Well why has currency trading been stressful while investing for retirement via an IRA seems relatively easier? I think it boils down to the idea that the strategy employed in the IRA (buy and hold) would ultimately work if given enough time. That could never be said about a short term system that could fall apart at any given time…

And so, as I was drafting some guidelines for the next iteration of my active trading, I read the Ivy Portfolio and noticed the launch of the GTAA ETF….

For the record here are those guidelines:

No leverage
No intraday trading
No drama (Usually related to gun-slinging “discretionary” trades with or without high leverage)
No monthly fees
No proprietary platforms

I have decided to use GTAA as a benchmark for my own Tactical Asset Allocation model. For now, I only have an absolute returns version of my model, and GTAA is an absolute returns fund, so the benchmark is appropriate. However, I also have my eye on the model over at MarketSci, which is likely to be geekier and more elegant than my absolute return model, but nevertheless, for now, I will be using that model as a secondary benchmark. The positions for my model for November 2010 are :

VEU / VWO

The model (if it can even be called that at this point) always purchases equal parts of the top two assets, so clearly there is some room for improvement.

To reiterate, this is never investment advice.

More to come…

“…

He was in the process of boiling wheat in 1894 in an attempt to create an easily digestible bread substitute when he came across a discovery that would lead to Kellogg’s Corn Flakes.

He had boiled some wheat with the intention of making dough with it and accidentally let it stand for several hours. The wheat became softened, tempered. He decided to put it through the regular rolling process anyway for baking. When he rolled it out, however, he noticed that the individual wheat berries in the mash would roll out into flat, wide flakes. He figured he’d bake them and see what happened. The result was a crisp, tasty, easy-to-eat cereal product.

”

I always like to think that if I tinker around long enough with trading systems, I”ll make a mistake that will prove to be a gold mine.

 Cable Glider is the result of adapting the Turtle System for an intraday market.  I accidentally discovered that by shifting the price channel and anticipating breakouts, the returns were massively improved.  I guess that this wasn’t an accident, but it was the result of testing some “out of bounds” parameters.

Euro Ranger has a MACD filter in it that, to this day, I have no blessed idea how I came up with it.  These types of things are usually the result of trying to fix code that doesn’t work the way that I envisioned it…meaning that I’m kind of lazy like W.K. Kellogg.

Darth Fader works very well in the natural gas market.  It’s only after I tested it there that I realized I was testing a parameter set that was taking the first available profit, and not holding for a longer period as I had originally intended.  It turns out that both methods work, but the accidental one works better.

Now surely trading systems are not rocket science, and they are nowhere near as important as antibiotics or Coca – Cola, but it sure is nice to think that if you just keep tinkering, you might one day be the one shouting “Eureka!”.

So I set out to create a simple system with a trend component and a trigger.  That system, which is so simple, it’s almost shameful, seemed to return positive results for a variety of trading instruments, including individual stocks.  So I purchased more data…NYMEX, COMEX, CBOT, and S&P Indices.  The system seemed to work well in most liquid markets even without optimization. 

The problem is that in some of the markets, almost all of the gains had been made in 2008, which was a banner year for trend followers.  After a conversation with a friend, I realized that I should be testing the long run trend follower alongside a short term “dip buyer” type of system.  So I coded a simple system up. 

After I put 2 and 2 together, I realized that the short term “dip buyer” was really trading from the same principles as my production system for the S&P, Darth Fader.  So then I started testing DF on my new data feeds.  It turns out that DF performs well for stock indices, energy futures, treasury notes, currencies, metals, etc…

Next I took a look at my actual trading results for the week.  3 wins, 0 losses trading Cable Glider and Euro Ranger, which are both intraday systems.  I’m up 12.4% year to date by the way…however, Cable Glider and Euro Ranger do not perform as well as optimized versions of Darth Fader work on the Euro and the Pound.

So I’m at a decision point.  On one hand, it is nice to trade systems with different methodologies to guard against a wholesale failure of Darth Fader’s approach.  On the other hand, I really would love to trade systems on the daily charts only.  This way, I wouldn’t have to worry about keeping my computers on all night long and possibly getting booted off the network at 3AM when CG and ER want to enter trades.

So I have a little bit of work ahead of me to determine the best course going forward.  Without the benefit of hard data just yet, I’m leaning toward trading Darth Fader on the S&P (or the Nikkei–to reduce correlation with my 401k), the Euro, and on Natural Gas Futures.  I’ll need to compare the risk profile of that approach to the current Troika of DF on S&P, Euro Ranger, and Cable Glider.  I also need to remember that CG and ER have proven themselves to be profitable in real trading, while DF has not…I mean, it has shown me a profit, but I’ve only been trading with it since the beginning of 2009.

These results are on the backtest page.  This is the power of trading uncorrelated systems.  Higher profit, lower variance.

My plans had no provision for System Death.  I plan on becoming more studied on this subject, and I will incorporate it into future system deployments.    There is some good information over at Vertical Solutions, which I hope will point me in the right direction.

In the meantime, I will keep Euro Ranger active.  It was virtually break-even last month, but it has shown several consecutive months of profitability. 

The linear regression system that I was testing on the Japanese Yen totally fell apart in out of sample testing.  It’s a little bit disappointing, but then again, that’s why we test strategies before deploying them.

I’ll also be tinkering with tick charts a bit…this is something that I’ve never considered before, but I believe that it makes good sense to use tick charts instead of time based bar charts in a 24 hour market where most of the ticks are clustered into a small time frame.  For example, if we use a 500 tick bar on the Euro FX contract, we’ll see that sometimes a single bar can span a few hours (usually right after the session opens), and as little as 1 minute in a fast market.  Very interesting indeed…and that’s what I need right now…new ways of looking at the data.  We’ll see what it yields.

Mechanical approaches are ideal for beginners, as they force us to trade what we see as we develop a feel for the market.  In my former life as a stock and option trader, I would bounce around from one idea to the next, blowing up my account several times in the process.  The problem with such a random approach was that sometimes luck did shine on me and save me from losing…and of course these random rewards are the most addictive and difficult to reproduce.

When I shifted focus to the currency markets and became aware of all of the great tools for developing mechanical trading systems, I firmly made the mental shift that a mechanical approach would be best for me…after all, my discretionary trading had led me to a track record of 8 consecutive losing years.  You know what they say…quitters never win and winners never quit, but those who never win and never quit are idiots!

Now that I’ve focused almost exclusively on the British pound for a few years, I believe that I’m starting to develop a reliable feel for the market.  I am certainly NOT ready to incorporate much discretion into my trading plan, because the mechanical approach has worked so well, but I’m becoming more interested in how an experienced trader moves from a mechanical approach to a discretionary one and consistently outperforms the mechanical approach.

Of course, when we talk about intuition, we must carefully define the term.  The first definition on dictionary.com comes closest to what we are referring, but it is still incomplete:  “direct perception of truth, fact, etc., independent of any reasoning process; immediate apprehension.”  Well in this case, we’re not really talking about understanding that is independent of reason, but rather immediate understanding derived from unconscious integration of our experience trading the market.  It’s quite a fascinating process, but one that I’ll admit that I’m at a loss to describe the mechanics of.

All that I can do is make a note of just how long it has taken me to begin to develop a market intution and remind beginners that the goal of consistent profitability is a mountain with an unseen (and uncertain!) peak.  If your trading plan keeps you in the game long enough, your “into-wishing” just might turn into intuition after all.

The second category of trading system that I think in terms of is the Home Run Hitter.  This type of system derives its edge from achieving greater average wins than average losses, but not necessarily from a high winning percentage.  The Cable Glider falls squarely into this category.  I’m much more comfortable trading this type of system, but no matter how well the Cable Glider works, I continue to feel vulnerable without a stable portfolio of other systems to back it up.  I’m sure I’ve mentioned this before, but the fact that the Cable Glider has worked so well makes my other attempts at system development feel like failures, even if these other systems are marginally profitable.

At any rate, my first goal for 2008 is to create a Grinder style system that will work for the British pound.  So far, even in the context of trying to create a grinder system, I’ve found it very helpful to stay out of the markets on days of significant economic releases.  It seems to me that on these days, the market ignores technical setups to a large extent as it schizophrenically adjusts to the new data.  Right now, I have a one parameter system on the 4 hour chart for the British pound that tests as well as the Euro system that I currently have in production.  The rules are so simple it’s almost stupid, but again, the system would not work well at all without the economic release date filters to keep it out of choppy waters.

So if I could offer anything as a word of advice to aspiring traders, it would be to examine how economic releases such as the employment report, the CPI, central bank interest rate decisions, etc, affect your trading strategies.  I continue to find a dual benefit to staying out of markets on those days…a chance to preserve my capital, and to recharge my mental batteries to trade clearly….LATER…in calmer conditions.

 Good luck to all in 2008.

Pain of loss

a) True Drawdown that SEEMS avoidable (1)

Whenever I’m in a drawdown where entry signals were BARELY triggered and then the trades reverse and take me out for a loss, this is the most painful, because it seems that optimizing my entry parameters differently could have saved me this pain.  This is the most painful, because a drawdown always feels like a threat to my ability to continue as a full time trader, and losses that I ALMOST avoided make me feel unlucky too.  The longer or deeper the drawdown, the worse the pain gets.

Antidote: I need to remind myself that my system at any given time represents my best attempt to make profit in the market and that drawdowns are unavoidable.  Indeed, I’ve recovered from several drawdowns, and even if this particular drawdown really does end my run as a professional trader, I will never regret my decision to try.  Anyway, I’m likely to recover from the drawdown, and losses are just what I need in order to be able to build better systems for the future.

b) True Drawdown that seems unavoidable (3)

If I suffer a drawdown that seems unavoidable, that is to say that most parameter sets would have still resulted in a drawdown, I don’t feel quite as bad, but I’m still in a real drawdown, which is a threat to my full time trading career.

Antidote: Same as (1)

c) Giving back profits on subsequent trades (6)

Giving back profits on subsequent trades is the least painful, because I know that losses are inevitable and it’s nice to initiate a drawdown from a new equity peak.  No antidote is necessary.

Pain of not participating (2)

This is the second most painful of all.  I trade a system that uses event-based filters to keep me out of the market.  When things are going REALLY well, these filters are counter-productive.  Missing a move is painful because I wonder whether I will experience the double whammy of missing a big profit and then participating in the subsequent losses.

Antidote:  Most of the time, my event filters work well.  They often allow me to have profitable months when a system without filters would have suffered losses.  Event filters allow me to take a break from the market and recharge my batteries.  A system that is in the market less of the time is inherently less risky.  Whenever the unfiltered system does well, the filtered system usually does well too.

Pain of Profit

a) Why didn’t I trade more contracts? (5)

This one always creeps into my mind a little bit at the end of each winning trade, but it’s easy to suppress as I know that if I were taking more contracts, my risk to reward would not be properly balanced.

b) Why didn’t I take profits at the top? (4)

This one is painful too, but it’s not a big deal.  My system does not attempt to catch tops, and will always give back some profit before exiting.

Here is the king of Thought Demons for a trading system developer…

The new system underperforms the old system in live trading.

The antidote for this is to realize that the system IS my edge and that system changes represent my best attempt to improve my edge.  Each change is either an attempt to increase profits or to limit risk.  A change that is intended to increase profit often has the side effect of increasing risk or drawdown in the short term.  If the new system passed all of my criteria to be traded live, then I simply have to go with it until such time that I discover a better alternative.  If the system change is meant to decrease risk, but it lowers my profit, I need to remind myself that that is exactly the price I pay for controlling risk, and in fact, risk has been controlled.  In fact, if I look at my list above, I’ll see that the pain of missed profits is less than the pain of real losses.

Very near the bottom of the equity drawdown, I decided to segregate some of my trading money into a living expense account so that I would feel less pressure from the day to day moves in the market.  I looked at the distribution of gains and losses for the prior year, and established a knee-jerk rule to add money to the living expense account whenever I experienced a “large” gain. 

In retrospect, this strategy amounts to a dynamic cash hedge, meaning that my total risk level fluctuates with the percentage of the total account that is held out in the living expense, or “hedge” account.  In truth, establishing the hedge at the equity trough was exactly the wrong thing to do, and it delayed my recovery from the drawdown.

From June to September, I blissfully ignored my hedge % as the Cable Glider posted gains in each of those months.  A losing October, however, has caused me to re-evaluate how I hedge my risk, and it’s led me to an interesting discovery that will affect how I handle money management with my systems going forward.

I have found that by picking a fixed hedge percent, say 25%, and resetting it to 25% at the end of each subsequent month, I can achieve better returns than by simply reducing my risk on each trade by 25%. 

Essentially, I’ve discovered the position sizing method that chaffcombe (http://www.futurestech.com.au/MonthlyReport_Sep2007.htm) refers to as Variable Fractional Position Sizing. (VFP).

Rather than re-invent the wheel, I’ll quote his comments on the subject below:

”

Although the drawdown was lengthy, it was extremely shallow – something that I am particularly pleased about considering the earlier major gains.   I largely attribute this to tight risk management and, specifically,  the variable fractional position sizing methodology (VFP) that I employ.  VFP increases or decreases the risk taken for each system in proportion to the realtime P/L  for the calendar month, thus allowing increased leverage during profitably months, without taking undue risks with what I consider to be ‘capital’ ie NAV booked at the end of the previous month.   VFP can make a lot of money very quickly in good trading conditions, while limiting losses very quickly in bad…

”

Armed with VFP, we can re-think the subject of drawdowns.  Traders will first and foremost consider the maximum peak to valley drawdown in their trading, but may not pay any attention to the timespan that those drawdowns cover.  Indeed, visualizing a drawdown as a lake whose volume is measured on the vertical axis by % drawdown and whose horizontal axis is measured by time, we can work on decreasing the pain of drawdowns by attacking both the vertical and the horizontal dimensions.

Most of the vertical value of a hedge is accomplished in the first down month.  If we reset the hedge to a fixed percent after a down month, we are effectively adding money to the trading account.  This is an anti-hedging technique meant to decrease the time taken to recover the drawdown.  With this methodology, we do risk a deeper drawdown if we experience consecutive losing months, but we retain the majority of the vertical value of the original hedge.

So, VFP has the potential to outperform fixed fractional in a drawdown.  What about in a run up to new equity peaks?  The beauty of VFP is its ability to lock in more of the gains from a positive outlier.  If we are happily moving along making 5% per month with no variation, VFP will reduce our total return by hedging out a portion of the gain each month.  If however, as has been the case with the Cable Glider, it is true that large positive and negative outliers are clustered together in time, VFP will outperform fixed fractional.  If we have a big gain followed by a big loss, VFP locks in more of the first month’s gain and reduces the subsequent drawdown. 

If we have a big loss followed by a big gain, the anti-hedging ability of VFP shines again by adding more capital to the system at the equity curve trough and speeding us out of the drawdown.

As with many other trading system parameters, the optimal (defined here as maximum net profit for any given level of risk) hedge percent is a moving target, and so this is not a simple topic.  For the Cable Glider in 2006, the optimal hedge % would have been in the 40′s, but this is because there were two cases of big run ups followed by drawdowns.  In 2007, the optimal hedge percent is much lower, because overall the equity curve has been much more flat.  When I combine the two years, 25% is the optimal hedge.  Over the past five months, my original dynamic hedge was moving in the 22% to 31% range.  Even though the length of this post and my long-winded explanation might suggest otherwise, the new method is simpler, cleaner, and easier to verify.

First of all, I do not want my fitness function to be dominated by recent results, so I do not “reinvest” my profits when testing a system;  I presume that I have a fixed starting equity for every trade and risk a percentage of that equity. 

Secondly, since the Cable Glider is quite profitable, the equity curve quickly looks like an exponential function, so I use logarithms whenever possible in order to normalize various pieces of the fitness function that would otherwise be on entirely different scales.  This is  an attempt at finding a fitness function that is not dominated by a single sub-component.

Thirdly, I think what I like and what I don’t like.  I like profit.  I dislike losses.  I like a high winning percentage.  I like the average win size to be greater than the average loss size.  I dislike drawdowns.  I want to be certain that I have a quantifiable edge that suggests my trading results are better than playing a coin flipping game.

I will post the specific function that I’m using in my next post.  In preparation, please read Ed Seykota’s page on risk management.  My fitness function incorporates the Kelly criterion described on that page.