Weekly Rotational Drawdown

[TaranveerSingh]

Hi guys,
I’ve been working on a weekly rotational system (on Indian Markets).

Approach 1 –

I have an index filter (C>MA)
A stock filter (C>MA)
ROC as momentum rank

CAGR – 18%
MDD – 37% (2008 and compounded losses of 2018-19)

I modified multiple aspects but it more or less remained the same.
CAGR – 17-20%
MDD – 37-40%

There were steep drawdowns (the markets were truly choppy in 2018-19)

2 suggestions by Nick made some significant difference –
1) Using regression as a momentum ranking tool. (reduced the drawdown to about 35ish % over a wide range of parameters)
2) Using a volatility filter
ATR < a percentage of close ^Now this got both the CAGR and drawdown significantly while the CAR?MDD ratio remained more or less consistent throughout.
(However, a stable 17-19% returns and 31-33% drawdown was maintained over a wide range of ATR lookback periods.)

While I understand it’s a never ending process, would anyone be kind enough to share their experience in how they continued to work upon their systems to reduce steep drawdowns?

I also understand that one’s gotta take drawdowns (which isn’t a problem). But any approach you took that made a major difference?

Thanks.

[ScottMcNab]

You could experiment with position sizing…instead of being all in or all out (with the index filter or stock filters) there are (volatile) situations where decreasing position size may be beneficial in reducing maxDD without harming cagr toooooo much

[ScottMcNab]

Another option, and one I am looking to employ in New Year is to take the same system and just trade it across 3 or 4 markets….eg divide equity into 4 and add US, Aussie and Canadian markets to the Indian market…trade all 4 using the same weekly rotational system (maybe 5 positions on each for a total of 20)…my testing has found this to be another effective in reducing maxDD without impacting cagr too badly

[TaranveerSingh]

Thanks Scott. The diversification across various markets definitely makes sense. I have easy and cheap access to Indian markets, and a viable one to American markets too. Hopefully within a year I’ll move to Canada so I’ve thought of having that too if Norgate provides that data.

I was also looking to have bi-weekly systems on rotational so the weekly would hopefully take me out of a falling market soon and the bi-weekly wouldn’t hurt me in a choppy market. (Don’t have the stomach for monthly.) Would love to know your opinion.

About the previous suggestion, I am rather comfortable mentally with the 100% or 0% involvement.

[ScottMcNab]

I have tried trading monthly rotational systems for several years but have stopped (again). I can see the advantages of a monthly system in the backtest (as I can, for example, in using margin in my MRV and MOC systems) but it just does not sit well with me either and so I have finally conceded that I will be unable to stick with it long term. I feel like I am strapped into a plane seat as the plane plummets and the flight attendant tells me to relax as the plane nearly always regains altitude. I am now re-working my rotational systems to weekly too. To be honest, I will probably need a year or so of live trading with that before I know psychologically if I am ok with it.

On a more important topic Taranveer, are you going to be moving close to a ski slope ? Love BC but there are so many ski fields it’s probably not that far regardless where you live.

[TaranveerSingh]

Scott McNab wrote:

I have tried trading monthly rotational systems for several years but have stopped (again). I can see the advantages of a monthly system in the backtest (as I can, for example, in using margin in my MRV and MOC systems) but it just does not sit well with me either and so I have finally conceded that I will be unable to stick with it long term. I feel like I am strapped into a plane seat as the plane plummets and the flight attendant tells me to relax as the plane nearly always regains altitude. I am now re-working my rotational systems to weekly too. To be honest, I will probably need a year or so of live trading with that before I know psychologically if I am ok with it.

What you say is absolutely understandable. If you’re interested in the middle ground and need to test for Bi-weekly rotation, you can try this –

//Rotate only on Monday, every 2nd one
Mon = DayOfWeek() countMon = Cum( Mon ); // count all first days
DofW = Mon AND (countMon % 2 == 0 ); //checks if the sum is an even number. takes a trade only then

I hope you came to this realization before a major drawdown? Either way it’s great that you are moving towards your comfort.

Scott McNab wrote:

On a more important topic Taranveer, are you going to be moving close to a ski slope ? Love BC but there are so many ski fields it’s probably not that far regardless where you live

Got a couple of acquaintances and relatives (not much difference haha) in BC but I have the nomination of Ontario.
So I’d most likely be moving to Brampton. Do you visit Canada frequently?

[ScottMcNab]

Thanks Taranveer. I will look into the bi-weekly.

We take one holiday a year..been going to Whistler for 3-4 weeks either side of Christmas for last 4 or 5 years ..approx 1/2 way between Australia and UK (where my brother lives) so meet in middle… not really seen a lot of Canada apart from that small region…sister in law used to live in Ottawa for quite a few years..they used to skate along river to work in winter !. I think she said it was quite important to also be able to speak French in that part of Canada for business

[TaranveerSingh]

Scott McNab wrote:

We take one holiday a year..been going to Whistler for 3-4 weeks either side of Christmas for last 4 or 5 years …

I’ve always been in India so really keen to see western Christmas.
Sounds like a great tradition.

Scott McNab wrote:

they used to skate along river to work in winter !.

Haha! just in Canada.

Scott McNab wrote:

I think she said it was quite important to also be able to speak French in that part of Canada for business

Yes, this changes from region to region and also depends on profession. From the conversations with people I’ve talked to, I should be good. Thanks for the heads up though. Cheers

[Anonymous]

Hey Taranveer – first off, thanks for that snippet on “every other Monday”, I will have to do some testing there and add that into my code snippet section. Have you thought about how to tell the backtester to use a Tuesday if Monday is a holiday? We have a few of those in the states yearly that may fudge the numbers.

I briefly looked at trading my MOC system on the NSE – only as far as realizing that I can’t do it because I’m not allowed! Hah!

Have you tried butchering the ROC and making some calculations of your own? Like taking a long time frame ROC and adding it to a short time frame ROC and ranking on your total? With this method you can also apply some weight to the longer or shorter period if you think one is more important.

Some of the podcasts that I have listened to have traders mentioning that with momentum the most recent month is mean reverting, so they omit that month from the calculation. On my momentum system that did not yield positive results, but it may for your market.

The other thing to try is adding a WorstRankHeld in your system. Say you have MaxOpenPositions = 5 and WorstRankHeld = 6, if one of your 5 stocks go to place #6, it will not exit. Some food for thought. I’m not at my workstation so I’ll link to the AFL reference (it is near the top): https://www.amibroker.com/guide/afl/setoption.html

Cheers!

[Anonymous]

Oh yeah, the other thing that I forgot: I trade my monthly rotational twice in two separate tranches. Half of the equity I trade on the 1st, and the other half I trade on the 16th. I’m doing this to try and eliminate signal luck – Nick’s idea. I like this method, for what it is worth.

[TaranveerSingh]

Seth Lingafeldt wrote:

Hey Taranveer – first off, thanks for that snippet on “every other Monday”, I will have to do some testing there and add that into my code snippet section. Have you thought about how to tell the backtester to use a Tuesday if Monday is a holiday? We have a few of those in the states yearly that may fudge the numbers.

Hey Seth. Thanks for such an elaborate response. My bad on calling the variable “Monday”. It’s basically the first trading day of the week. So whether it’s Monday or Tuesday, we’re good. What I personally noticed in a few tests was that it was close to the weekly rotational in performance but with less trades. However, monthly does noticably better.

Seth Lingafeldt wrote:

I briefly looked at trading my MOC system on the NSE – only as far as realizing that I can’t do it because I’m not allowed! Hah!

Ya? Do you trade any market other than Australian/American?

Seth Lingafeldt wrote:

Have you tried butchering the ROC and making some calculations of your own? Like taking a long time frame ROC and adding it to a short time frame ROC and ranking on your total? With this method you can also apply some weight to the longer or shorter period if you think one is more important.

Some of the podcasts that I have listened to have traders mentioning that with momentum the most recent month is mean reverting, so they omit that month from the calculation. On my momentum system that did not yield positive results, but it may for your market.

Yes I had this discussion with Nick. Theoretically it does make sense but I haven’t tunneled into it to confidently accept/reject it just yet. I will however. This served as a good reminder, Thanks!

Seth Lingafeldt wrote:

The other thing to try is adding a WorstRankHeld in your system. Say you have MaxOpenPositions = 5 and WorstRankHeld = 6, if one of your 5 stocks go to place #6, it will not exit. Some food for thought. I’m not at my workstation so I’ll link to the AFL reference (it is near the top): https://www.amibroker.com/guide/afl/setoption.html
s!
Cheer

So currently I’m at 20 open positions and worst rank at 35. I tested at multiple worst rank held parrameters and it didn’t budge much so I’m comfortable about where I am there.

Next on my agenda is this –
I’ve coded multiple permutations of 2 stock filters together and testing that with ranking being ROC. I think I’ll take this weekend to see how it differs if I keep a varied/weighted ROC as the ranking.

Thanks once again.

[Anonymous]

Only US market for me. No ASX or any others. Some cryptocurrency, wherever you would put that.

[Author]

Posts