GDX Trading Systems - 2

Following Thursday's post I managed to finish 3 more long strategies for trading the daily chart of Market Vector's Gold Miner ETF (GDX). These strategies use the Force Index (FI), Volume Zone Oscillator (VZO) and True Strength Index (TSI) indicators.

The results (below) were obtained using the Think or Swim platform and reflect the use of each individual strategy from the first day GDX traded in 2006 until today. 

These strategies have been optimized (read 'curve fitted') to the max. As you will see, the number of trades is modest, the overall win percentage (70%) is nothing to get terribly excited about, but the winners pack a heck of a punch. The average return per trade (including the losing trades, of course) is a juicy $275. This kind of return was worth an average gain of around 5% each and every time it was attempted.

GDX – Market Vectors Gold Miner ETF Individual LONG Strategy Results (2007 – 2012)
Strategy		Trades	Win/Loss	Win/Loss %	Total Gain	Avg Gain per Trade
6	Force Index	12	8W / 4L	66.6%	$4,001	$333.33
7	VZO	19	14W /  5L	73.7%	$5,606	$295.05
8	TSI	26	18W /  8L	69.2%	$6,042	$232.35
Total		57	40W / 17L	Avg 70.17%	$15,649	Avg $274.54
Strategy Indicators Used
Strategy Name
6	Force Index	Uses the Force Index and True Strength Index
7	VZO	Uses the Volume Zone Oscillator
8	TSI	Uses the True Strength Index
GDX – Market Vectors Gold Miner ETF              Cumulative LONG Strategy Results (2007 – 2012) (Strategies 6-8)
Year		Trades	Win/Loss	Win/Loss %	Total Gain	Avg Gain per Trade
2012		11	6W / 5L	54.5%	$2,563	$284.77
2011		9	4W / 5L	44.4%	$1,398	$139.80
2010		7	6W / 1L	85.7%	$2,670	$106.42
2009		11	8W / 3L	72.7%	$2,153	$121.32
2008		10	9W / 1L	90.0%	$4,269	$460.80
2007		9	7W / 2L	77.7%	$2,596	$288.44
Total		57	40W / 17L	Avg 70.17%	$15,649	Avg $274.54

My hunch is that once I code my GDX long strategies into TradeStation and start throwing grenades at them the results will slip, perhaps considerably. It is one thing to manipulate indicators and variables such that one gets the desired outcome and quite another when the defined rules confront new realities as time marches forward.

A simple example that comes to mind is the use of a target price. Suppose a strategy executes a BUY order under a specific set of conditions and is then set to SELL GDX if it reaches a target price $6.87 higher than the entry price. That worked great when using the past 6 years of data but suppose in the future the same set of BUY rules get one into a trade but the highest GDX reaches after entry is $6.86 before then turning back down. The trade that had every reason to make a killing may well become a painful dud when another SELL rule closes the trade at a much lower price. 

Fortunately, there are techniques to deal with this statistical/probability issue that I will be able to address once I code the strategies into TradeStation. 

As I understand this at the moment, I will run my strategy code against a significant portion of the 6 year GDX time frame (perhaps 2006 - 2011) and optimize those results. Then I will run that optimized code against the time frame of GDX that was not included (2012) in the first test run....and see what happens. The name they have for this is "Walk Forward Analysis".

If, by chance, the strategy flew through this 'unknown' data and did well that is encouraging. In any case, the strategy variables will be further adjusted using another significant time frame (perhaps 2008 - 2012), optimized, then run against the time frame ommitted in this second test (2006-2007). 

And on and on, and so on, etc. etc. (I guess I already mentioned how extremely time consuming this challenge to create profitable strategies can be).

Hopefully the results of the final strategy version (attained when I have have hit the limit of my patience) will generally work well whenever it is run against any time frame it has not previously seen. Wish me luck?

I share this information with you for a couple of reasons. First, I would like for the reader to be able to evaluate these strategies with an understanding of the process used in their creation. And second, I would like for the reader to understand if/when it turns out these strategies are not as profitable as they now appear, there is actually a decent reason to have more confidence in them than now, not less.

The 5 long strategies I presented in the previous post and the 3 long strategies in this post yielded a gross gain of $46,573 ($30,924 + $15,649). This was accomplished by running each strategy on the GDX daily time frame since the first day it began trading at $38.75 a share on May 23, 2006 using 100 shares for each entry and exit.

GDX closed this past week at $48.74 - showing a 25.78% 'buy and hold' gross gain for its entire history. This amounted to a gross gain of just $999 for 100 shares.

$46,573 vs. $999.

Incredible contrast. But not realistic whatsoever. 

But heck, what if we could make $30K vs. $1K, or $20K vs. $1K  or even $10K vs. $1K realistically? Wouldn't that be worth pursuing? 

I think so. 

And I'll keep you posted, OK?

John
tsiTrader@gmail.com