How to trade the Unholy Grails system?

[craft]

Craft, again no argument from me. Thats why I take the discretionary way.

And you had better get back to it and me to reading some annual reports

Cheers.

 

[qldfrog]

A system has to be evaluated across all market conditions – cherry picking the return from a trend system in a bull market gives you a false sense of reality as does Monte Carlo simulations if you destroy a serial correlation that actually exists.

true but you can backtest your system against a falling market period, a choppy one and a bull one
then based on your risk appetite you choose parameters with fits best in all situations;
my current "system" (a work in progress ) is designed that way, will not be the best in any specific market but should offer nice enough result over the long range and i stress this the long range, I am on for a 5 years or so trip and will see how it goes
I will probably go for other variant etc each with their own portfolio and I am determined to see how it ends up on the long term

 

[sinner]

But no trigger is perfect and getting whipsawed in sideways markets is the norm.

The classic paper "Dynamic Strategies for Asset Allocation" by Perolde and Sharpe went into this topic deeply. I think what you are describing is essentially the result of a crowded trade (at least from the perspective of the paper).

http://www.stanford.edu/class/msande348/papers/PeroldSharpe.pdf

 

[tech/a]

true but you can backtest your system against a falling market period, a choppy one and a bull one
then based on your risk appetite you choose parameters with fits best in all situations;
my current "system" (a work in progress ) is designed that way, will not be the best in any specific market but should offer nice enough result over the long range and i stress this the long range, I am on for a 5 years or so trip and will see how it goes
I will probably go for other variant etc each with their own portfolio and I am determined to see how it ends up on the long term

Attempting to design a system which will perform in all market conditions is futile in my view

You only need 2
Trending
Or
Ranging.

You need to be trading both.
That way you'll catch trends and profit in ranging markets while your trending system is switched off or under performs.

 

[aarbee]

Table 39 on pg 107 of "Unholy Grails" outlines the performance of the Techtrader system after 1000 iterations in Monte Carlo simulation.
CAGR Avg/21.29% Range Min/14.53% Range Max/30.82%
maxDD Avg/39.3% Range Mon/49.9% Range Max/32.7%

In terms of the Drawdowns, the figures are quite high as such. Beginning to trade it with 3x leverage would be sheer madness based on the figures presented. Even with the knowledge that Tech turned a 30k float to 360k, it would not be prudent to trade the system with any leverage based on the results outlined above. The lower DD figure of 32.7% would still be way outside the comfort zone of nearly all traders.

I am currently working on an EOD system showing a Win Rate of 82%, about 110 trades per year, with a DD of around 3% and annual return of 4.7% (not compounded). With these figures, leverage is a great tool. Even with 5x leverage the DD would be around 15% which would be acceptable to most traders, while the returns can be quite juicy. This is generally not possible with Trend Following systems which by their nature have relative higher drawdowns and leverage would only make these worse. One should not forget that DD is the number one reason most traders start second guessing or quit trading their systems. TF systems by their nature are difficult to trade by most people, primarily due to whipsaws and DDs (length and depth thereof).

Nick has outlined quite a few TF systems in Unholy Grails for all to see. However, I will be very surprised if there would be too many people who will end up trading these consistently over a period of time even if they get attracted by the nice looking compounded equity curves showing final equities in millions of dollars. I for one never backtest with equity compounded but those realistic curves don't look sexy in books and publications touting particular systems.

As for the topic of the thread, the systems in the book can be easily traded. They are not hard and do not require much time. I have been trading one of the systems and will get a good idea of it's performance after trading for atleast another one or two years. With ideas presented in the book, even better systems can indeed be designed.

Cheers

 

[kid hustlr]

Good discussion and aarbee's thoughts about the techtrader are very good.

Using leverage on a system with a high maxDD can obviously create problems.

In regards to the actual question in the OP, I actually consulted several different sources with the exact same question and Nick himself was kind enough to reply to me. Effectively I asked him 'I've read your book, what do I do now?'

His reply was below:

You have a variety of options to move forward.

You can set up a watchlist and do it manually. Obviously you'll need faith yourself that it works, so doing it this way is somewhat 'geusswork' and very laborious. Perhaps its not psychologically the strongest way to proceed.

You ocan download our turnkey code that operates with Amibroker. This enables you to run the systems as designed with the knowledge the codde is correct (many have sent theor own versions and much of it is garbage). Doing it this way allows you to test and validate it for yourself, but also tinker and change parameters to better suit your own objective. You will need to buy Amibroker and then also the data to run it.

Lastly, you can subscribe to our Growth Portfolio, which is how I manage my very own SMSF and manage money for others. There is no need for software or data, you know its being monitored and personally traded by me, and we also offer ongoing research and support.

Anyway, not trying to sell you - just saying you have a variety of options to think about if you wanted to proceed.

I think which option one decides upon is going to vary depending on the individual.

- If I was running my own SMSF and had a decent balance, I'd just subscribe to the Growth Portfolio. This seems like an obvious option.

- Purchasing the turnkey system and validating ideas yourself is ideal if you have the time/skill/ability. Any concerns about how the system operates can be answered as you can explore them yourself.

- In theory, if you don't have the time or expertise, you could just pick one of the systems and trade it. If you have a trading account, paid subscription for data and the ability to set up a watchlist (i'm sure you could do this in Ninjatrader or some other software) in theory you can trade it. You just won't have done all the validation work yourself and may not have the same 'peace of mind' during a drawdown.

EDIT: Nick's attitude is really really good by the way.

 

[chops_a_must]

But no trigger is perfect and getting whipsawed in sideways markets is the norm.

I here lots of discussion on Monte Carlo but how many chunk the actual market equity curve in varying block sizes to preserve the serial correlation?

As soon as you Monte Carlo resample using a period of one you are making an implied assumption of no correlation or full randomness and that’s’ just not how the market works. It’s quite feasible that a trend system will turn on multiple times to trends that fizz out early – and you can’t know in advance which one will be the real deal so you have to take every signal. Under these false starts its easy to get multiple losers across a correlated portfolio and have it happen two, three, four or more times before a sustained trend gets under way. The reality of what correlated instruments do is almost inconceivable under a random distribution.

A system has to be evaluated across all market conditions – cherry picking the return from a trend system in a bull market gives you a false sense of reality as does Monte Carlo simulations if you destroy a serial correlation that actually exists.

I think you're missing a main point here...

Trading systems aren't, or shouldn't be traded in isolation.

Like tech says, he's range trading as well.

Personally, if you're a tech trader, I reckon you want three time frames: long term, mid term and short term. Complementing each other, and smoothing out your overall returns. And working in different market conditions.

From memory tech a like many of us bought bottom drawer stuff un leveraged in the gfc.

Although I'm not sure how to develop a range trading system, as most of the types of trades for that I've experimented with end up having to rely on mean reversion, and aren't strictly range trades.

Anyway, I've made my point and I'll save my other opinions for discussion elsewhere on the topic.

Really good thread and discussion though.

 

[tech/a]

Table 39 on pg 107 of "Unholy Grails" outlines the performance of the Techtrader system after 1000 iterations in Monte Carlo simulation.
CAGR Avg/21.29% Range Min/14.53% Range Max/30.82%
maxDD Avg/39.3% Range Mon/49.9% Range Max/32.7%

In terms of the Drawdowns, the figures are quite high as such. Beginning to trade it with 3x leverage would be sheer madness based on the figures presented. Even with the knowledge that Tech turned a 30k float to 360k, it would not be prudent to trade the system with any leverage based on the results outlined above. The lower DD figure of 32.7% would still be way outside the comfort zone of nearly all traders.

In reply.

The maximum D/D was in the GFC.

My own filters had me out in July 2008.
The system actually was stopped out of ALL trades ( the one I was trading ) before the Max % was reached.
Nick left T/T in its raw state something he didn't do with the others---making suggestions for improvement with all others but not T/T

Adding that it is probably the most tweeked method around as its been available for all to play with.
It also shows only 1 negative year in 14 yrs (2008) with 2009 returning 113%
There is a 97% chance of being profitable in 30 mths

I have also played with it and have 4 T/T methods which have its base premise but far better results.
Drawdown can be minimised (19% through the GFC) is my best with the T/T based methods.

Was never meant to be the system of systems but was and is meant to show that anyone can build one from scratch.---Even a DUCK.

I urge everyone to learn from the journey many of us have been through and are going through.
A few of us put stuff up for the general populace to rip apart---that's what its for ---if you didn't then there would be nothing learnt.

I am currently working on an EOD system showing a Win Rate of 82%, about 110 trades per year, with a DD of around 3% and annual return of 4.7% (not compounded). With these figures, leverage is a great tool. Even with 5x leverage the DD would be around 15% which would be acceptable to most traders, while the returns can be quite juicy.

Aarbee

LTCM had a very similar premise based upon an ARB strategy.
Very high win rate in its 90% s I believe. Drawdown--low single figures.
Ideal for Leverage.
I guess you know what happened to LTCM?

 

[chops_a_must]

Attempting to design a system which will perform in all market conditions is futile in my view

You only need 2
Trending
Or
Ranging.

You need to be trading both.
That way you'll catch trends and profit in ranging markets while your trending system is switched off or under performs.

Agreed on your first point. Half agree on your second. Because I'm not sure on your terminology.

 

[Gringotts Bank]

I think there are much better systems out there than those in the Unholy Grail, no disrespect intended to Nick.

2003-2007 bull run was highly unusual in it's length, strength and lack of volatility. Any trend following system would have done well in that time frame. For those interested in turn-key, the following systems might be of interest because they adapt to the conditions, and you're not guessing as to when the market is trending or ranging. None of those massive DD's to worry about.

http://www.adaptivetradingsystems.com/trading-systems.html

 

[tech/a]

I think there are much better systems out there than those in the Unholy Grail, no disrespect intended to Nick.

2003-2007 bull run was highly unusual in it's length, strength and lack of volatility. Any trend following system would have done well in that time frame. For those interested in turn-key, the following systems might be of interest because they adapt to the conditions, and you're not guessing as to when the market is trending or ranging. None of those massive DD's to worry about.

http://www.adaptivetradingsystems.com/trading-systems.html

Having the best trading system on earth in "The Unholy Grails" was never Nicks intention.
But if you'd have read the book you'd know that.

Its actually designed for people like you who are drawn to equity curves like the one shown on your link.
In the hope you maybe better equipped to keep your $2995 plus updates in your pocket.
But if you'd have read the book you'd know that.

Not every trend following systems trader or discretionary trader made a profit in those years--
But you'd also know that.

 

[aarbee]

Hi tech,

My remarks were put forth as a reminder that use of leverage should be predicated on sound analysis of numbers for a system. GFC happened and so for anyone considering trading this or any other similar system has to base their judgements on the basis of backtesting results including that period. Of course, OOS testing would be part of the validation process. Like I mentioned, I am trading a properly tested momentum based system tweaked from ideas presented in Nick's book.

In reply.

Aarbee

LTCM had a very similar premise based upon an ARB strategy.
Very high win rate in its 90% s I believe. Drawdown--low single figures.
Ideal for Leverage.
I guess you know what happened to LTCM?

I know what happened to LTCM but you are pre-supposing too many things in lumping the system I mentioned in passing to that of LTCM's ARB strategy. I have been a trend-follower and it is still my prime strategy. However, there are many short term EOD strategies that are very viable. Of course, like any other, they need to be properly tested (IS and OOS) and validated before trading with real money. More importantly, the performance of the strategy needs to be monitored on continuous basis. Contrary to what you appear to be inferring, all these strategies don't have to go down the LTCM way.

For profitabe trading through all seasons, the trader's arsenal should have multiple strategies trading different markets in different time frames exploiting the different traits of the market's structure.

Cheers

 

[tech/a]

Sorry I seem to be missing something.

What is it that you believe bought about the demise of LTCM ?

 

[aarbee]

Sorry I seem to be missing something.

What is it that you believe bought about the demise of LTCM ?

No idea what you are missing tech. Matter of fact, I am quite confused as to what brought on LTCM in the discussion. I'll take a shot. It must be the mention of high win rates and leverage in my posting. Your earlier reply seemed to indicate that (and I might be completely wrong here) high win rates, low DD and leverage means going down the toilet like LTCM.

There is reams of literature on what caused the demise of LTCM. No point rehashing the whole thing. However, I'll say when events like LTCM happen, it is more than the strategy that paves the way. There were systemic problems in LTCM. Even strategy wise, they were using quite complex ones dreamt up by academics and Nobel winners. They also ended up using very high leverage.

Quote
At the beginning of 1998, the firm had equity of $4.72 billion and had borrowed over $124.5 billion with assets of around $129 billion, for a debt to equity ratio of over 25 to 1. It had off-balance sheet derivative positions with a notional value of approximately $1.25 trillion, most of which were in interest rate derivatives such as interest rate swaps. The fund also invested in other derivatives such as equity options.
Unquote

Contrast this with what I talked about. A quite simple Mean-reversion strategy with high win rate, low DD, and a simple 5:1 leverage trading very liquid instruments. For the life of me I can't think of how this would equate to risks of going down the LTCM way.

There are a lot of inefficiencies in the market that can be exploited to extract profit. Trends (Momentum) and Mean-Reversion are two of the most prevalent. As a retail trader, we should have systems in our arsenal to trade for both to smooth out our equity curve. Between the two, Trend-Following is by far the more difficult to trade due to whipsaws, deeper drawdowns, long string of losses etc. In difficult times it does require a lot of fortitude to stay the course.
Other short-term, high win rate, low DD strategies are easier to trade, complement the TF ones and most importantly are far easier to monitor the health of during trading. One has a very prompt feedback in such systems when they go out sync with the market because of frequency of trades etc.

It is rather disingenuous of you to defend 3:1 leverage in a TF system (with relatively high DDs) which by it's very nature is hard to monitor for performance but condemn to ignominy a 5:1 leverage of a frequently trading system with low DD and is far easier to monitor and based on a very persistent inefficiency in the market structure.

Notwithstanding above, I will record here my appreciation of all the benefits the live discussion of TechTrader system has brought to many traders and doubters of Trend-Following, and systems trading.

Apologies to the starter of this thread for taking it off-topic.

Cheers

 

[CanOz]

Apologies to the starter of this thread for taking it off-topic.

Cheers

No worries, great post aarbee. Cleared that one up for me too.

CanOz

 

[sinner]

Although I'm not sure how to develop a range trading system, as most of the types of trades for that I've experimented with end up having to rely on mean reversion, and aren't strictly range trades.

If you have a look at the Sharpe paper I posted he explains "Constant Mix" trading strategies, these are strategies that perform best within ranging environments. Please consider that "mean reversion" is a constant mix style strategy.

 

[tech/a]

LTCM's demise was leverage.

Even a 2% move against ---which was all but never heard of--- would smash it.
While if it could have been funded it may have survived---it couldn't and didn't.

My point was as is yours that poor use (Perhaps cavalier use) of leverage will if things turn pear shaped have the potential to ruin any trading method.

Back to T/T.
and leverage.

I wont ask you read the complete topic BUT.
Position sizing was 10% of capital and risk was 10% of entry price.
so 1% on leveraged capital/trade and 3.33% on initial capital.
That to me was acceptable. risk of ruin on testing was Zero and Potential for profit
was 100%.

Testing which I completed in 2002---didn't have the GFC in the test.
In fact from memory the worst case draw down from testing was 23-24%
at the time of completion of testing.

Plan was that if the System performed under the numbers returned that I would not trade it.
Youll find quite a bit of discussion where the system was out performing the mean for quite
sometime and the question asked----why keep trading it if its outside the test series.
Answer was that it was still in the tested ranges.

It did trade lower than the 23% and when it did I sold out.
We kept the live one going and in the end all positions crossed the exit with no new triggers forth coming.
It eventually made a new equity high in 2010 according to Nicks testing--although I don't know--the live portfolio is very likely to have as well.

Thanks for the discussion enjoyed it.

 

[chops_a_must]

If you have a look at the Sharpe paper I posted he explains "Constant Mix" trading strategies, these are strategies that perform best within ranging environments. Please consider that "mean reversion" is a constant mix style strategy.

Fair enough.

 

[skc]

Table 39 on pg 107 of "Unholy Grails" outlines the performance of the Techtrader system after 1000 iterations in Monte Carlo simulation.
CAGR Avg/21.29% Range Min/14.53% Range Max/30.82%
maxDD Avg/39.3% Range Mon/49.9% Range Max/32.7%

Aarbee, do you know what risk per trade were these drawdown based on?

Position sizing was 10% of capital and risk was 10% of entry price.
so 1% on leveraged capital/trade and 3.33% on initial capital.
That to me was acceptable. risk of ruin on testing was Zero and Potential for profit
was 100%.

How does 3.33% risk per trade relates to the % risk per trade used in testing?

 

[odds-on]

You probably need to recalibrate your BS detector. The market itself did better than that at times. Even a term deposit back in the 90s could do that...

Fair point, so I did some checks. The RBA website has monthly bank term deposit rate history going back to 1982. I downloaded the data and checked how many months the term deposit rate was greater than 15%, the results are below:

1 month - 0 cases
3 month - 9 cases (approx. 2% of the sample set)
6 month - 12 cases (approx. 3% of the sample set)
1 year - 7 cases (approx. 2% of the sample set)
3 year - 0 cases.

According to the data the maximum term deposit rate was 17.25% (!!!) on a 3 month term deposit rate in Oct 1989!

You make a fair point but due to the low occurrence of >15% rates I do not think my detector needs recalibrating.