Friday, October 30, 2009

A Qomplete Reversal

Yes. . .I know how to spell. I post this only because it reflects a complete change of character and momentum in the Qs. . and it happened in one day.
We got another nice clean VIXEN signal (per yesterday's post) at 10:10 this morning, this time to the upside, and it held for the remainder of the day.
Plus, there were two nice bull trend confirmations to support the probability of an uptrend day.
1 . .the 3 composite MAs all remained upslope for the duration of the day, and
2. . the NYAD never fell below the initial opening pullback line. This is particularly significant and reflects solid buying strength (or short covering) throughout the day.
Although volume was about 15% sub par today, there was solid buying into the last 15 minutes so I'm expecting a carry over of the buying into Friday.
The Qs month end Tickler system is flashing a late buy signal.

Thursday, October 29, 2009

An Old Bear Pattern

Taking a little break from my pair trading exploration, here's a snapshot of yesterday's Qs and the NYAD as seen on 5 minute bars. Quite a bit of quant blog traffic has been in an anticipatory mood, awaiting a bounce off oversold levels, but this chart should clearly show how unlikely such a move was on Wednesday.
For my own part, I was hesitantly bullish going into Wednesday's open if only because the NYAD typically falls to .08 to .10 levels prior to upside reversals and the previous intra day lows of the NYAD over the past few days have not fallen below .30. That changed dramatically yesterday as the NYAD made a steady descent to close at .11 as the VIX rose a full 12.5%.
For daytraders the real clue this was going to be a downtrending day was confirmed by the early VIX/Qs cross at 11:00. That was followed by a little whip saw and then another kiss off of the VIX at 12:00. The significance of these crosses has been examined in previous posts and reflects a pattern I call the VIXEN.
There were actually about 2 weeks of VIXEN related posts so if the concept intrigues you, scroll back in the archives for more details and examples.

Wednesday, October 28, 2009

FXY Dashboard Results

First, a quick update of the PDQ Dashboard signals for the Qs as of the 10-27 close. The signal line has a strong alignment of shorts, although today looks like it could present a potential pivot low reversal.
Now, back to the real work at hand. . . .
Following up on yesterday's FXY dashboard spreadsheet, here's a snapshot of the signals generated by a consensus of 4 of more of the seven pair signals.
The green bars reflect LA signals, while the red bars reflect SA signals. A closer look at yesterday's spreadsheet reveals some slippage between the actual entry and exit dates for the various pair trades, which is to be expected as the inherent volatility of each pair is unique.
I've also shown some of my classic technical indicators on the chart, just to see how they align with both price dynamics and the dashboard signals.
While the FXY dashboard didn't capture all of the FXY price cycles, it did manage to trigger 8 of 8 winning Long trades and 3 of 4 winning short trades. Those are pretty respectable results IMHO over a 6 month backtest period for a 7 pair dashboard that averages a 13.5 N day cycle, almost 2 weeks to the day.
In future iterations of the dashboard, I'll probably cull N day pairs greater than 10. This will increase the net number of confirmed trades and, hypothetically, reduce the overall risk exposure over the duration of the open position.
So far this little experiment using a woven basket of highly correlated FXY pairs to trade the underlying FXY has proven successful and I believe we have the makings of a robust trading tool that can be applied to many other ETFs and stocks.
And, now that we have a basic template, we can track the signal line from this point forward.
Up until now I've manually executed the signal line, but by the end of the week we hope to have a working template in place that will automatically compute the true signal line for each trade based on some scalable loss stops and trailing stops and well as the critical P&L/R2 status line.

Tuesday, October 27, 2009

Pairs Basket Weaving

Sorry to display this spreadsheet in such small format but this seemed like the best way to show the correlation between the trade trigger dates generated by the z-score band reversals.
Double-click to zoom up and you'll see the trade dates for each of the seven FXY pairs.
Obviously, the UUP is the real chatterbox here, although we can easily adjust (reduce) the trade frequency by ratcheting up on the width of the Z-score band.
The whole argument behind using a basket of ETF/stock pairs to trade a single ETF (in this case FXY), is that the more high correlation pairs fire a confirming signal, the lower the risk of a losing trade. There's been some confusion in recent emails between the terms "correlation" and "convergent". These are 2 different concepts. ETFs/stocks can be convergence or divergent . . they typically can move together or move contrary to one another. . these are directional vectors. Correlation is a measure of how closely the ETFs/stocks move in relationship to one another . . this is a scalar.
We've clearly got some strong trade confirmation signals here and tomorrow I'll line up the PDQ signals with the FXY price chart.

Monday, October 26, 2009

Monday VIXology

Today I'm using a one-two punch for the VIX forecast.
The upper charts reflect the trend per the LR30 channel in conjunction with the Qs trend.
Technicals are now looking bullish for the VIX while the Qs ride the LR30 channel mean in the face of deteriorating technicals.
But what's at least as interesting is the behavior of the VIX PDQ Dashboard.
Using some of the pairs that were profiled originally in the VIX basket post, we can detect a pretty strong consensus from the Dashboard that VIX Long positions have not fared very well, with 4 of 4 Longs showing negative Momo . . . and these are not little numbers.
We're still fiddling with a more definitive Dashboard format, including a variety of programmable stops and risk management controls but, per our earlier posts, the basic model would have exited 3 of these positions once Momo reached 1.5 % while XLE would never have triggered a Long due to the BF status.
Notice that current z-scores of the 4 (hypothetically) long positions are all only 50% of the way to the band reversal values, suggesting more of the same to come.
Also keep in mind the previous caution that the VIX is a statistical value, while the pair ETFs reflect actual prices so you can't actually make these trades other than as options and the VIX values shown are not option prices, so the plot gets even thicker.
Our premise in using the VIX basket is not to execute these trades . . you can't . . but to use the power of the VIX/pair Z-scores to forecast market momentum.
At the present time the VIX Dashboard is showing exceptional alignment.

Friday, October 23, 2009

PD FXY Update

I'm back . . with a corrected N day algorithm thanks to Jeff's prompt turnaround. The current signal is actually not LONG but net OUT since the Firing Line filter (still on progress) would override the FXA and FXF long signals short term and the FXE long signal longer term.
Tomorrow we'll line up the momentum swings in the FXY relative to the trade reports of both the short term and long term models. I've been holding off this task until the Dashboard was in a more stable condition but, with the latest changes to the database, I believe we're there.
Stay tuned. . .this could be interesting.

Thursday, October 22, 2009

PD FXY Dashboard

Here's another example of how using the PDQ Dashboard can provide forecasts for a variety of ETFs. Keep in mind that the PDQ is really a condensation of the my PairDicator Qs model (hence PDQ), so for today's post we're looking at the PD FXY Dashboard.
Using pairs for currency trading is nothing new. . there's always the Forex for those who care to venture there . . I've been there, done that and found too many broker shenanigans to keep my interest.
This is a different approach although I'd be interested to hear from active FOREX traders on how these signals align with their technicals.
As with the latest iteration of the PDQ, the optimized N (lookback days) values are presented in ascending order. . . in this case providing both a short and longer term forecast for FXY momentum.
And, just a clarification. . .although the Position may be listed as OUT there are still # Days listed. . . these are the number of days out of the trade . . so every Trade position is regarded as active.
I'm showing 2 Dashboards here. .the upper is as of Wednesday's close, the lower is as of Wednesday's open, so we can see the developing momentum in the MOMO values, the current z-score (showing proximity to z-score band reversals and the current FXY beta.
There is a little glitch with the results as seen with the case of FXE. The N values and # of trades don't reconcile and, as usual, this reflects the work in progress nature of this project. Actually, both results are valid, the problem is locking the optimization algorithm in either a long or short term mode until that model executes a P&L/R2 failure.
And you thought this trading thing was easy?

Wednesday, October 21, 2009

PDQ Dashboard Update

I've reorganized the sequence of Qs pairs to reflect the ascending order of the number of N days (optimized lookback period). This should yield a short term and longer term probability model when I can locate a few more reliable Qs pair candidates with 12-18 N day characteristics.
With the discounting of the BD type signals (P&L below R2), we're left with a residual of 3 signals, 2 longs and a short. Since the short is showing negative MOMO, the inference is that the lon side is the valid signal.
I'm still working with Jeff to implement the "Firing Line" filter for the trade signals that would cancel trades if the P&L/R2 status is below and descending (basically, a roll over of the equity curve). That filter should help resolve some of the current ambiguities in signal alignment.
There's still a few bugs with the PDQ Dashboard but these are being resolved and the forecasting model continues to be refined and made more robust at the same time.
As I've mentioned before, there are some very exciting possibilities presented by the PDQ Dashboard concept, both for daytraders and system traders, ETF and option traders.
I'll be exploring a few of these further in the weeks to come.

Tuesday, October 20, 2009

Pair Scans on ETNs

Bill Luby had an interesting post about often neglected ETNs (exchange traded notes) yesterday so I thought it might be interesting to run a pairs linearity correlation with the SPY against the ETNs just to see if anything popped out.
To back up a little bit, Bill ran a relative performance scan of the ETNs and found the top performer to be CRO, followed by NFO so we have a little benchmark to work with.
Above is the result of my pair scan using only SPY as the overriding index. I could have also plugged in DIA, VTI, QQQQ, etc, but just to keep it simple I stuck with SPY.
Mirroring Bill's results, CRO turns out to be the star of this little basket, but in this case we actively trade it against the SPY.
Keep in mind these are the results for the past 6 months only and not all trades are shown in the Trade Report. . . .but that is one nice looking equity curve!. Only 2 losing trades out of 42 -- .36 and .72 %. Great drawdown control.
But, before you jump hook, line and sinker into this trade there's a leeeetle problem: volume. Just as JJT turned out to be kind of a fantasy trade, I'm afraid CRO falls into the same black hole of illiquidity. In fact, of this dirty half dozen ETNs only NFO (IMHO) has sufficient volume to warrant trading but, another problem, it has a HTB (hard to borrow) status with several brokers, including Schwab.
Nevertheless, CRO is something to keep an eye on if volume picks up to a tradeable level.

Monday, October 19, 2009

Monday VIXology

These are the relative positions of the Qs and the VIX as of Friday's close. The VIX has now closed below the LR30 lower channel band, hypothetically putting it in oversold territory. At the same time, the Qs are sittng dead on the lower LR30 channel band.
A couple major differences here -- the VIX LR30 has been downslope for the past 2 months while the Qs LR30 channel has been upslope. The other technicals of note are the VIX's RSI, MAs and MACD, which all look poised to rebound, while the Qs similar technical all look poised to collapse. Net signal is bullish for the VIX and bearish for the markets.
Below is the new look for the PDQ Dashboard thanks to Jeff's handiwork.
Of particular note is that the PDQ was flashing SELL signals for the Qs on Wednesday's open and the signal got progressively more bearish going into Thursday and Friday . . so I'm encouraged about this little tool's forecasting value.
We'll be adding one more metric to the "Qs Position" section and that's what I call the "Firing Line" which will take into account whether the current equity curve of each pair is above or below the R2 slope and whether the current open trade position is showing positive or negative momentum.
Basically this is a stop loss and/or trailing stop to keep us out of entering trades whose equity curves have begun to collapse and at the same time help us retain any positive gains that the open trade may have captured.
As such, a new signal will need to be added to our repertoire of long, short, or out an that will be "close open trade" or COT. Those of you who have followed the evolution of the PDQ Qs pairs forecaster should already be familiar with these trade conditions but, in an attempt to avoid any ambiguity in the signals, this seems like a wise addition.
This PDQ reflects values and signals as of Friday's close . . still short the Qs although XME is giving us a contrary signal on a new entry.

Friday, October 16, 2009

PDQ . . 2 Models

Here are a couple views of the PDQ Dashboard as of Wednesday's close.
The model I use most frequently simply involves reversals off the Z-score bands and generates the signals noted above.
BTW, I've simplified the position of the P&L relative to R2 into 3 categories in order to minimize confusion.
Per yesterday's comments, the signals are continuing to turn negative as the open Q longs all show negative momo while the net short positions of the Qs both show positive momo.
The fact that the Qs/UUP pair has an ascending P&L in the face of a short signal is not inconsistent as our goal is simply to keep adding to our net gain however we can.
The Qs/VIX pair continues to provide a contrarian signal and, as mentioned previously, should not be considered tradeable but simply used as a momentum indicator.

Now this is another version of the same pairs, but with the Switch Trade option turned on. Adding this option increases the frequency of trades since the trigger is pulled at the first band kiss rather than at the reversal. The result is a bump in net gains at the expense of a bit more risk exposure

The Z&R model does demonstrate more volatility than the Z only model and shows some disparity among the signals. For the near term I'm sticking with the Z only model for upcoming basket testting.
What's not clear is whether this disparity of performance is unique to the Qs basket of whether this is more universal. Additional pair basket tests should help resolve this question.

Thursday, October 15, 2009


Above is a 15 minute bar chart of he NYAD. Long term readers know the NYAD's my hands down, #1 marker for determining short term market direction and momentum. Without it, the success rate of my trades would suffer considerably.
Now every chart is subject to personal interpretation and one man's bull is another man's bear but what my feeble brain derives from this NYAD chart is basically a failure of momentum carry through over the past 5 days.
Although we've gotten some nice pops at the open, these moves have been faded more often than not into the close.
This type of follow through failure leads me to be suspicious of longer term strength in the markets. Yes, the headlines each day remind us that the recession is over and that the good times are once again just around the corner (without any jobs, unfortunately).
But a bit a caution is warranted here.
Over at the Quant Wizards site I've been posting bullish forecasts for the past 2 weeks based solely on the markets' proclivity towards whole number reversals. . in this case Dow 10K and SPY 1100. With Wednesday's action we are there and the odds for retracement increase exponentially. Thursdays are statistically the highest probability pivot high cusp days so we may get some up side spill over before the pullback, but I'm personally going to 90% cash for a bit.

In further support of my cautious to bearish scenario . . the PDQ dashboard as of Wednesday is showing negative momentum in each of the 4 open Qs long positions. This is unprecedented in the 15 market days I've been monitoring the Qs basket positions and the net conclusion is once again failure of the bullish algorithm as maximum 6 month historical intraday drawdowns approach.

Wednesday, October 14, 2009

PDQ Dashboard

Continuing my work on the PairDicator theme, this is the PDQ Dashboard reflecting most of the database metrics I need in order to apply some more sophisticated system trading algorithms and the resultant signals.
Over the past 2 months I've put together a number of pair baskets that I intent to track and trade on a systematic basic. Although my original focus was on ETFs, I've found that adding a few correlated stocks into the basket mix can definitely pump the net return numbers.
My current stable of ETF pair baskets includes Qs, UUP, KRE, RTH and XHB.
The stock based pair baskets include GE, NEM, EXPD, COST and LUV.
Sticking with the PDQ template as shown above, each basket contains 9 pairs.
As with most of my projects, this is a work in progress and has consumed a few hundred hours of fiddling around with these data fields that I hadn't really anticipated.
Nevertheless, I'm encouraged by the current product and optimistic that ongoing refinements of the PDQ will provide a robust and adaptable predictor of market momentum.
Watching the various pairs Z-score reversals intraday for the past 2 months has given me a real appreciation for the forecasting possibilities of this rather simple algorithm. At the same time I've come to understand why several large prop shops focus enormous capital on pair trading strategies as pair trading can significantly handicap the risk otherwise assumed by taking a directional position. Pair trading is not about trading market direction, but rather about trading the relative strength and momentum of various sectors and themes. Clearly a different approach to gaming the markets and one that, as shown briefly already, can provide an effective risk management tool.
An additional benefit of using the PDQ basket format is its scalability both in terms of capital deployment and signal confirmation (add-to positions).
Just a caveat . . the data shown above is from 10/9/09 and does not reflect current PDQ signals.

Tuesday, October 13, 2009

PDQ . . .Qs PairDicator

This is the initial wrap-up of the Qs pair basket study. After testing over 100 ETFs and a few stocks against the Qs, I've settled on these pairs for the time being. As the markets develop over time I fully expect the basket composition to change both focus and momentum.
I've color coded the various sectors reflected by the ETFs/stocks and you'll notice a current emphasis on materials. I'm not going to venture a reason for this high correlation, I'm just going with the flow and use the correlation to provide a trading edge.
Trying to put this information into a systematic tradeable format has presented some challenges and (as a first draft) my first signal template is presented below . . the PDQ Qs PairDicator.
Now this could be the NEM PairDicator, the FAS PairDicator, etc., depending on the particular focus of the matrix. Although there are lot's of interesting pairs that may catch your attention as you glance at the Qs matrix, for purposes of my study I really only care about how the Qs interact with the matrix components.

I've reorganized the REWIND data fields and presented them in a (hopefully) easy to visualize format. While this format is not currently available on the REWIND site, Jeff may eventually take pity on me and offer the above layout as part of the expanded Trade Report.
I''ve added several metrics which seemed useful to short term traders like myself:
. . First, the number of days we are into open trades.
. . Second, MOMO, which describes what's happening in real time to open positions, either positive or negative.
. . Third, the position of the P&L equity curve relative to the R2 slope (which I have have discussed extensively in previous posts).
From this point it's just a matter of developing consensus among the long/short signals and the positive/negative momentum in open position.
I'm reserving discussion of those fine points until later this week, when I intent to produce a streamlined version of the PairDicator.
Until then . . here's a site that I really love featuring some out of this world photos. If you click on the "Discover the Cosmos" link at the top of the page you'll find a gallery of several thousand pictures that are truly unique. I've posted my personal favorite on the lower right panel of the blog. . M17. Looks like it's been painted by Michaelangelo, but this is natural phenomenon.

Monday, October 12, 2009

Monday VIXology - A Different Spin

Continuing with last week's introduction of the VIX Pairs Advantage I've run the VIX against a basket of index and Spyder ETFs and UUP . . twice.
In the upper matrix I've turned on the pairs correlation. . . in the lower matrix I've used the Ticker A only setting to evaluate the z-score reversals.
In both cases I've set the linearity filter to 90% and the P&L filter to 1.5.
A quick scan of the 2 matrices will should convince you that using the pairs approach is a superior tactic for both managing risk and optimizing net returns.

Thanks to readers who participated in my little poll. Although only about 20% of my daily readers voted I appreciate your feedback. It helps me understand your interests and focus. As I expected, the major areas of interest were daytrading and system trading. I'll endeavor to continue my focus on those themes but keep in mind that I have specific goals for my own trading, which is really what it's all about.
#1 . . Job One is always capital preservation.
#2 . . I'd rather trade a system with a 95% chance of making 25% than a system with a 75% chance of making 50%.
#3 . . I hate drawdowns, both intraday and daily and craft my systems and trading tactics to minimize both the frequency and magnitude of negative bars.
#4 . . I trade what works. . . and this means adapting to market conditions as they evolve. I'm always looking for an edge or a niche because as a small retail trader I realize that the deck is pretty much stacked against me by the funds, commercials and big prop shops.
#5 . . I really trade. This is how I support my family and I take it seriously. I've known too many traders whom I respected that have blown up their accounts and are no longer in the game. My goal is not to suffer that fate, but to survive and to thrive.
I don't sell books, tapes, self-help videos and such that make outlandish claims based on hypothetical trades of thinly traded stocks that I somehow managed to buy at the day's low bid and sell at the day's high ask. I'm strictly into exploring what's practical and realistic for the average retail trader who's willing to work at it.
That being said . . I've got to comment that my 2 month pairs study and 6 week posting has convinced me that this is an excellent low risk strategy for trading daily bars. I've yet to explore the possibilities of daytrading pairs on 5, 10 and/or 60 minute bars. I have buddies who do daytrade pairs . . very happily I might add, and I'm looking forward to testing some of my pair basket on shorter time frames to reflect the majority of my readers focus on both daytrading and system trading.
And, while there are other pair trading algorithms on the market, my view is that this is the best value by far. You've got to do a little work to find pairs that fit your style and risk tolerance, but nobody ever said trading was easy. If nothing else, pair trading can provide a relatively low exposure consistent revenue stream to supplement your other trading.

Friday, October 09, 2009

Qs Pair Basket - Part 4

Continuing the Qs pair basket study, I'm looking at 2 different themes today.
With the DIG and UNG Qs pairs I wanted to highlight the correlation between the differential beta and the number of trades for these 2 pairs.
Note that the lookback (N) periods are not dramatically different, as might be expected.
Also of interest, the equity curve position relative to the R2 slope for both charts is supportive of maintaining these pairs in our Qs basket.

Here's a little different spin . . using a sector ETF . .IYT (transports) versus a sector component stock EXPD to enhance the pairs net return. With the EXPD showing twice the beta of the IYT relative to the Qs, we would expect a bit more volatility and we're not disappointed.
EXPD trades on a 3 day lookback whereas IYT trades on a 5 day cycle. And the total number of trades is 40 with EXPD as opposed to 10 for IYT. Net return from the EXPD is about 50% higher than the IYT.
The caution with these 2 pairs is that the equity curve has fallen below the R2 slope producing negative returns on the last 2 IYT trades and from a risk management perspective we should postpone entering any new trades until the hypothetical equity curve once again turns positive.
And, if you haven't voted in the poll in the upper right corner of the blog, please do so.

Thursday, October 08, 2009

Qs Pair Basket - Part 3

Continuing my Qs pair basket study, I've shown 4 Qs pairs from the Qs matrix related to the materials/metals. The optimized lookback (N) periods vary from 4 -8 days and the resultant trades number 6-19. Perhaps surprisingly, there's not a direct correlation between the value of the (N) and the number of trades. The variation in differential beta between the pairs may account for this disparity and, as mentioned in several previous posts, paying attention to these beta values is an essential component of managing risk on these trades.
My argument in developing the Qs pair basket is that by maintaining a number of correlated pairs focused on the Qs when the trigger signals align, the probability of successful trades increases correspondingly.
The other pair metric that requires confirmation is the alignment of the P&L equity curve relative to the R2 slope, and in the 4 charts posted here we have a close correlation. It's important to also track the execution dates of the trades to see if the trade is open or closed as that will determine precisely where the equity curve slope is riding relative to the optimized R2 line.

If you haven't already voted in the poll on the upper right corner of the blog, please do so.

Wednesday, October 07, 2009

NYAD Reverse

Just a quick note on yesterday's NYAD dynamics. Shown on 10 minute bars, the NYAD momentum over the past 2 days is clearly evident. Where it gets interesting is just before noon when all our technical align to the downside on both the Qs price chart and the NYAD. With a SELL confirmation of the parabolics on the NYAD, this is a high probability short. Also of interest, and reflective the relative over bought levels, R1 on the NYAD yesterday was at 7, a number normally associated with an R4-R6 reading.

Finally, a little digression today before resuming my extensive study of Qs pairing opportunities. I've mentioned this free e-magazine before and, despite the name, it's not only for futures and options traders, but stock traders as well who may creatively apply some of the featured concepts, indicators and systems to the equities markets.
Download the current edition and/or sign up for a free subscription.
The article that caught my attention this month was the double butterfly setup (pages 20-21) that has shown itself to be consistently profitable for the last 5 years (the period backtested).
This particular setup uses calls (net credit spread) whereas I have previously noted double butterfly setups using puts (net debit spread), which is just my preference.
Contrary to normal expectations to let the position ride into expiration, the F&O model exits after only 11-14 days. Yet, this simple approach and aggressive risk management program yields an 81% win/loss ratio and a 154% return.
Worth a look IMHO and did I mention . . . it's free.
To help me better serve my readers please take a minute and check your primary trading interest in the poll at the top right corner of the blog. The poll will be open for a week.

Tuesday, October 06, 2009

VIX / VXX Pairs Advantage

This is a continuation of Monday's VIXology showing some of the nuances of VIX versus VXX when run against the same pairs basket. As mentioned yesterday, Bill Luby explains the differences between the VIX and VXX, but the important thing to keep in mind in that VIX is a statistical value, not a price . . . whereas VXX reflects the price of the VXX exchange traded notes of the VIX futures.
VIX can be traded using options, VXX has no options. VXX trades about 1.5M shares daily with a nickel spread . . VIX options reflect high open interest and robust daily volume.
This is just a teaser to illustrate the differential between the VIX and VXX against a fixed pair basket. The VIX pairs display a positive edge, both in correlation linearity and P&L, but keep in mind you can't actual trade the VIX at the indicated values.
Some of my more resourceful option savvy readers may find merit in these VIX pairs using ATM options, especially on 4-7 day cycle trades when premium decay may not negate potential capital gains.
For my own purposes my intent here is to use the VIX long/short signals as confirmation for otherwise indicated momentum moves.
More on this theme next week . . . . .

Monday, October 05, 2009

Monday VIXology

Some dramatic moves in the VXI and Qs charts this week as the markets suffered a technical retreat.
The Qs hit our first level of support at 41 on Thursday and continued down a bit more into the Friday close. Friday's close also puts the Qs eerily in a position identical one month to the day when the LR30 lower channel band was touched and then violated. . .to be followed by a 3 week rally to new highs.
With the Qs RSI2 at the zero level there's a solid argument for suggesting that current oversold level will lift the Qs back to at least 42.50 next week.
The VIX also moved up to 29 levels not seen since early September and aggressively violated the upper LR30 channel band. As with the Qs deja vu' pattern from September 1&2, The VIX
is displaying a similar pattern.
The $64 question now . . .mean reversion or channel kiss-off?
One of the tools I'm working on involves using the VIX to gauge the probability of various system trade setups and I'll give you a little preview of how it looks.
Using the pairs basket concept, I've set VIX as my target pair core and run a variety of correlated and non-correlated ETFs against it.
Keep in mind that you can't actual trade the VIX except as options and as the ETN VXX, but my intent is to not really to trade that side of the pair anyhow, so no handicap there. Instead, we use the momenum signal of the VIX in much the same fashion as the VIX Advantage system to keep us on the right side of the trade and to mimimize draddown.
Also see Bill Luby's relational study of the VIX and VXX.
This is just a preliminary VIX study but with huge implications and I'll be reporting on various refinements as they develop.

Friday, October 02, 2009

Qs Pairs Basket - Part 2

Despite the JJT and PTM diversions this week, I'm still working on the Qs pairs basket and have expanded the preliminary Qs matrix as shown above. I had fully intended to conclude the Qs pairs study this week, but sometimes you've just got to go where the research takes you if you want to uncover the nuggets.
Next week I'll run the complete Qs pairs basket analyses and define several metrics for determining what's hot and what's not on the basis of a daily update.
Going forward into November I've got two new projects that have consumed much of my time time recently and will no doubt consume much more going forward:
# 1. A series of studies I'm calling GEconomics, including systems and pair trading approaches.
Since I actually trade GE more frequently than the Qs and hold bigger positions, it seems wise to devote some investigative efforts to assure maximum risk management controls are in place. Being a stock, GE is a little more problematic to trade than the average index ETF, but initial forays have been encouraging. Because of it's consistently high daily volume (almost always in the NYSE top 10) and it's relatively high daily ATR , GE offers some unique trading opportunities and I'll endeavor to profile a few of them.
# 2. Another series of studies involving VIX and VXX pairs. The preliminary research on this front is stunning to say the least and really devolves into a variation of my VIX Advantage studies a while back. The difference here is that by creating a VIX/xxx pair and using the reversals of the VIX z-scores to trigger trades we goose our single sided Ticker A performance considerably.
For those that haven't been following my pairs-related posts for the past month this may sound like a lot of mumbo jumbo. Sorry, but that's where I'm headed.
Reread earlier September pairs posts to catch up and/or get a 3 day trial subscription to the Rewind matrix and pairs trader with no obligation to try out the possibilities for yourself.

Thursday, October 01, 2009

All That Glitters. . . . .

Continuing with our failed JTT pairs model, the original ETF commodity pairs scan suggests that E-Tracs Platinum PTM might, in fact, offer some better trading opportunities.
The PTM pairs basket shown is just to illustrate the range of ETFs and stocks that work well with PTM. This type of robustness is a necessary component of most longer term pairs.
In addition, this particular basket provides a nice spread of lookback (N) days and in the 3 selected pairs analyses shown below both the max drawdown and percent winners are more than encouraging.
There is a downside here and it's the same one that doomed JJT . . . the lack of volume.
Bid/ask spreads are also normally in the .06-.08 range and alas, there are no options.
That being said, PTM does typically trade 100K shares daily and while the big prop shops aren't likely to pile on this one, a small retail trader can probably pull a few bucks out of these low risk pairs without a lot of exposure.