Dump it Here

[qldfrog]

Skate's probably off on hols for a whilte ArtMaster. Everyone needs some downtime

Good to see some green shoots in the market. Tricky times, but yes, most market filters probably indicating buy signals from recent activity (weekly timeframe).

Nick Radge has been known to pop in from time to time here - I'm not sure that they actively trade/follow those strategies any more, but would be interesting to hear from anyone that does actively trade variations on the Unholy Grail themes.

FWIW I have 3 weekly strategies
1 relatively aggressive and which started buying a few weeks back, the other 2 more conservative with their index filter
Only one of these 2 had 1 buy on a parcel this morning....we are not in a clear long trend yet for weekly

 

[qldfrog]

Ps I assume Mr Skate is enjoying a well deserved rest at the moment..

 

[Sid23]

@ArtMaster , @Newt , I've been running BBO from the book since mid 2021 after learning Amibroker, thought about Flipper but at the time I was chasing the highest CAGR.

Plenty info here on the 20% Flipper, here is one link from Skate end 2021- https://www.aussiestockforums.com/threads/dump-it-here.34425/post-1152285

Drawdown of the original BBO was pretty high so soon after added the filter BBO(F), as per Nick's book, which at the time some of the backtesting I did showed lower maxDD (pg 124 showing difference between filter & no filter, 11% lower DD, 2009-2011, pg 126 comparison, Flipper with filter even lower DD). Without the filter, these strategies seem to have fairly high DD, which can be hard to handle emotionally when you're starting out, though if I had better understanding of DD, I would probably start by testing 20% Flipper vs BBO, both with & without additional filter. Would be interesting to see if anyone else has Flipper and able to run backtest of previous year.

Now onto the 3rd version of the BBO, as Amibroker knowledge increases & new ideas arise, you can start making improvements on the basic system. Version 3 now has other parameters and doesn't rely solely on the bottom band to sell, though still got hammered last year. Programming it yourself really helps understand it's performance and where you can make tweaks. Improving DD for me became more important.

Backtesting BBO & BBO(F) over multiple time frames (whether 5, 10 or 20 years), surprisingly having filter didn't consistantly improve maxDD.

Backtest over the past 18 months (from when I started trading BBO, 01/06/21-17/1/23), original BBO (no Filter) shows CAGR 0.2% and maxDD of -20.28%, whereas BBO (F) had a CAGR of -4.86% & maxDD of -22.29%.

SID.

 

[Newt]

Great stuff thaks Sid23. Tough times indeed to be starting strategies, but wondeful info and thanks for sharing. Seem to recall Nick saying the Unholy Grails strategies were no longer traded for liquidity reasons. Not sure what that means TBH, but still stands as a seminal book on system trading.

Skate has put out many posts through this thread about various market filters. Nothing ever perfect, and balancing "robust" against "responsive" is always best judged in the eye of the beholder.

 

[Richard Dale]

Trust nothing. Provie it for yourself.

i.e. Backtest.

 

[farmerge]

Trust nothing. Provie it for yourself.

i.e. Backtest.

And trust nobody, my case I have that very wise sage as my "assistant/adviser" the bloke in the mirror.

 

[Nick Radge]

Avoiding drawdowns leads to data mining and curve fitting, which in turn decreases system robustness. We've seen a regular user on here show the merits of data mining to lower the drawdown, but that will backfire eventually. It 'appears' to look good on his short test period, but will most likely fail over longer periods.

The robust way to lower drawdown is to diversify strategies. When I have more time I'll take a weekly strategy with a 16% CAGR and 42% maxDD, i.e. one that nobody would touch. Then add the same strategy but on a monthly timeframe, and combined the CAGR is 24% with a maxDD of 22%. So rather than modify, or 'adapt' (read curve fit) the drawdown out, we use it to retain the robustness.

If I combine 8 of my strategies, shortest being a weekly swing system, I get a CAGR of 24% with a maxDD of 12%. Now one of those 8 strategies includes the weekly CAGR 16% and maxDD -42%. Rough and ready is robust.

Discarding a strategy because of a poor profile is not necessarily the correct action. That poor performer could well be the best performer over a certain window and keep the full portfolio on an even keel.

Using a variety of BBO versions 'could' do that, so long as they're different timeframes and different markets. Shake them up - different lookbacks, different regime filters, different parts of the market and different markets etc etc

 

[qldfrog]

Avoiding drawdowns leads to data mining and curve fitting, which in turn decreases system robustness. We've seen a regular user on here show the merits of data mining to lower the drawdown, but that will backfire eventually. It 'appears' to look good on his short test period, but will most likely fail over longer periods.

The robust way to lower drawdown is to diversify strategies. When I have more time I'll take a weekly strategy with a 16% CAGR and 42% maxDD, i.e. one that nobody would touch. Then add the same strategy but on a monthly timeframe, and combined the CAGR is 24% with a maxDD of 22%. So rather than modify, or 'adapt' (read curve fit) the drawdown out, we use it to retain the robustness.

If I combine 8 of my strategies, shortest being a weekly swing system, I get a CAGR of 24% with a maxDD of 12%. Now one of those 8 strategies includes the weekly CAGR 16% and maxDD -42%. Rough and ready is robust.

Discarding a strategy because of a poor profile is not necessarily the correct action. That poor performer could well be the best performer over a certain window and keep the full portfolio on an even keel.

Using a variety of BBO versions 'could' do that, so long as they're different timeframes and different markets. Shake them up - different lookbacks, different regime filters, different parts of the market and different markets etc etc

I like the focus on the different timeframes as this is something I have not really gone into yet
I did try adding daily to the weekly strategies I use but the logistic was too hard..aka too difficult to maintain vs life.
I might consider a monthly one now

 

[DaveDaGr8]

I like the focus on the different timeframes as this is something I have not really gone into yet
I did try adding daily to the weekly strategies I use but the logistic was too hard..aka too difficult to maintain vs life.
I might consider a monthly one now

Different timeframes and strategies. If you play around and get the right balance you can more than halve your overall MDD. Even with a dog of a system like the NASDAQ in the mix. This is from 2021 to present too, so it wasn't even in the 2000 tech wreck.



Not only the MDD, but also the Average Drawdown and the depths of drawdowns are significantly reduced.

The main trick is to find systems that are uncorrelated so that they will all behave positively, just at different times. Hopefully ( in theory) as one goes down, another goes up and balances them out and reduces the drawdown.

2 great books that go into this in detail
Andreas Clenow - Trading evolved / Rob Carver - Systematic Trading.

 

[Nick Radge]

Excellent example @DaveDaGr8. Thanks

 

[DaveDaGr8]

Excellent example @DaveDaGr8. Thanks

I spend too much time playing with excel i think !!!.

 

[Sid23]

It certainly makes sense if you can build a suite of systems, and as you mention @DaveDaGr8 ,that are uncorrelated, to smooth out total MDD rather than trying to avoid DD as Nick mentions.

Knowing when you're improving or curve fitting your system seems to be a "hazy" area. You're only ever going to make improvements when system doesn't seem to be performing (which I did after a few large trade DD) during a current timeframe.

When is your system good enough to be considered robust? Or is it better to focus on a reasonable system (accepting occasional large DD), with aim of have a bunch of uncorrelated systems for an overall robust portfolio?


SID

 

[DaveDaGr8]

OMG ... there are about 10 loaded questions in there .

I'll try and be brief.

YOU need to be confidant that your system is robust. If you're not confidant then why ? What made you change your system (apart from 2022 being a dog of a year) ?

What do you consider an occasional large DD ? 20%, 30%, 50% ? Again, what can you handle ?

Running multiple systems comes with it's own headaches. You need more money to fund them to start with, you also have a lot more work to do managing several systems, how do you segregate the individual portfolios in one account etc. The reward is theoretical much lower account Drawdowns and less risk if one system blows up.

 

[Gringotts Bank]

Ideally a system will tap into something so fundamental that it won't have periods of not working. Standard indicators like MA's aren't fundamental. Buying and selling volumes are fundamental; time is fundamental.

 

[DaveTrade]

OMG ... there are about 10 loaded questions in there .

I'll try and be brief.

YOU need to be confidant that your system is robust. If you're not confidant then why ? What made you change your system (apart from 2022 being a dog of a year) ?

What do you consider an occasional large DD ? 20%, 30%, 50% ? Again, what can you handle ?

Running multiple systems comes with it's own headaches. You need more money to fund them to start with, you also have a lot more work to do managing several systems, how do you segregate the individual portfolios in one account etc. The reward is theoretical much lower account Drawdowns and less risk if one system blows up.

I'm no expert on systems but I do know a bit about trading and I have to agree with @DaveDaGr8 that there are many ways to skin a cat. Multiple systems looks to be a valid solution but it may not suit everyone, those of us that can't take the multiple systems approach for whatever reason will just have to find the way that does suit our circumstances and personality, making the required compromises in order to do so.

 

[Sid23]

OMG ... there are about 10 loaded questions in there .

I'll try and be brief.

YOU need to be confidant that your system is robust. If you're not confidant then why ? What made you change your system (apart from 2022 being a dog of a year) ?

What do you consider an occasional large DD ? 20%, 30%, 50% ? Again, what can you handle ?

Running multiple systems comes with it's own headaches. You need more money to fund them to start with, you also have a lot more work to do managing several systems, how do you segregate the individual portfolios in one account etc. The reward is theoretical much lower account Drawdowns and less risk if one system blows up.

Yeah, 2022 not too good for most, let alone starting new systems (my BBO version started mid 21 & WTT beginning 22), first dog was a big dog.

Thanks for highlighting some of those questions, especially - "Again, what can you handle ?", as did some more backtesting and the drawdowns through 2022 still fell within system expectancies, even if they were at the max. Still learning to accept that markets go down, who would've thought?

There is a third being Chartist Growth which I have no input so just follow signals. It will make for interesting comparison in time and help deciding whether to drop one in future as there are similarities to my BBO, though I suppose always other ways to find something a bit more uncorrelated, like US market, platforms also now charging a lot less to trade there.

It is a bit tricky whether to stretch funds to cover extra system or not. They are all big enough (just) to not be too affected by transaction costs it is interesting to see how each performs in comparison and whether 1 system carries the others too often. A bit more time will tell if they can stand on their own whilst also working together, or not.


SID

 

[MovingAverage]

Happy belated new year everyone. 2023 off to a good start for breakout traders (especially if you're skewed to miners). But let's not get too confident as we have some key index resistance levels approaching.

 

[MovingAverage]

There is a third being Chartist Growth which I have no input so just follow signals. It will make for interesting comparison in time and help deciding whether to drop one in future as there are similarities to my BBO, though I suppose always other ways to find something a bit more uncorrelated, like US market, platforms also now charging a lot less to trade there.

It is a bit tricky whether to stretch funds to cover extra system or not. They are all big enough (just) to not be too affected by transaction costs it is interesting to see how each performs in comparison and whether 1 system carries the others too often. A bit more time will tell if they can stand on their own whilst also working together, or not.


SID

The Chartist Growth is just another breakout system...whether you use BB, X week high or some other breakout metric they are all breakouts and will always have a high degree of correlation. Want to diversify then select a system that is not breakout based and reduce correlation.

 

[Sid23]

Ideally a system will tap into something so fundamental that it won't have periods of not working. Standard indicators like MA's aren't fundamental. Buying and selling volumes are fundamental; time is fundamental.

I guess there are many ways to find uncorrelated systems, One really needs to open up the mind, so easy to spend time trying to fine tune further what already seems to work fairly well.

Certainly has me now thinking (with renewed enthusiasm) ways to looking into fundamental indicators and not just a breakout as @MovingAverage mentioned.

These responses have been gold, thanks to all. It really does open up some ideas and different ways to approach building a system that I hadn't thought too much about.

SID

 

[Gringotts Bank]

I guess there are many ways to find uncorrelated systems, One really needs to open up the mind, so easy to spend time trying to fine tune further what already seems to work fairly well.

Certainly has me now thinking (with renewed enthusiasm) ways to looking into fundamental indicators and not just a breakout as @MovingAverage mentioned.

These responses have been gold, thanks to all. It really does open up some ideas and different ways to approach building a system that I hadn't thought too much about.

SID

Probably uncorrelated instruments is easier than uncorrelated systems.

Anyone remember skc with his pairs trading? I remember he did quite well from that. Not many people talk about that style