Taranveer’s Journal

[TaranveerSingh]

Hello good people. I felt I’m gonna have too many questions to keep adding new topics.

Been a part of the course for quite a while now. I’ve been trying to develop a momentum system for over 2 months now. Initially wanted to trade Indian markets but due to lack of reliable data providers, US markets seemed a better option.

My objective was to have a rotational system that traded a broad index’s constituents (currently looking at R1000).
My desired CAGR was 15-20% and I’d accept an MDD of less than 30%.

Starting from the basics, I’ve tried multiple permutations. Initially I did get mildly acceptable systems. Thanks to Nick’s guidance and the forum, I added some difference makers like ROC/ATR ranking, multiple lookbacks (in both ranking and index filter) while keeping stock filter rather simple.

I reached a more than desirable stat in 16% CAGR and 19% MDD. But this was a monthly system so I tested for when the rotation would be fixed for other days in the month. Say 5th or 10th.

While the system sustained on the 5th day, on the 10th trading day it went to 11% CAGR and 32% MDD (courtesy, 2020 March).

For those of you using rotation (especially in trenches), does your system sustain the performance through other days of rotation?

I’m using the following bit of code to test for the same. If you do use rotation, would you mind sharing how your performance changed?

EOM = Month() != Ref(Month(),-1);
delay = Param(“Rotation day of month”, 1, 1, 20, 1);
refdelay = delay-1; //to rotate on nth trading day, we’d have to use Ref(eom, -(n-1))
tradeday = Ref(eom, -refdelay);
PositionScore = IIf(Year()>=1985 AND tradeday,score,scoreNoRotate);

Since I don’t know what I don’t know, any unsolicited advice is most welcome!

[Anonymous]

I don’t bother checking for other rotation days. I can hear Nick in my ear saying “law of large numbers” and “next 1000 trades” and something something “luck”… so between the trading days (I trade on the 1st and 16th), I don’t bother opening the broker to look at the account! And I refuse to check it, too… because I don’t want to FOMO and change my strategy!!

[Anonymous]

There’s absolutely some rule that you could institute in the backtest that would give you stellar results, something like:
Exit every third Wednesday and reenter the following Monday at the Close.
I see this as no different to me picking the 18th and the 29th for rotation days, just because they look good.

[ScottMcNab]

Taranveer Singh wrote:

While the system sustained on the 5th day, on the 10th trading day it went to 11% CAGR and 32% MDD (courtesy, 2020 March).

For those of you using rotation (especially in trenches), does your system sustain the performance through other days of rotation?

Since I don’t know what I don’t know, any unsolicited advice is most welcome!

I looked into this a bit in Jan 2019 (after my rotation systems got hit hard towards the end of 2018) and I there are some results posted in my journal regarding my findings. I found some systems were relatively stable regardless of the day selected for the rotation while others had quite large fluctuations in the cagr and maxdd…at the time I was wondering if this stability (or lack of) that could be used as a measure of robustness of the system….?

[TaranveerSingh]

Seth Lingafeldt wrote:

There’s absolutely some rule that you could institute in the backtest that would give you stellar results, something like:
Exit every third Wednesday and reenter the following Monday at the Close.
I see this as no different to me picking the 18th and the 29th for rotation days, just because they look good.

So the doubt was pretty legit I think. Not looking for the best day here, exactly the opposite in fact. looking for consistency (instead of the peak) as we do in parameters?

I do feel however that change in rotation day is far more significant than a change in parameter even in an otherwise robust system. A market can continuously trend downwards on any date of the month. If it does so between 1st and 15th, the 15th date rotation will suffer and vice versa.

That isn’t necessarily a system problem I feel. It’s the concept of monthly rotational. No matter the system, if you’re 100% invested and a downturn starts, you can’t do much. It’s just signal luck and as Nick told me in one of our initial calls, “we can’t avoid it, we can only dilute it”. Which is fair enough.

Scott, thanks for redirecting me to your journal.
I actually did something very basic after noticing this inconsistency myself. I went through the course lessons about robustness and if I’m not mistaken, in the module “Is my system broken?” there’s some text that says –
(paraphrasing)
If there’s a one-off drawdown that can be easily explained due to the market conditions, the system isn’t broken.

And well this is where true “acceptance” of risk gets tested, despite the knowledge of risk.

I also noticed the following in many parameter variability tests I’d run on some systems –
Even if the drawdown of some parameters was large, the returns would actually be similar to the rest of the parameters. I don’t know if any confident conclusion can be made here but maybe over time it works out?

Would love to know your thoughts.

[TaranveerSingh]

So my IB account just go approved. I was prompted to select some settings.
I’ve selected “Non professional” > free delayed data subscription.
Base currency is USD (Indian rupee wasn’t an option)
I also changed my brokerage from fixed to tiered.

Does the data subscription play any role in executing rotational systems? I hope not?
Any other word of advice? It’s one of those global accounts.

[JulianCohen]

Data subscription doesn’t play any part in rotational systems….unless you want to sit in front of your computer and watch your money accumulate and or disappear in real time that is

Don’t worry you can place orders the way you have it set up.

[TaranveerSingh]

Julian Cohen wrote:

….unless you want to sit in front of your computer and watch your money accumulate and or disappear in real time that is
.

Wait why would I not want that?

Julian Cohen wrote:

Don’t worry you can place orders the way
you have it set up.

Thanks

[TaranveerSingh]

Execution of volatility based position sizing

Hey guys.

Could anyone redirect me to a relevant post or share an efficient workflow of how you go from exploration to actually listing down what buy/sell orders you manually execute?

[ScottMcNab]

Taranveer Singh wrote:

Seth Lingafeldt wrote:

There’s absolutely some rule that you could institute in the backtest that would give you stellar results, something like:
Exit every third Wednesday and reenter the following Monday at the Close.
I see this as no different to me picking the 18th and the 29th for rotation days, just because they look good.

That isn’t necessarily a system problem I feel. It’s the concept of monthly rotational. No matter the system, if you’re 100% invested and a downturn starts, you can’t do much. It’s just signal luck

Another idea you could experiment with and hopefully improve on is the idea of an intra-month kill switch or bailout if it hits the fan (again, it is questionable if gets activated enough times to have a good sample size though…in which case may just be curve fitting to dodge 1 or 2 large draw downs)

BailOut = !IndexPass;
AnyDayofWeek = DayOfWeek() != Ref(DayOfWeek(),-1);
TradeDay = (AnyDayofWeek AND Ref(BailOut,-1)) OR EOM;
PositionScore = IIf(Year()>=1955 AND TradeDay,score,scoreNoRotate);

May well work better if instead of having

BailOut = !IndexPass

having some other kill switch (eg Index drops 10% in 5 days or hits insane vola level) ?

eg

IndexKill = IndexVola>60 or IndexROC5<-10;
Bailout = IndexKill;

[JulianCohen]

Taranveer Singh wrote:

Execution of volatility based position sizing

Hey guys.

Could anyone redirect me to a relevant post or share an efficient workflow of how you go from exploration to actually listing down what buy/sell orders you manually execute?

I know there is a post here somewhere as I remember writing one a few years ago, but I’m not sure where.

Basically best bet is do the following.

Run your scan in AB with the updated account balance.

If this is the first entry for the system, or after an Index Filter lay off, then paste the results into a spreadsheet so you can manipulate them.

Add up the totals (number of shares * close) of the first few entries and put that in a column by the side, then add them up looking to see that your account total is used up. The last entry will probably have to be cut back (# shares) in order to allow the total to use up all the spare cash.

If this is the second month/period of entry do the same thing but make sure that you look to see which positions you are already holding, and adjust the share amounts to reflect that position first. Then look to do the same thing.

I have deliberately left this a little unspecific as the best thing is to create your own workflow so you understand exactly what you need to do.

I’m sure Nick is available should you get really stuck.

[GlenPeake]

…Just to add to the 2nd part of what Julian was describing i.e. “2nd Month/Period of Rotation”….

Have a read through the following post…. I hope it helps.

https://edu.thechartist.com.au/kunena/running-your-trading-business/615-end-of-month-rotational-process.html

[JulianCohen]

That’s the one I was looking for!

[GlenPeake]

Julian Cohen wrote:

That’s the one I was looking for!

The only reason I knew the thread your were referring to is the fact that I was the ‘Rotational Rookie’ asking the question lol

[JulianCohen]

Yup I can see that

[Author]

Posts