Emergency Exit for Weekly System

qldfrog said:

Do you check that 4.5% fall daily or weekly.
If you do the check weekly, are you not at the mercy of timing?
And should you not do a daily check similar to:
Exit all if fall on the last n session is above xx %?

Click to expand...

Weekly at this stage. You can see from the sims that it halved the March DD to around 8-9% which is pretty respectable and something I can live with. I’m definitely interested in doing a daily check as you suggest, but haven’t had time to code and sim yet, but I will do.

MovingAverage said:

Weekly at this stage. You can see from the sims that it halved the March DD to around 8-9% which is pretty respectable and something I can live with. I’m definitely interested in doing a daily check as you suggest, but haven’t had time to code and sim yet, but I will do.

Click to expand...

interested in the results..obviously not ideal as the aim of weekly is to keep serene until EOW... but choosing an arbitrary date like Friday night to decide doing the check could just bet on a repeat timing of a crash..obviously not a given.
Anyway, very interested and good job btw, especially after having 2 catastrophic weekly results on my own in a row

qldfrog said:

interested in the results..obviously not ideal as the aim of weekly is to keep serene until EOW... but choosing an arbitrary date like Friday night to decide doing the check could just bet on a repeat timing of a crash..obviously not a given.
Anyway, very interested and good job btw, especially after having 2 catastrophic weekly results on my own in a row

Click to expand...

I've been thinking about this this morning and given my emergency exit based on the weekly time frame has reduced the March drawdown from around 20% to around 8-9% I'm not sure moving my emergency exit from weekly to daily will yield much of a dramatic DD improvement on the current 8-9%. My feeling is that expecting to reduce the DD even further is probably optimistic as I think an 8-9% DD is pretty respectable. Still, I'm interested in trying so will see how it goes. Having said all of that, I think moving my emergency exit to daily timeframe could well benefit the performance of my daily swing system so I'm going to focus my effort on seeing what impact my emergency exit will have on my swing.

MovingAverage said:

I've been thinking about this this morning and given my emergency exit based on the weekly time frame has reduced the March drawdown from around 20% to around 8-9% I'm not sure moving my emergency exit from weekly to daily will yield much of a dramatic DD improvement on the current 8-9%. My feeling is that expecting to reduce the DD even further is probably optimistic as I think an 8-9% DD is pretty respectable. Still, I'm interested in trying so will see how it goes. Having said all of that, I think moving my emergency exit to daily timeframe could well benefit the performance of my daily swing system so I'm going to focus my effort on seeing what impact my emergency exit will have on my swing.

Click to expand...

What i mean is that your weekly check is a daily check done on friday night:
is there a way your DD reduction is not a strike of luck because the last crash was timed well for that check?
What if the crash had started 2 days later, would your same code not have missed the whole exit by a week? and so you would have been hit much harder?
That's all what i mean.Conceptually, if i was to trigger such an exit, i would do it along these independence from crash start.just because we have too few historical data with such crashes, means we can not sorely rely on thatmarch backtest for timing..yet need to take the scale of the crash into account
And genuinely interested in your work.
Hope i am not confusing you, or seem to belittle you, in no way the idea.
being in a middle of a move, i do not have enough free time to spend researching properly that area..yet as you, i believe it is needed now

qldfrog said:

Hope i am not confusing you, or seem to belittle you, in no way the idea.
being in a middle of a move, i do not have enough free time to spend researching properly that area..yet as you, i believe it is needed now

Click to expand...

All good...taken as a constructive observation. Although I am a little confused as to understand what your suggesting so forgive me if I'm misinterpreting you. But here goes. I'm not the sharpest tool in the shed so maybe if you can give me a simpler actual example I can better tell you how my emergency exit will respond to that scenario.

qldfrog said:

What i mean is that your weekly check is a daily check done on friday night:

Click to expand...

Are you suggesting that on Friday night I check each and every daily bar for that week; that is, separately check each of the five daily bars for Monday to Friday and if any of those bars has a 4.5% drop then trigger my emergency exit? If this is what you're suggesting then I think it would be prone to false triggers. For instance, assume Wednesday of a given week had a 4.5% drop but then had a strong rebound on the Thursday and Friday to recover Wednesday drop and put on gains--that to me would incorrectly sell of my entire portfolio despite the market finishing the weekly strongly. Like I say, maybe I am misunderstanding your suggestion.

qldfrog said:

What if the crash had started 2 days later, would your same code not have missed the whole exit by a week? and so you would have been hit much harder?

Click to expand...

My code (and my emergency exit) is not dependent on when the crash started during the week--it can start Mon, Tue, Wed, Thu or Friday. So long as any single week closes 4.5% (and I should say a week is Mon to Friday) down from open then it will trigger. Sure a decline in the market might be slower and spread over many weeks and in that situation my emergency exit will not be triggered because the 4.5% decline occurs over many weeks with each week seeing only a 1% or so decline. I'd mentioned this in one of my earlier posts, my emergency exit will only capture a "flash crash" scenario...it will not catch slower sell offs.

qldfrog said:

That's all what i mean.Conceptually, if i was to trigger such an exit, i would do it along these independence from crash start.just because we have too few historical data with such crashes, means we can not sorely rely on thatmarch backtest for timing..yet need to take the scale of the crash into account

Click to expand...

I am keen to better understand the scenario you're talking about as I'd like to make sure my emergency exit is solid. Thanks for the input.

simply speaking:
if the index is stable all week but fall 4 % on Friday: you stay in
then on Monday it falls 4% and stay there all week, you are still in: after a 8% fall which can repeat itself again and again;
I acknowledge your normal exit might trigger but in that scenario:

the market has fallen 8% in 2 sessions, yet all is good for that weekly detection
and potentially if tuesday to friday have been disastrous you would still wait till the next Monday to react
All good for a long slide down, not for a big crash
I would prefer a : if the index falls by more than say 4.5% (could be different value now) in 5 last session, sell all;
if the week had been shifted to the blue lines I added below (roughly, just to share the idea), would you have exited?That is my worry in a nutshell

have a great week end

qldfrog said:

simply speaking:
if the index is stable all week but fall 4 % on Friday: you stay in
then on Monday it falls 4% and stay there all week, you are still in: after a 8% fall which can repeat itself again and again;
I acknowledge your normal exit might trigger but in that scenario:

the market has fallen 8% in 2 sessions, yet all is good for that weekly detection

Click to expand...

Absolutely, my threshold being 4.5% then yes my emergency exit would not trigger because you'll have two bars each with a decline of less then 4.5%. But in that first week if the Friday fall was 4.6% then my emergency would trigger. If I look back at XAO during March my emergency exit only kicked in halfway into the decline, but that was early enough to reduce the drawdown from 20% to 8-9%.

You got me thinking @qldfrog

I just changed my code so the XAO decline is measured across 2 bars (two weeks); that is, I now compare the close of the current week to the open of the prior week. I ran some optimizations and it now appears the sweet spot is a decline of 5% over two bars (weeks). In summary, the overall profit took a bit of a hit but the March COVID drawdown is improved. Here's the sim results, which is over the same period as my original post.

The equity chart below shows I've sacrificed some equity for improved DD (no surprise there) but nothing major.



You can see from the below that the DD has now been reduced to around 6% for the COVID March.





And the MC results seem a lot tighter than the single bar strategy, which I really like.



I have a feeling that this multi-bar approach will probably turn in an improved DD over the 08/09 crash. I'll be running some sims over that period shortly. I'm also going to look at the performance of this over a range of bars not just 2 so will let you know how that goes. Thanks for pointing me in this direction--it looks promising.