technical analysis of market internals is used most effectively with the trend following model to give the trader or investor the best chance of success. The chart shows the original application of the McClellan oscillator using the Advances-Declines [Adv-Decl]. However, instead of using the apparently random selection of 19 and 39 day moving averages, the six McClellans [Macs] use moving averages that are all at the same ratio [.38] and have the same ratio to each other providing a wider coverage of varying cycles within the trading data. The first subgraph shows the shortest 2 Macs [3:8 red dots, 8:21 green short dash], and the 2nd subgraph shows the 3rd, 4th and 5th length Macs. The longest oscillator is with Dow Industrials weekly price in the 3rd subgraph [21:55 purple, 55:144 short dasy dark magenta, 144:377 golden green dash, 377:987 cyan]. The purple line is closest to the original Mac calculation. This weekly chart of the Dow Industrials shows the various lengths of McClellan oscillators of Adv-Decl. The important issue seen on this chart of the intermediate term Dow, is that every one of the 6 Macs [numbered from short to long term] are showing what is called positive divergence. That means the oscillator lines are showing higher lows while the Dow showed lower lows. The strength of a buy signal is rated by giving the longer term greater importance than the short term. So each numbered Mac gets the same number of points as the number listed. Since all of the Macs are showing positive divergence to price, the buy signal for the intermediate term is a maximal and rare [for the weekly chart] +21 = [1+2+3+4+5+6]. Thus, all of the oscillators which are the calculation of the difference of two moving averages for two data sets are showing progressive buy signals from short to long.
Below, the weekly chart of early 2009 shows a progression of buy signals summating to a rare +21 or a 100% buy indication from very short to long term moving averages of the advances-declines. This indicates there are less and less stocks declining and that the intermediate term trend is soon to change.
To see how soon this trend change might occur, one looks at the daily McClellan set-up. This showed a +10 score at the low, then became very overbought over a month long consolidation following an extremely oversold situation. The ability of the market to become strongly overbought after a prolonged selling period is an indication of strength rather than weakness. The market followed with a retest and positive divergence of the shortest McClellan.
The retest on the 15 minute chart shows a +10 score of positive divergences, suggesting a strong likelihood that the retest may need more time to be completed.
Indeed, it did take longer and the maximally positive bullish composite signal progressed to the daily by early February 2009. At this point, few believed the market was ready to take off as the Dow Industrials retested a second time and went to new short term lows while the A-D again positively diverged giving the +21 composite signal.
Looking at the 60 minute chart and comparing the McClellan oscillator lows with the market lows, one sees all of the Macs showing positive divergence except the shortest two [top], for a +18 composite rating. Still, the market was not quite ready to turn.
When both the 60 and 15 minute charts give the full composite score, the retest should be completed.
On 2/20, the weekly SP McClellan oscillators of the A-D are still all intact: all 6 still showing positive divergences compared to the Nov low in price, IFF we move up from here: ie, the max +21 buy signal is still present [progressive increase of 1 point for each os from 1->6 for a total of 21 maximum].
It is unbelievable, but the same is still true for the daily, with all 6 Macs holding up above their Nov lows while the price went to new lows. A turn is a must at this point, however, or the oscillators will start to break down consecutively requiring the need for further lows and new progressive positive divergences to develop, which would take months to accomplish. The market is holding on by a thread at the edge of a precipice, imo.
A Donchian confirmation of the McClellan buy signals would occur at 872.90. That is the daily system short exit level. A second and stronger confirmation of the correctness of the signals would be a print of 942.10. That level is the system buy signal [green line]. There had been no buy signal on the daily SP system since the sell signal on 9/5/08 at 1232. The chart also depicts the improving McClellan oscillators of SP-EURJPY, showing positive divergences of all but the longest two for a +10 buy signal.
Importantly, the EuroYen cross [EURJPY] which the SP tracks, was close to a breakout of its month long consolidation without new lows. A close above 120.332 was a very bullish for long stocks and the SP.
McClellan oscillators have a much broader application than just a comparison of the 19 and 39 day exponential moving averages of Advances- Declines. We have seen how multiple moving averages can be applied to produce a progression of buy or sell signals using divergences with price. The most important aspect of McClellan oscillators uses the fact that they can be used to compare any two sets of related data. They then become an extremely powerful method for making an inter-market analysis of a related data set with multiple McClellans on a spectrum of time frames. Also, one can make projections on a 3rd data set which is highly correlated with the pair being analyzed, such as the A-D series shown previously. Below is an example of the McClellans applied to Tick-Trin and used as an indicator for the S&P. This daily chart shows multiple positive divergences of the full series of Macs, giving a composite score of +21 [1+2+3+4+5+6].
The 60 minute chart shows a +21 composite indication of the McClellan set, calculated from the 6 McClellan sets of Tick-Trin.
Here, Up Volume – Down Volume is used to see if the retest turn is imminent. It looks very good, but the fact that the 3 longest Macs have not turned up yet makes the call of a turn a successful retest still incomplete.
McClellan Oscillator sets can be used on any time frame in which data is available. It can even be a useful adjunct to daytrading as it shows the typical progression of divergences through the set of Macs as a turn approaches and occurs, even on this 2 minute chart. The underlying data of Tick and Trin are shown in the lowest sub-graph.
Usefulness of VXZ to VXX Spread Ratio Spread line of long VXZ and short VXX shows a very nice tracking with the Dow futures [graphed for regular day hours only], with an excellent signal multiple positive divergences to enter 8/23 on break above recent high of the spread line. At the present, the spread is not moving down with the stocks and has gone to retest the old highs. An exit signal has not been given for the spread because of no negative divergences and no breakdown of the spread’s trendline. We shall see if the negative divergence of stocks to spread line means anything.
The spread line tracks nicely on the 15 min chart as well as the 60 min chart, again, all are charted for day hours only.
Longer term, the spread shows excellent profitability, measured from the 2/5/10 low. The spread would have made 168% vs. 16% for the SP with drawdowns of 24% and 16% respectively. A nice feature of this trade is the way the spread line trends much better than the indices. It may not always track with the indexes as the first 60 min chart shows, but if one can trade the trend and avoid the larger drawdown, possibly by using regression channels, this appears to be a viable method of trading.
ETF Monthly Close Moving Average System The system and testing are from ETFreplay.com 5. The portfolio is my own. Switches within the system are made based upon the monthly close only, or the close of the last trading day of the month. If the close of an ETF is equal to or above the 9 month MA, an entry of one unit [1/10 of the portfolio] in that ETF is made. If the monthly close is below the MA, an exit of that ETF is executed. Portfolio:
Backtesting: From 2003
From 2008
From 2009
From 2010
“Death-Cross” Indicator- Lots of Hype But Poor Results A Google search shows lots of references to the so-called “Death-Cross” indicator or system. With a fairly bearish consensus about the market at this time, this indicator is frequently shown as evidence to support a bearish theme, since it is felt by most to be an indication of a looming bear market. The sell signal occurs when the 50 day simple moving average [SMA] crosses below the 200 day SMA. What is the success of this signal over the past 10 years? I set up a TradeStation system that buys the Dow Industrials at the next day’s open when the 50 day SMA crosses above the 200 day SMA. It also sells the long and sells short on the next open following a crossover down [50d. SMA < 200d SMA]. The short is covered when a crossover up occurs and the long is again initiated. There have been 16 trades over the past 10 years, and they have only been 31% successful. Trading $100,000 per trade generated 28K$ of closed trade profits over the 10 years, or 2.8% per year. The largest winner generated more than the entire net profit of all trades. It had 5 losing trades in a row in ’04 and ’05 and it spent the first 8 years chopping back and forth without consistent gains. This indicates a very poor system over time that would be very difficult to follow consistently.
The equity curve shows somewhat better results because a still open trade from the 7/7/09 buy has generated over 20K$. The problem is, this system is not tradeable when one notes that the maximum drawdown from peak to valley of equity is 42K$ . This is greater than the entire net profit of the 10 year test period.
As a comparison, a Donchian channel Hi-Lo system is presented. This buys to cover a short position and buys long on a 15 day high. It sells/ sells short on a 20 day low. The success of the trades are 52% and it has half the maximum drawdown from peak to valley of the MA system. It trades more often, but it does not have all of its profits concentrated within 1 or 2 trades. The stated annual rate of return is 8% and the system more than doubled the equity within the 10 year test period.
The equity curve is more smooth indicating a more consistent profitability.
I conclude that the â&#x20AC;&#x153;Death-Crossâ&#x20AC;? indicator / system has a poor profitability and reliability profile based upon this test, and should be used as a subjective warning rather than as an objective system. This moving average crossover system compares poorly to another trend-following system, the Donchian Hi-Lo methodology.
Secular Bear Market We have certainly been in a secular bear market of equities since 2000 if the S&P is valued in gold units as well as based upon the definition of no new all-time highs. A secular bear of stocks does not need to see dramatic declines in stocks, as in the market of the 1930s. More commonly, valuation wave revisions occur. It appears that the Curse of the Dow Trading Range, written in 2002, is still in effect, and if similar in length to the last valuation wave revision, it could last another 10 yrs. Adam Hamilton has written two subsequent articles on this topic [contents list pages 6 and 3]. The chart below shows the secular bear market of stocks in gold units. The percentage decline of the ratio of the S&P/GC would take gold above 1300 if the 88% decline of the ratio [blue, 4th sub-graph] in â&#x20AC;&#x2DC;76-â&#x20AC;&#x2DC;80 was repeated in this secular bear market.
Developing and Practicing Patience Developing Patience [much summarized from Steve Scoresby] Patience is the pause between stimulus and response that allows us to get the result we are really looking for. Patience is a virtue because we remain steadfast despite difficulty, and through this we gain the experience necessary to develop character. * Accept yourself and others * Use a consistent philosophy based upon a value system * Stop dwelling on past mistakes and focus freshly NOW on what I can do to change future results * Confront your fears * Believe in your goals and have faith in your value system * The consequences of rational choice improve with the change introduced by patience -------------------------------------------------------------------------------------------------------------------------------------------Practicing Patience [much is summarized from Sue Annabrooke Jones] * Get real- adjust to the situation * Resist emulation of demanding behaviors * Determine where your attachment lies, then develop a plan to remove yourself from its grip * Restrain impulsiveness by listing the benefits of holding back * Self-speak to calmness with mantras and questions toward mastering patience * Creatively visualize toward a beneficial conclusion * Write notes of helpful quotations * By forcing oneself to be patient, the reward IS patience * Affirm restraint of rational choice from immediate gratification ------------------------------------------------------------------------------------------------------------------------------------------Practicing Patience Patience is practiced by delaying rational response long enough to focus on the systematic method that has been proven correct in the past. Focus is accomplished by: * Rhythmic breathing in a 4:8 ratio, accentuating the expiration * Gazing at the situation with crossed eyes * Visualize the correct conclusion *Repeat a mantra of resisted impulsiveness and release from the attachment to action ---------------------------------------------------------------------------------------------------------------------------------------General Guidelines for Developing Patience * Make patience your goal each day * Slow down * Practice delayed gratification * Practice thinking before you speak
Bibliography 1 Rational Choice Theory: Wikipedia: http://en.wikipedia.org/wiki/Rational_choice_theory 2 Investopedia: http://www.investopedia.com/terms/m/mcclellanoscillator.asp 3 Precedent: http://en.wiktionary.org/wiki/precedent 4 Phantom of the Pits: http://www.webtrading.com/phantom/chapter5.htm
5 Monthly close moving average portfolio testing: http://www.ETFreplay.com