Dump it Here

[Skate]

Wimps
In fact, I’ll go so far as to claim that a smart wimp who runs and hides when the going gets tough generally produces better results than brave souls who are proud of their ability to suffer great monetary pains while they wait for their convictions to be rewarded. The problem with the “macho” approach to the markets is that the consequences of being wrong are so onerous. The stronger your convictions and beliefs, the more invested and braver you are, the greater the chance for a backbreaking loss. The wimpy trader knows that the key to success is staying in the game for the very long term.

@peter2 it's those with the experience that make the most sense. The secret in this game (since COVID) is to grab profits & cut your losses quickly. By cutting those losses quickly results in a lower drawdown with increased profits. (of which you are an expert BTW).

Weekly system traders don't enjoy the luxury of exiting early during midstream
Systematic trend traders need to have this added layer of protection coded within their buy condition & sell strategy (so you don't buy false breakouts whilst ensuring you exit quickly as possible). With a low 36-45% strike rate the exit strategy has to be sharp & responsive. Any stalling of momentum is good enough for me to pull the pin. Relying on a two-stage trailing stop just won't cut it these days. Also, drawdowns don't hurt as much when you are playing with their money. Getting deep into profits is the tricky part of trading.

Skate.

 

[peter2]

Getting deep into profits is the tricky part of trading.

Yes, and buying as early as possible is so important for two reasons. An earlier entry provides a better R:R outcome and an earlier entry allows price to wobble a bit after the start.

Thank you for juxtaposing my name with your wimp quote. LOL. I'll take it as a compliment.

I am a trading wimp. If my trades don't start well almost immediately, I cull them. I had a close call with the BRN trade last week. My entry was 0.495 with the iSL at 0.445. The iSL could have been placed at 0.46, just below the HVBB that got me interested. Being a wimp, I placed it below 0.45 (extra cautious).

When BRN closed 0.475, I'm thinking oh no looks like I'm going to sell this one soon. Considered selling at 0.46 (to reduce loss) and re-buy > 0.52. Luckily (was it luck or did I correctly assess the probability of the opportunity presented by the price, volume data ?) price bounded higher soon after.

Long may the trading wimps last.

 

[Skate]

When BRN closed 0.475, I'm thinking oh no looks like I'm going to sell this one soon. Considered selling at 0.46 (to reduce loss) and re-buy > 0.52. "Luckily" (was it "luck" or did I correctly assess the probability of the opportunity presented by the price, volume data ?) price bounded higher soon after.

let's not mention the word "luck" again as it flames a different range of opinions. Let's use the words "good fortune" instead of "luck" as it might be more palatable to some.

Was it Luck?
@peter2 we try very hard to refrain from using the word "luck" in the "Dump it here" these days. So there is no "further debate" around this word, let's just say you were blessed with "good fortune" in holding (BRN) enjoying a bounce that was not predicted.

Skate.

 

[Cam019]

The secret in this game (since COVID) is to grab profits & cut your losses quickly cut your losses short and let your winners run.

Perfect.

 

[Skate]

The secret in this game (since COVID) is to grab profits & cut your losses quickly. cut your losses quickly and let your winners run.

@Cam019, I do agree with the premise & hear it quoted so often it makes my ears bleed. It's similar to someone saying "buy low & sell higher" without telling others how to do it.

The "Dump it here" thread is the perfect platform
This thread is perfect for you to explain "so others can understand" how you go about "cutting your losses quickly and letting your winners run".

The markets have changed
I often hear that markets have changed and new rules are needed for the new game, well the markets do change but the underlying fundamental rules for success don’t seem to.

With changed trading conditions since COVID
I, personally had to change the filters & parameters of many of my trading strategies "to adapt to the increased volatility". To my surprise, a "Take Profit Stop" generated larger returns since early 2020 & I attribute this to "changing" trading conditions. Also, including an "Ulcer Index Indicator" to my (entry & exit condition) has paid off handsomely.

Getting out is when you make the money
These days I've learned to be a wimpy trader & using the Ulcer Index Indicator (that only measures the downside risk) is important to the risk-averse trader like @peter2 & myself. I'm sure it will be invaluable if adopted in any trading strategy & "be blown away" with the improvements it makes.

The Ulcer Index focuses on downside risk rather than the upside
Traders care about the downside risk because of the stress it causes. The Ulcer Index Indicator focuses on the "bearish volatility" that compares the highest point reached with the current price. All I'm saying, if you want to be a great mechanic - have the necessary tools to do the job.

Skate.

 

[Cam019]

@Cam019, I do agree with the premise & hear it quoted so often it makes my ears bleed. It's similar to someone saying "buy low & sell higher" without telling others how to do it.

The "Dump it here" thread is the perfect platform
This thread is perfect for you to explain "so others can understand" how you go about "cutting your losses quickly and letting your winners run".



With changed trading conditions since COVID
I, personally had to change the filters & parameters of many of my trading strategies "to adapt to the increased volatility". To my surprise, a "Take Profit Stop" generated larger returns since early 2020 & I attribute this to "changing" trading conditions. Also, including an "Ulcer Index Indicator" to my (entry & exit condition) has paid off handsomely.

Getting out is when you make the money
These days I've learned to be a wimpy trader & using the Ulcer Index Indicator (that only measures the downside risk) is important to the risk-averse trader like @peter2 & myself. I'm sure it will be invaluable if adopted in any trading strategy & "be blown away" with the improvements it makes.

The Ulcer Index focuses on downside risk rather than the upside
Traders care about the downside risk because of the stress it causes. The Ulcer Index Indicator focuses on the "bearish volatility" that compares the highest point reached with the current price. All I'm saying, if you want to be a great mechanic - have the necessary tools to do the job.

Skate.

@Skate, first and foremost - I hope you get something to fix your bleeding ears.

Moving right along.

This is where you and I are going to disagree, which is fine. To each their own.

Now, I am not sure over how many years you backtest with all the systems you design and build. Most of the backtest results for the systems of yours I have seen are from memory, 5 years or less. Most of them also seem to have very good metrics and quite good CAGR/MaxDD ratios. Sometimes it seems like they are a little too good to be true. Which brings me to my next point.

I agree that markets and market volatility do change over time, however I would be very skeptical about someone modifying system parameters just because they didn't like or couldn't handle the drawdown of the system due to one, two month black swan event.

In my opinion, if the system design was robust, and a long enough backtest with a large enough sample size was run before the systems implementation, the backtest would have shown that there was the potential for a drawdown of such a magnitude within the systems expected results.

However, if the system/s have been curve-fitted or market regime fitted, (and to me that sounds like what you might be doing if you're changing your systems parameters due a drawdown you'd prefer to avoid and only back testing your system changes from early 2020), and a backtest had only been run over a smaller window of time, e.g. 3-7 years then the system might have had a drawdown that the system designer was not aware could happen.

You said that

To my surprise, a "Take Profit Stop" generated larger returns since early 2020 & I attribute this to "changing" trading conditions.

however my question to you would be; add in that change and run a backtest over 20-28 years of historical data free of suvivorship bias and see what effect the change has on the system.

If you're only making changes to your systems and testing over the period of which you want to reduce that uncomfortable drawdown - in my opinion, your system is not robust.

 

[Skate]

I agree that markets and market volatility do change over time

@Cam019 what a great post. Well constructed comments with a variety of personal views are welcomed, alternative views even more so. When others post it's refreshing than reading information slanted from my experience & views I hold. When members respond in such a detailed way as yourself, @peter2, @ducati916, @qldfrog, @Newt, @cynic, @MovingAverage, @KevinBB, @othmana86 & @DaveTrade just to name a few off the tip of my tongue adds tremendous value to this thread.

Everyone is entitled to express their view
Whether a poster is right or wrong has never been the underlining reason for encouraging others to make a post. Placing yourself in another person's shoes is how we learn to understand a different point of view. The only thing I would suggest, don't explain why you think others are wrong but rather express an alternative.

Is this still on the cards?
I for one would be very interested in reading how you would suggest to others how they should go about accomplishing "cutting your losses quickly and letting your winners run". My ears only bleed when hearing catchphrases. Explanations rather than catchphrases go further in educating others. Education is the main "purpose" of this thread.

This thread is perfect for you to explain "so others can understand" how you go about "cutting your losses quickly and letting your winners run".

Skate.

 

[qldfrog]

@Skate, first and foremost - I hope you get something to fix your bleeding ears.

Moving right along.

This is where you and I are going to disagree, which is fine. To each their own.

Now, I am not sure over how many years you backtest with all the systems you design and build. Most of the backtest results for the systems of yours I have seen are from memory, 5 years or less. Most of them also seem to have very good metrics and quite good CAGR/MaxDD ratios. Sometimes it seems like they are a little too good to be true. Which brings me to my next point.

I agree that markets and market volatility do change over time, however I would be very skeptical about someone modifying system parameters just because they didn't like or couldn't handle the drawdown of the system due to one, two month black swan event.

In my opinion, if the system design was robust, and a long enough backtest with a large enough sample size was run before the systems implementation, the backtest would have shown that there was the potential for a drawdown of such a magnitude within the systems expected results.

However, if the system/s have been curve-fitted or market regime fitted, (and to me that sounds like what you might be doing if you're changing your systems parameters due a drawdown you'd prefer to avoid and only back testing your system changes from early 2020), and a backtest had only been run over a smaller window of time, e.g. 3-7 years then the system might have had a drawdown that the system designer was not aware could happen.

You said that



however my question to you would be; add in that change and run a backtest over 20-28 years of historical data free of suvivorship bias and see what effect the change has on the system.

If you're only making changes to your systems and testing over the period of which you want to reduce that uncomfortable drawdown - in my opinion, your system is not robust.

Really? let's put that into context:
You have a system doing great for the 20 years backtest 1, what do you think will be the results in the attached picture, same backtest in 2
I can somehow bet you a lot that if you have a good system for bt period 1 or 2, you will get pathetic results on ongoing live systems 1 and 2..
Conditions change, markets change..
And your systems should IMHO remember that whatever you do if you use backtests as part of your system design, you assume that the past will somewhat repeat in the future..
That is the assumption number one of system trading..
So we will disagree?
Nikkei index....

 

[Cam019]

Really? let's put that into context:
You have a system doing great for the 20 years backtest 1, what do you think will be the results in the attached picture, same backtest in 2View attachment 132852
I can somehow bet you a lot that if you have a good system for bt period 1 or 2, you will get pathetic results on ongoing live systems 1 and 2..
Conditions change, markets change..
And your systems should IMHO remember that whatever you do if you use backtests as part of your system design, you assume that the past will somewhat repeat in the future..
That is the assumption number one of system trading..
So we will disagree?
Nikkei index....

@qldfrog, I didn't say it was bulletproof. We can never know how the markets are going to perform in the future. However in my opinion, backtesting your systems parameters over a backtest period that gives you a large enough sample size to ensure robustness and longevity of the system is very important. The larger the sample size of trades the better.

Why not backtest between 1980 - 2000 or 2010? Oh yeah, because that's not beneficial to your point of view.

You have also just shown us perfect examples of confirmation and hindsight biases.

We will continue to disagree. ?

 

[Sir Burr]

let's put that into context

OK, are you guys backtesting a daily, weekly or monthly. Less data for daily, significantly more for a monthly system. Also, what's your average trade length to consider a fair backtest period.

Middle of the road weekly having an average trade length of 40 weeks. I guess no less than 5 years be OK and when do you test it, on latest data or during multiple market types?

Many ways and you probably already know but how about running an auto walkforward? Optimise for a few years then backtest, repeat etc etc. Then there's Monte Carlo plus that nice 3D optimiser image AB has. Try find yourself a flat plateau

I think trying all, so the confidence to place the cash is there.

Survivorship mentioned, don’t even bother without Norgates expensive top priced data, especially for the longer-term systems where you need plenty of data to test.

 

[qldfrog]

OK, are you guys backtesting a daily, weekly or monthly. Less data for daily, significantly more for a monthly system. Also, what's your average trade length to consider a fair backtest period.

Middle of the road weekly having an average trade length of 40 weeks. I guess no less than 5 years be OK and when do you test it, on latest data or during multiple market types?

Many ways and you probably already know but how about running an auto walkforward? Optimise for a few years then backtest, repeat etc etc. Then there's Monte Carlo plus that nice 3D optimiser image AB has. Try find yourself a flat plateau

I think trying all, so the confidence to place the cash is there.

Survivorship mentioned, don’t even bother without Norgates expensive top priced data, especially for the longer-term systems where you need plenty of data to test.

Sure, daily and weekly with short holding period as i do make systems less susceptible to flash crash..a bit, increase sample size on short period and should make system performing on the short term, with need to be dynamic on changed conditions
And no need for 20y of backtest

 

[othmana86]

Great conversations, so I think this is a good opportunity for me to showcase my system, for the start I will highlight the system over a 20 year period, 10 year period and a 5 year period, this is to satisfy both schools of thought. The first set will be w/out a 'take profit' and the second set will be with 'take profit'. You make the judgement. lets go..

w/out TP 20 year w/out TP 10 year w/out TP 5 years


now lets introduce the accidently discovered "Take profit" .. take note of 4 things 1. CAR 2. MDD 3. No. of trades 4. Win rate %

w/TP 20 years w/TP 10 years w/TP 5 years


Before i post my thoughts..id like to hear yours. have at it

 

[Skate]

Great conversations, so I think this is a good opportunity for me to showcase my system, for the start I will highlight the system over a 20 year period, 10 year period and a 5 year period, this is to satisfy both schools of thought. The first set will be w/out a 'take profit' and the second set will be with 'take profit'. You make the judgement. lets go..

w/out TP 20 year w/out TP 10 year w/out TP 5 years
View attachment 132855View attachment 132857View attachment 132858

now lets introduce the accidently discovered "Take profit" .. take note of 4 things 1. CAR 2. MDD 3. No. of trades 4. Win rate %

w/TP 20 years w/TP 10 years w/TP 5 years
View attachment 132859View attachment 132860View attachment 132862

Before i post my thoughts..id like to hear yours. have at it

@othmana86, gee those are impressive results, thanks for sharing. There is a lot of information to digest & personally, I would like time to compare the metrics over the three-time periods.

Also, so much generated from so little. The commission of $6.60 should have killed this strategy over time but it didn’t. Nah, something not right.

Just a few questions if I may
Is this a strategy that you are currently trading?
Has historical data been used for the Backtest?

Skate.

 

[othmana86]

Hi Skate
Why do you somethings not right?

Answers
Yes, currently live.
Not sure what you mean by historical data? But I’m using norgate..

 

[Skate]

Why do you think somethings is not right?

Think about this
In the last 5 years, your backest reports that you have made $800,091 or $176,018 each & every year from a $20k investment (from the last report). Nah, something not right.

Not sure what you mean by historical data?

It appears you need to do some research. This matter of (historical constituents) has been canvassed to death in this thread alone.

Skate.

 

[othmana86]

oh, yes i have used historical constituents. ive got a platinum membership with norgate

ok, i should explain. So all profits are reinvested with NO limits on position size. This is somewhat unrealistic and there is no way i will trade that way. eventually position size will be max @ 10k.

 

[Cam019]

I for one would be very interested in reading how you would suggest to others how they should go about accomplishing "cutting your losses quickly and letting your winners run".

@Skate, I think we are getting lost in translation here are little, so let me try and clarify my alternative point of view. My view of the world is that if I am trading a classic trend following strategy I want to "cut my losses short and let my winners run", apologies for the ear bleeding again.

That would mean using some sort of initial stop loss and then a trailing stop that follows price once the trailing stop is greater than the initial stop, as an example. I'm a simple man, I like simple trading systems.

I am not saying that a "cut your losses short and grab your profits" trading system cannot be profitable or have an edge in the markets. I am certain a system like this can be profitable - mean reversion strategies come to mind. My alternative view comes in the form of how strategies are backtested.

Everyone is entitled to express their view
Whether a poster is right or wrong has never been the underlining reason for encouraging others to make a post. Placing yourself in another person's shoes is how we learn to understand a different point of view. The only thing I would suggest, don't explain why you think others are wrong but rather express an alternative.

My alternative view is this:

When backtesting, I want to run the backtest over the longest period of historical constituent, suvivorship bias free data that I have (June 1993). I also want to make sure that all the backtest parameters in the system reflect the realistic trading of the strategy as closely as possible. Entries, exits, position sizing, position shinking, next day exits, etc.

Why?

Because I like to look back and see what has happened in the past, even though we all know that past performance is no indication of future performance - as system traders, it's all we have. It gives me confidence in what range of results I can expect when I am trading a system in the future.

Let me show you what I mean with an example based on the COVID-19 sell off. If I based my system expectations on a 7 year backtest, I get the following underwater equity curve prior to the COVID-19 sell off:

01/01/2013 to 31/12/2019


So my system expectation is that my maxDD is 17% and my avDD over that same period is 4.1%. Great, right? Not in my view. Because, and we are using hindsight bias here, what happens if we push the backtest finish date out to the 12/11/2021? Well we get this:

01/01/2013 to 12/11/2021


Now all of a sudden we are concerned because we have seen a drawdown that was outside of our backtested expected results. Most likely we are now tinkering with the system to try and reduce that heinous drawdown that we weren't expecting. However, if we actually tested our system parameters over as much historical data as we can get, we get the following:

01/06/1993 - 12/11/2021


So when the COVID-19 sell off happens (and don't get me wrong - it's not ideal, but drawdowns are a part of trading), I'm not shocked at the 38% drawdown because the system had a 44% drawdown during the GFC and the 38% COVID-19 drawdown is within the systems expected results.

This is why I backtest this way. I hope my alternative view has been helpful to some.

 

[qldfrog]

@Skate, I think we are getting lost in translation here are little, so let me try and clarify my alternative point of view. My view of the world is that if I am trading a classic trend following strategy I want to "cut my losses short and let my winners run", apologies for the ear bleeding again.

That would mean using some sort of initial stop loss and then a trailing stop that follows price once the trailing stop is greater than the initial stop, as an example. I'm a simple man, I like simple trading systems.

I am not saying that a "cut your losses short and grab your profits" trading system cannot be profitable or have an edge in the markets. I am certain a system like this can be profitable - mean reversion strategies come to mind. My alternative view comes in the form of how strategies are backtested.


My alternative view is this:

When backtesting, I want to run the backtest over the longest period of historical constituent, suvivorship bias free data that I have (June 1993). I also want to make sure that all the backtest parameters in the system reflect the realistic trading of the strategy as closely as possible. Entries, exits, position sizing, position shinking, next day exits, etc.

Why?

Because I like to look back and see what has happened in the past, even though we all know that past performance is no indication of future performance - as system traders, it's all we have. It gives me confidence in what range of results I can expect when I am trading a system in the future.

Let me show you what I mean with an example based on the COVID-19 sell off. If I based my system expectations on a 7 year backtest, I get the following underwater equity curve prior to the COVID-19 sell off:

01/01/2013 to 31/12/2019
View attachment 132875

So my system expectation is that my maxDD is 17% and my avDD over that same period is 4.1%. Great, right? Not in my view. Because, and we are using hindsight bias here, what happens if we push the backtest finish date out to the 12/11/2021? Well we get this:

01/01/2013 to 12/11/2021
View attachment 132876

Now all of a sudden we are concerned because we have seen a drawdown that was outside of our backtested expected results. Most likely we are now tinkering with the system to try and reduce that heinous drawdown that we weren't expecting. However, if we actually tested our system parameters over as much historical data as we can get, we get the following:

01/06/1993 - 12/11/2021
View attachment 132878

So when the COVID-19 sell off happens (and don't get me wrong - it's not ideal, but drawdowns are a part of trading), I'm not shocked at the 38% drawdown because the system had a 44% drawdown during the GFC and the 38% COVID-19 drawdown is within the systems expected results.

This is why I backtest this way. I hope my alternative view has been helpful to some.

We will agree to disagree.
my view is that
-if you have enough trades to be statistically relevant in your backtest,(we agree here);
Covid crash allows you a unique opportunity to create backtests from a few months before that event to present and so capture a condensed timeline of market conditions which do reflect current fundamental changes: qant and robot trading, super funds and overseas traders and money, retail internet trading and social media impacts.
Up to you to get that opportunity or discard it.
But i agree on monthly or low rotation/ long holding period, it will not be possible to get statistically relevant numbers and you will so be constrainted to either change that characteristic, or rely on obsolete to my view market pasts .
Note that I see the market covid crash as a blessing in that context as otherwise, i would have to follow your thought process.
I nevertheless will move to full..kind of...historical data and will run my systems thru.
Hope this explains clearly the 2 sides of the argument for whoever is new to this debate

 

[Skate]

@Skate, I think we are getting lost in translation here are little, so let me try and clarify my alternative point of view. My view of the world is that if I am trading a classic trend following strategy I want to "cut my losses short and let my winners run", apologies for the ear bleeding again.

@Cam019, I know what "cut my losses short and let my winners run" means & the process. Your post explained it for those who didn't. Also, you did a mighty fine job in doing so.

When you don't know you don't know
Recently I was explaining to a seasoned trader that we were buying a position that was under a scheme of arrangement "because the proposal does not reflect the significant valuation upside & is currently at a discount to industry peers based on EV/revenue or EV/FUA multiples". If I was talking to someone new to trading I would have used language they "would better understand".

My posts & post format
Sometimes my posts can be lengthy & exhausting to read but they are written for those who are on the lower level of experience. Even the format of my posts has been "ridiculed" many times publicly & privately. But in my defence, a bold heading with a short snappy passage in "my opinion" tends to hold a reader's interest over a short novel.

Skate.

 

[Newt]

oh, yes i have used historical constituents. ive got a platinum membership with norgate

ok, i should explain. So all profits are reinvested with NO limits on position size. This is somewhat unrealistic and there is no way i will trade that way. eventually position size will be max @ 10k.

Please forgive my backseat driving othmana86, but are you trading the All Ords (top 500) universe, and if so, do you have something along the lines of these 2 statements in your BUY conditions to ensure stocks are actually in the XAO at time of purchase?


NorgateIndexConstituentTimeSeries("$XAO.au") AND
NOT OnLastTwoBarsOfDelistedSecurity;