My Hedged Fund - Another "Trend-Following" Post

[meanreversion]

If you fancy being blinded by science -

http://www.activetradermag.com/index.php/c/The_AT_Interview/d/Jonathan_Kinlay,_Systematic_Strategies

 

[michaelguan326]

Take two systems, both with positive expectancy, and add them together. If the correlation is negative, then risk adjusted return for the portfolio will be superior to either of the two, provided

1. Correlation of returns doesn't change adversely (a big if) and
2. Markets behave in a vaguely Gaussian fashion

In other words, a trend system could/should be complemented by a mean reversion system, but the trick is to find a consistently profitable system of this nature.

(Another option would be to consistently sell low delta out-of-the-money option spreads on the S&P)

Yes, the option strategy is used by some CTAs like Kevin Bruce, it will help reduce volatility of portfolio. It all boils down to equity curve smoothness. There's a guy in blox forum that did a experiment with combining several systems. You can see the combination really improved MAR. (the guy was sluggo)

To be honest, I have absolutely no idea in terms of how to develop a mean reversion system. Using oscillators seems to bring me back to think whether it is possible to use them to initiate trades as when I was a day trader before, Implementing oscillators was a fail..... Any entry exit ideas?

Also not sure if a combined portfolio should have more then one TF system. Implementing 2 TF systems seems be like double betting.

 

[meanreversion]

Yes, the option strategy is used by some CTAs like Kevin Bruce, it will help reduce volatility of portfolio. It all boils down to equity curve smoothness. There's a guy in blox forum that did a experiment with combining several systems. You can see the combination really improved MAR. (the guy was sluggo)

To be honest, I have absolutely no idea in terms of how to develop a mean reversion system. Using oscillators seems to bring me back to think whether it is possible to use them to initiate trades as when I was a day trader before, Implementing oscillators was a fail..... Any entry exit ideas?

Also not sure if a combined portfolio should have more then one TF system. Implementing 2 TF systems seems be like double betting.

You sound dangerously like you know what you're talking about, are you sure you're on the right website??

The conceptual issue with a mean reverting system working over time is that it relies on picking tops/bottoms to get in, then similarly you need to pick a nearby bottom/top to get out. Trend following recognises that picking turning points is hard to impossible (which it is), whereas mean reversion needs you to do it twice with each trade!

In addition, a trend system will have a handful of big winning trades which make up the bulk of returns, whereas pure mean reversion will have fixed profit:loss ratio, i.e. you can never have big winners by definition.

So whilst it's superficially true to say that "markets are range bound most of the time", this in itself doesn't lend itself to a robust, long term positive expectancy system.

Here are some options -

1. Wait for a failed breakout in order to go short, e.g. if the market breaches but then closes below the 10 day high, sell on the open. But where's the stop and profit target? You may have to go with a 1:2 win:loss ratio.

2. Incorporate some element of trend following, e.g. sell 2 units at the high channel, then buy 1 back at 1 ATR profit, move the stop to breakeven for the other unit, then operate a trailing stop loss. Very messy.

3. Systematically sell option spreads. Very tough to backtest.

The mean reversion system, if it's directional, needs to trade against the trend, otherwise it will correlate to the trend system. That's the tough part .... not only are you trading against the trend, but you need to pick local highs and lows.

 

[michaelguan326]

You sound dangerously like you know what you're talking about, are you sure you're on the right website??

The conceptual issue with a mean reverting system working over time is that it relies on picking tops/bottoms to get in, then similarly you need to pick a nearby bottom/top to get out. Trend following recognises that picking turning points is hard to impossible (which it is), whereas mean reversion needs you to do it twice with each trade!

In addition, a trend system will have a handful of big winning trades which make up the bulk of returns, whereas pure mean reversion will have fixed profit:loss ratio, i.e. you can never have big winners by definition.

So whilst it's superficially true to say that "markets are range bound most of the time", this in itself doesn't lend itself to a robust, long term positive expectancy system.

Here are some options -

1. Wait for a failed breakout in order to go short, e.g. if the market breaches but then closes below the 10 day high, sell on the open. But where's the stop and profit target? You may have to go with a 1:2 win:loss ratio.

2. Incorporate some element of trend following, e.g. sell 2 units at the high channel, then buy 1 back at 1 ATR profit, move the stop to breakeven for the other unit, then operate a trailing stop loss. Very messy.

3. Systematically sell option spreads. Very tough to backtest.

The mean reversion system, if it's directional, needs to trade against the trend, otherwise it will correlate to the trend system. That's the tough part .... not only are you trading against the trend, but you need to pick local highs and lows.

Apologies for taking the thread discussion off topic! To be honest I don't know what I am talking about! :cheesy: it's just that I've reading here and there and found certain things to make sense but I actually haven't tested much ideas out! I just stumbled upon the threads about systematic trading here at T2W forum, I was before at forex factory. Here at college I don't get a tad bit interested individuals; the interested ones only dabble in stocks so I find discussions in threads to be great areas where I can learn from other experienced members like you and the others.

I'll go take a look at your ideas and try to test then. Thanks so much thou

 

[myhedgedfund]

Apologies for taking the thread discussion off topic! To be honest I don't know what I am talking about! :cheesy: it's just that I've reading here and there and found certain things to make sense but I actually haven't tested much ideas out! I just stumbled upon the threads about systematic trading here at T2W forum, I was before at forex factory. Here at college I don't get a tad bit interested individuals; the interested ones only dabble in stocks so I find discussions in threads to be great areas where I can learn from other experienced members like you and the others.

I'll go take a look at your ideas and try to test then. Thanks so much thou

Well... go at it boys.

I am glad someone here knows that "The conceptual issue with a mean reverting system working over time is that it will have fixed profit:loss ratio. So whilst it's superficially true we should wait for a failed breakout in order to go short."

Please don't take this the wrong way... I feel like I'm learning a lot... or maybe I'm just oscillating here. I'll just take a deep breath and say "just go with the trend... it's simpler that way."

Boston

 

[michaelguan326]

Well... go at it boys.

I am glad someone here knows that "The conceptual issue with a mean reverting system working over time is that it will have fixed profit:loss ratio. So whilst it's superficially true we should wait for a failed breakout in order to go short."

Please don't take this the wrong way... I feel like I'm learning a lot... or maybe I'm just oscillating here. I'll just take a deep breath and say "just go with the trend... it's simpler that way."

Boston

I am 100% with you in Following the trend as its really dumb not to. The reason I am wary about TF is in the long run will the system I employ turn to be like obsolete just like the turtle system?

From reading your "my hedge fund" blog, you seem to be using a ATR band breakout with some MA filters. ( just a guess, you don't have to comment it it). I was also wondering which backtesting software you use and how extensively did u backtest your system?

 

[meanreversion]

Well... go at it boys.

I am glad someone here knows that "The conceptual issue with a mean reverting system working over time is that it will have fixed profit:loss ratio. So whilst it's superficially true we should wait for a failed breakout in order to go short."

Please don't take this the wrong way... I feel like I'm learning a lot... or maybe I'm just oscillating here. I'll just take a deep breath and say "just go with the trend... it's simpler that way."

Boston

Right, but it's about the quality of returns. Let's say you've seen 35% DD in your testing, maybe you actually experience 45% DD in real life at some point (I hope not!). Imagine if in three years you've had a good run and are up to $500k and the big drawdown strikes.. that's >$200k of "money out of the window" you'll have to contend with, equal to your original starting pot..!

Now let's say you could reduce your expected return by 5% and simultaneously reduce your max DD by 15%, you'd probably say yes to that.

That's where adding another system comes in, and it's something that all the systematic funds do. It's Markowitz's portfolio theory.. add an uncorrelated asset and improve the risk adjusted returns...

So now the question is, what asset is un- or negatively correlated to a trend system, and yet still has a positive expectancy. It used to be the stockmarket, maybe it still is. Or maybe there's another systematic approach that trades countertrend..

 

[michaelguan326]

Right, but it's about the quality of returns. Let's say you've seen 35% DD in your testing, maybe you actually experience 45% DD in real life at some point (I hope not!). Imagine if in three years you've had a good run and are up to $500k and the big drawdown strikes.. that's >$200k of "money out of the window" you'll have to contend with, equal to your original starting pot..!

Now let's say you could reduce your expected return by 5% and simultaneously reduce your max DD by 15%, you'd probably say yes to that.

That's where adding another system comes in, and it's something that all the systematic funds do. It's Markowitz's portfolio theory.. add an uncorrelated asset and improve the risk adjusted returns...

So now the question is, what asset is un- or negatively correlated to a trend system, and yet still has a positive expectancy. It used to be the stockmarket, maybe it still is. Or maybe there's another systematic approach that trades countertrend..

http://www.tradingblox.com/forum/viewtopic.php?t=8342&highlight=combining+systems

The above link should support both our point...

One thing I see myhedgefund doing to reduce volatility is by implementing some option hedge strategy. The only glimpse i ever came across is from Kevin bruce in this article:

http://www.michaelcovel.com/pdfs/bruce1.pdf

The point I am trying to get across is that there are other ways to reduce volatility. ( ie, options and trading uncorrelated markets with different systems, invest a couple of millions in winton capital and other cta will do the job)

Also a source of ideas I was referred to by a great mentor is Nelson freeburgs formula research....

 

[meanreversion]

The piece by Sluggo is almost an exact rehash of this link -

http://www.emccta.com/correlation/

Ok, so now we are all in agreement on the need for diversification, what exactly is this mythical system H ???? That's the million dollar question, but as sluggo puts it so well, you need to break a lot of rocks to find gold.

 

[michaelguan326]

The piece by Sluggo is almost an exact rehash of this link -

http://www.emccta.com/correlation/

Ok, so now we are all in agreement on the need for diversification, what exactly is this mythical system H ???? That's the million dollar question, but as sluggo puts it so well, you need to break a lot of rocks to find gold.

I wish I knew.....

Btw, from reading, you use a breakout model too right? Do you have other systems you implement? Also do you program your strategies and backtest? Which software do you use? I read your name and it suggests you probably know a lot on The subject of mean reversion...

His thread is probably the most I interesting one I've encountered haha,

 

[meanreversion]

Breakout systems seem to offer the best long term chance of success, probably because most people find it tough to trade this way, and most traders tend to lose money (I know the causality is not great, but you get the point).

I use Amibroker for backtesting and running algos, the language takes a little getting used to but for $280 it's an incredibly powerful piece of software.

 

[michaelguan326]

http://www.tradingrecipes.com/files/csmith.pdf

thought you may be interested, great article

 

[myhedgedfund]

Week 10: Well, it was just "The Quiet Before the Storm...."​

Funny how the market changes from week to week. It was only last week when I wrote about how it had been a quiet week for The Fund from a trading perspective, selling only one position in Week 9. This week (Week 10) the market sold heavily, and it triggered my selling or reducing of 14 positions. I am at week's end holding over 70% cash and returns are basically back to where they were at the beginning of the year; more on that shortly.

Before I get into the specifics though, let me write a couple of sentences on what I consider to be one of the greatest advantages of following trends. Fluctuations such as the ones we witnessed this week would have in the past created some anxiety for me, and I suspect the same would have happened for some of the readers. Trading based on fundamentals would have caused me to question my analysis, my decisions and maybe even the data which led to those decisions. Was my analysis correct? Did I miss anything? Was the decision the correct one based on the analysis? But wait... Was the data which I used in the analysis accurate? What if someone is "cooking the books?" Or what if they are misleading us? Does someone have access to information that I don't (well of course)?

It was interesting to me how relaxed and calm I was the entire week. I followed the system's signals to a "T." I sold what I needed to sell, when I needed to sell it... with absolute trust in the system. With trend-following, stages like these feel detached, unemotional... or at least with the range of emotions you feel when watching a movie. We may have some interest in the characters and the story, but the fact that we have a good sense of how the movie will end reduces the full range of emotions.

The fact that "Trend-Following" as a practice allows for a reduced spectrum of emotions cannot in my opinion be overstated.​

Transactions for the week
As previously stated, a fairly active week:

- Sold positions in Copper (JJC), Grains (JJG), Semiconductors (SOXL), Technology (TYH), Timber (CUT), Japan (EWJ), Financial Services (FAS), Nasdaq (TQQQ) and Metals and Mining (XME)
- Reduced positions in Energy (ERX), Canada (EWC), Natural Gas Exploration (FCG), Nuclear (NLR) and Oil & Gas Exploration (XOP)​

A good number of these positions (particularly Semiconductors and Financial Services), exhibited the well known "whipsaw" characteristics common in many TF systems. I bought (or added to existing positions) as the trend hit certain indicators, only to see it quickly reverse in a short amount of time. In situations like these we just shrug our shoulders and move on.

Performance for the year is now basically flat (-0.15%). Charts with weekly performance and benchmarks follow:

Current sector allocation breakout:

Specific holdings in more detail can be seen here.

What Next Week May Bring:
The reader may have noticed that I retain some energy positions, I would not be surprised if I receive buy signals in energy stocks this week, the trends in the sector remain solid. The same could be said for Commodities and the broad market S&P 500; I may buy there at some point this week. I also remain keenly aware of the developing opportunities in the short side, several developing country ETF's (Indonesia, Vietnam, Turkey) have shorting potential at some point.

Questions as always welcomed...

Happy Birthday to me!

Boston

 

[michaelguan326]

I really admire that you have the courage to be honest...

not many people can keep up doing this after seeing 2months of hard work erased in a single week

Good trading~

 

[leonarda]

Week 10: Well, it was just "The Quiet Before the Storm...."​

Funny how the market changes from week to week. It was only last week when I wrote about how it had been a quiet week for The Fund from a trading perspective, selling only one position in Week 9. This week (Week 10) the market sold heavily, and it triggered my selling or reducing of 14 positions. I am at week's end holding over 70% cash and returns are basically back to where they were at the beginning of the year; more on that shortly.

Before I get into the specifics though, let me write a couple of sentences on what I consider to be one of the greatest advantages of following trends. Fluctuations such as the ones we witnessed this week would have in the past created some anxiety for me, and I suspect the same would have happened for some of the readers. Trading based on fundamentals would have caused me to question my analysis, my decisions and maybe even the data which led to those decisions. Was my analysis correct? Did I miss anything? Was the decision the correct one based on the analysis? But wait... Was the data which I used in the analysis accurate? What if someone is "cooking the books?" Or what if they are misleading us? Does someone have access to information that I don't (well of course)?

It was interesting to me how relaxed and calm I was the entire week. I followed the system's signals to a "T." I sold what I needed to sell, when I needed to sell it... with absolute trust in the system. With trend-following, stages like these feel detached, unemotional... or at least with the range of emotions you feel when watching a movie. We may have some interest in the characters and the story, but the fact that we have a good sense of how the movie will end reduces the full range of emotions.

The fact that "Trend-Following" as a practice allows for a reduced spectrum of emotions cannot in my opinion be overstated.​

Transactions for the week
As previously stated, a fairly active week:

- Sold positions in Copper (JJC), Grains (JJG), Semiconductors (SOXL), Technology (TYH), Timber (CUT), Japan (EWJ), Financial Services (FAS), Nasdaq (TQQQ) and Metals and Mining (XME)
- Reduced positions in Energy (ERX), Canada (EWC), Natural Gas Exploration (FCG), Nuclear (NLR) and Oil & Gas Exploration (XOP)​

A good number of these positions (particularly Semiconductors and Financial Services), exhibited the well known "whipsaw" characteristics common in many TF systems. I bought (or added to existing positions) as the trend hit certain indicators, only to see it quickly reverse in a short amount of time. In situations like these we just shrug our shoulders and move on.

Performance for the year is now basically flat (-0.15%). Charts with weekly performance and benchmarks follow:

Current sector allocation breakout:

Specific holdings in more detail can be seen here.

What Next Week May Bring:
The reader may have noticed that I retain some energy positions, I would not be surprised if I receive buy signals in energy stocks this week, the trends in the sector remain solid. The same could be said for Commodities and the broad market S&P 500; I may buy there at some point this week. I also remain keenly aware of the developing opportunities in the short side, several developing country ETF's (Indonesia, Vietnam, Turkey) have shorting potential at some point.

Questions as always welcomed...

Happy Birthday to me!

Boston

I know exactly how you're feeling right now, although my account is peanuts compared to yours, it's been a hard week for me too, for very similar reasons...

 

[meanreversion]

Boston, have you traded a trend system for any length of time prior to this, on a consistent basis?

The only reason I ask is that - ok, sure, you're following rules so less stress and all, but if you go back and look at your backtests, you might see drawdown periods of 9 months or more.

Now maybe in ten years time you'll be fine, but do you think you'll stick to one trading system only, with the large volatility it entails? I'm only guessing, but right now I reckon you're running 30%+ volatility, which could wear you down over time.

 

[myhedgedfund]

Boston, have you traded a trend system for any length of time prior to this, on a consistent basis?

The only reason I ask is that - ok, sure, you're following rules so less stress and all, but if you go back and look at your backtests, you might see drawdown periods of 9 months or more.

Now maybe in ten years time you'll be fine, but do you think you'll stick to one trading system only, with the large volatility it entails? I'm only guessing, but right now I reckon you're running 30%+ volatility, which could wear you down over time.

Meanreversion,

Sorry for the delay in responding to your question.

In short, I do not rely on one trending system only. I also use another more conservative system (not described here, maybe I'll create a separate blog on that one later this year) which combines covered call writing and trends. Basically I identify a trend, and "lock in" short-term profits (one month out) by selling covered calls slightly out of the money; performance is almost as good as with the main TF system, but with a lot less volatility.

Boston

 

[myhedgedfund]

Week 11: Reducing Exposure...​

Trends have for the most part dissapeared, and may at best be in transformation mode. The chart below is typical of previously trending investments which have seen deteriorating trends for ~2 weeks now.

Transactions for the week

The deteriorating trends resulted in several sell signals early in the week:

- Sold positions in Agriculture (DBA), Russell 2000 (TNA), S&P 500 (UPRO), Aerospace (ITA), Midcaps (MWJ), Russia (RSX), DJIA (UDOW), Industrials (XLI), Media (PBS), Broker-Dealers (IAI), Real Estate (IYR), Water (PHO), Materials (XLB) and Consumer Discretionary (XLY).
- I added to my existing positions in Energy (ERX), Natural Gas Exploration (FCG) and Canada (EWC)​

I am at this point heavily into cash, waiting for trends to develop.

Performance for the year is sligthly positive (1.52%); charts with weekly performance and benchmarks follow:

What Next Week May Bring:
The energy sector came back after a short pullback; I may find myself adding more blocks to the energy-related positions. Short opportunities may also materialize.

Questions as always welcomed...

Boston

 

[michaelguan326]

what happened? did this discontinued?

 

[meanreversion]

The website says he's pretty much all in cash as of March 20th, waiting for trends to develop. I think there have been trends since then, so I'd be interested to see an update.