PMKing Trading LLC

Whenever I see the word “free” with anything to do with the financial industry alarm bells go off in my head. That’s why when I saw the other day that Interactive Brokers were offering trading of FactorShares^TM ETFs “commission free” I knew something must be going on. FactorShares ETFs are a new kind of ETF that is 2x leveraged, and represents a spread between 2 things rather than simply a collection of underlying instruments. For example, FSE is the FactorShares 2X: S&P500 Bull/TBond Bear which is long S&P E-mini futures and short US treasury Bond futures. The fund will increase in price if the S&P 500 goes up more than the the US Treasury Bond goes down, or vice versa.

Let’s put aside the fact that anyone who has a little bit of futures knowhow can easily go long S&P futures and short US treasury bond futures and avoid paying the 0.75% “management” fee the fund charges for a moment. If you don’t understand how to trade futures or what going short means then why would you contemplate owning an ETF you don’t have the first idea what it’s actually doing and paying for the privilege? The main question you should be asking is:

“If Interactive Brokers aren’t charging a trading commission for these ETFs, who’s really paying it?”

The answer, of course, is that you (the ETF purchaser) is - just not directly to Interactive Brokers. If you look in the fund prospectus at the “breakeven table” you’ll see that the 0.75% management fee is only part of the fee picture. Another question I find myself asking is “Why would anyone pay a 0.75% per year fee for someone to go long X and go short Y for you anyway?” but maybe that’s a question for another blog post. Anyway, when all the rest of the fees from the prospectus are included, like “brokerage commission and fees” and “operational expenses” the total fee is well over 1% per annum.

The moral of the story? When you see the word “free” used by anyone in the financial industry, simply substitute for “well hidden” and you’ll probably be closer to the truth.

Paul King

A little while ago an independent publishing company asked me to write a book and I’ve been working on it for a while and it’s almost finished now.  It’s called “Protect Your Wealth From Inflation” and it outlines a three step process for doing exactly what the title says.  The steps are:

1. Maintain the purchasing power of an emergency fund
2. Make a good real rate of return on your savings
3. Make a good risk-adjusted return in your investment accounts

I’ll be talking a bit more about it in future blog entries, but I’m sure that it’s going to be very useful for a lot of people especially in today’s interesting economic environment.

Paul King

The most recent book I read was Little Bets by Peter Sims. The book itself is a “little bet”, it’s not very expensive, not very long, and the downside is small, but you may just learn something that helps make the difference between a venture being successful or a (hopefully) graceful failure. I was a little surprised and disappointed when I came to the last chapter and the book was over (I was reading it on my iPad and “feeling” how far you are through a book by how thick each half is in your hands is just about the only feature of a paper-based book that can’t be reproduced digitally yet) which either means that it was a bit thin on content, or the content that was presented was so compelling that I wished there was more of it; I’ll let you decide which.

Little Bets: How Breakthrough Ideas Emerge from Small Discoveries

 Paul King

Even if you believe that markets are fractal in that they exhibit similar price  movement in different time frames what implications does this have for successful trading? Most traders following the rationale below:

1. If markets are fractal then they exhibit similar price movement in different time frames
2. Therefore I should trade in as small a timeframe as possible
3. This will give me more opportunity, shorter trade duration, a statistically significant sample of trades much quicker and lead to generally more consistent results

This all sounds like a good idea and a logical line of reasoning except for the problem that the empirical evidence from testing tells me the following:

Markets are fractal and there is similar price movement in different timeframes

However, shorter timeframes exhibit more “noise” ie random ro chaotic price movements and less signal ie trends.  Couple this with the increase in implementation costss (spread, slippage, commissions) relative to the size of winning trades, and also the increased likelihood of implementation errors or deviations from your method as defined, it means that trading in a short timeframe actually generates more volatile and poorer risk-adjusted return.

The solution?  Trade in as big a time frame as possible but allow for long periods of “normal” drawdown so you don’t suffer operational failure due to cashflow problem even if you accurately implement your method.

Paul King

My UK trading partner recently sent me a link to an article about making money trading with the phases of the moon.  I think it was mostly tongue-in-cheek, but I always like to see if any of these weird, off-the-wall trading ideas actually work.  Inevitably, in the actual article the way the “system” was tested was simply coparing returns to a buy-and-hold strategy which is uselessz in determining whether there is actually any trading edge to be found.  Any results that only look at return don’t mean a thing.

Therefore, I set about testing this idea in my own way (I was already very dubious that it held any merit, but until I’ve tested something I won’t dismiss it).  The main trading idea was to buy when there was a new moon and then sell when there was a full moon.  Fortunately all the data for moon phases is readily available from NASA at http://eclipse.gsfc.nasa.gov/phase/phasecat.html.  Armed with Excel, a quick web query, XLQ to get data for the SPY since 1993, and a bit of nifty VBA programming in a couple of hours I had the following results.

For “buy and hold” the SPY since March 1993 the CAGR% was 6.09%, the maximum drawdown was 55.19% (ouch) which gives us a MAR ratio of 0.11.  Obvioulsy these results suck so it’s not much of a target to beat anyway.

For the moon phase strategy over the same period the CAGR% was 3.01%, the maximum drawdown was 33.1% giving a MAR of 0.09.  So the results were worse and this didn’t even include increased transaction costs versus buy and hold.

The moral of the story?  When anyone says, “hey this trading idea works great” there are two main things to think about:

1. Exactly how has the performance been tested?
2. What is the risk-adjusted return like compared to a simple “buy and hold” strategy?

What I tend to find is not that people want to deliberately mislead everyone, it’s that their method of determining what works and doesn’t work in trading is flawed becuase it does not test the risk-adjusted return of a method.

Paul King

The other day someone asked me what my definition of “medium term” trading was, and I realized that I didn’t really have one.  Typically “medium term” would imply an average trade duration of weeks or months rather than days or years, but the interesting question is really “does it matter?”.  I take a minimalist approach to trading and never include anything in a trading system that does not definitely, absolutely, have to be there.  WIth this in mind it was interesting to me that I never considered trade duration when I build my trading programs.

I suppose the first this to think about it what actually determines the average trade duration.  For my systems (which don’t have any time-based stops) it’s basically how wide the stops are that is the main factor in trade duration.  If I used a very wide stop (or no stop at all) then, on average, I would stay in a trade far longer than if I used a very tight stop (less than 1 ATR, for example).  When I tune a trading system, the average trade duration is not a consideration - I simply want the best risk-adjusted return possible with whatever instrument class I’m dealing with.

So the real answer to why I am indifferent to average trade duration, or classifying systems into “short”, “medium” or “long-term” systems is becuase I only care about net risk-adjusted return that already takes into account realistic implementation costs - whether the best answer is that trades end up being days, weeks, months, or years makes no difference as long as I can accurately implement the trading system as defined.

Paul King

One of the golden rules of trading is “let profits run”.  This helps to maximize the average size of a winning trade which is the main way a trading program makes money.  Obviously to actually optimize the size of a winning trade one should use no stops at all on winners - but this is impractical since the profit for winning trades would never be realized.  For this reason the number of exit rules applied to winning trades should be minimal (but not zero) - a volatility-based trailing stop that is outside the “noise” of the timeframe you are trading in is normally sufficient.

However, there is another type of exit that is useful for winning trades and that is one which measures the rate of profit on the trade and exits when the trade is no longer making a sufficient rate of return to justify continuing to take the risk.  Exactly where this rate of profit threshold should be set depends on all the other aspects of your trading system and can only be tuned correctly if you simulate your trading methods in a sophisticated back-testing environment.  If you don’t do this then you’re just “taking a shot in the dark” regarding where the parameter settings should be for your exits and there is no way you can determine if you’re generating the best risk-adjusted return from the instruments you are trading.

My advice: Never trade a system with a full-size allocation of real money until you have simulated the historical performance of the system with your current parameter settings and tuned them to achieve the best risk-adjusted return possible.

Paul King

Often traders contact me and are frustrated with the (huge) difference between the historical simulation results (which inevitably look great) and their actual trading results (which inevitably don’t). Even with the most sophisticated testing software it’s very easy to fall into the trap of developing a system that only works well on past data and will perform erratically in real trading. This is generally not a fault of the software, or the markets being traded, but in the methods used by the trader to design, develop, and tune the trading program.

One key mistake is developing a trading program that relies on a high winning percentage of trades to make money. Typically, these kind of systems are short-duration, average winners are smaller than average losers, and the expectation is only significantly positive when the percentage of winning trades is greater than 60%. This is basically a system that makes money by trading an anomaly in past data and no matter what you do, real-time trading results will not look anything like historical testing because your actual winning percentage is likely closer to 50% (i.e. random chance) than the 60% (or higher) in your simulation. When your winning percentage drops to it’s long-term reasonable average, and implementation costs are taken in to account, you are lucky to break even on your trading.

Getting stuck in this cycle of developing a system that looks good in back-testing, trading it with real money and slowly losing your account, then going back round the development cycle again is a recipe for long-term failure of your trading business and the sooner you stop the vicious cycle the better it will be.

Paul King