Thoughts on "The A to Z of trading career - musings of a professional trader !!"

[madank]

You guys are much more informed and well-versed than me with backtesting methodologies and the parameters to look for but nevertheless, I would like to share my thoughts on it. Please be warned that the content of that post could sound amateurish to some folks but had to cover this topic in the series anyways (as I deem it indispensable to any trading related discussion).

Here's the link to the post -

http://www.traderji.com/community/t...of-a-professional-trader.105523/#post-1264151

 

[mindgames]

hi madan,

a forex trader recently mentioned that his institutional clients reject any model that has CAGR/MDD worse than 12:1. i couldn't digest the fact and thought that he was probably making it up. and then i bumped into your post that mentioned a thumb rule of 7:1 or better for intraday.

of course I am not questioning the correctness of your post!

considering that most trend following systems will have at most a 50% win rate, they will be profitable if the average win/loss size is higher. assuming a win rate of 40%, with avg win:loss in the ratio 1x:2.5x, how do we achieve an MAR (CAGR/ Max DD) as high as 7:1 (or better). this appears to be very difficult to achieve. Add leverage to the mix and it appears even more difficult.

In fact most managed hedge funds do worse than to 1:1 - https://www.managedfutures.com/top_cta_rankings.aspx

so, is money management and compounding the secret sauce? OTOH, smaller bet sizes = lower DD but also possibly lower CAGR!

would like to know your thoughts.

ps: i recollect we discussed this a few months ago but still find it difficult to wrap my head around the fact that some traders achieve such high CAGR/MDD performance (which also appears to fly in the face of higher risk = higher reward thumb rule).

 

[mindgames]

Hello Madan

I have read the book although quite a while back, will surely refer back to it.
If you have some specific script for "mental rehearsing in trading", please share it.

Thanks
Happy

Steve Ward's book "Tradermind" is good. It is like psycho-cybernetics for traders. Also contains an 8 week plan.

 

[madank]

hi madan,

a forex trader recently mentioned that his institutional clients reject any model that has CAGR/MDD worse than 12:1. i couldn't digest the fact and thought that he was probably making it up. and then i bumped into your post that mentioned a thumb rule of 7:1 or better for intraday.

of course I am not questioning the correctness of your post!

considering that most trend following systems will have at most a 50% win rate, they will be profitable if the average win/loss size is higher. assuming a win rate of 40%, with avg win:loss in the ratio 1x:2.5x, how do we achieve an MAR (CAGR/ Max DD) as high as 7:1 (or better). this appears to be very difficult to achieve. Add leverage to the mix and it appears even more difficult.

In fact most managed hedge funds do worse than to 1:1 - https://www.managedfutures.com/top_cta_rankings.aspx

so, is money management and compounding the secret sauce? OTOH, smaller bet sizes = lower DD but also possibly lower CAGR!

would like to know your thoughts.

ps: i recollect we discussed this a few months ago but still find it difficult to wrap my head around the fact that some traders achieve such high CAGR/MDD performance (which also appears to fly in the face of higher risk = higher reward thumb rule).

Hi mindgames,

Good to see you here !! Hope you are having fun in your vacation.

I mentioned that 7:1 for traders who are looking into compounding their account (esp intraday..it is difficult to realize it in swing due to overnight fluctuations). So, it entails that the trader's account size is small and would like to scale it up. But, this kind of ratio is not possible when the account size grows above 2-3 mil USD. If you look at the CTA's in the website you shared, their AUM runs in 100s of millions of dollars atleast. If they can accomplish this 7:1 or even 3:1, am sure their AUM will ‘go through the ceiling’ to 4-5 billion USD in couple of years. With larger account, it’s is not possible and risk mitigation takes the hot-seat when we grow in size.

Hence, as we move up in size, our returns will diminish for 2 reasons - lack of opportunities to deploy the fund and a predilection for mitigating the risk to bare minimum (that depends on the manager's comfort level and the nature of the instruments he's investing in).

Yes MG -lower DD will give us enough elbowroom to play around with MM. We cannot influence the trading parameters of our system but we can play around with MM and attain our anticipated/estimated goals. I just gave one way of managing MM and there are 1000's of ways to do it.

I also vibrantly remember the conversation in our chennai meetup and thanks to you for making that meetup happen We can discuss more on it once you come back to India !!

Steve Ward's book "Tradermind" is good. It is like psycho-cybernetics for traders. Also contains an 8 week plan.

Thanks for the recommendation. I did not know about this book. Sounds interesting !!

 

[ncube]

hi madan,

a forex trader recently mentioned that his institutional clients reject any model that has CAGR/MDD worse than 12:1. i couldn't digest the fact and thought that he was probably making it up. and then i bumped into your post that mentioned a thumb rule of 7:1 or better for intraday.

of course I am not questioning the correctness of your post!

considering that most trend following systems will have at most a 50% win rate, they will be profitable if the average win/loss size is higher. assuming a win rate of 40%, with avg win:loss in the ratio 1x:2.5x, how do we achieve an MAR (CAGR/ Max DD) as high as 7:1 (or better). this appears to be very difficult to achieve. Add leverage to the mix and it appears even more difficult.

In fact most managed hedge funds do worse than to 1:1 - https://www.managedfutures.com/top_cta_rankings.aspx

so, is money management and compounding the secret sauce? OTOH, smaller bet sizes = lower DD but also possibly lower CAGR!

would like to know your thoughts.

ps: i recollect we discussed this a few months ago but still find it difficult to wrap my head around the fact that some traders achieve such high CAGR/MDD performance (which also appears to fly in the face of higher risk = higher reward thumb rule).

@madank , I think mindgames is asking about the logic behind the DD:Return ratio and feasibility of such a system.

@mindgames , I understand your point, yes comparing DD with total returns is confusing and difficult to understand as they are independent factors.
However these ratios suggested by Madan are guidelines to measure the quality of trading systems.Though at first look these ratios seem
difficult to achieve, in reality its not, any positive expectancy system can easily meet this requirement.

Lets take the example of the system you mentioned which has an accuracy of 40% and RR of 1x:2.5x. Assume the risk we take for each trade is 10, i.e R:R = 10:25, and the system generates 100 trades in a year for easy calculation. At the end of the year the result will be as follows:

Total Win Points = 40 * 25 = 1000
Total Loss Points = 60 * 10 = 600
Total Returns = 1000 - 600 = 400

Which means after 100 trades this system will give a total return of 400 points.

Now lets apply this to our DD:Return requirement of 1:7 (200:1400)

To get 1400 points we need 400 * 3.5 = 1400, which implies the systems should generate 100 * 3.5 = 350 trades in a year.

As our trade risk is 10 points, we can have max 200/10 = 20 consecutive losses.

Hence this system will comply to the 1:7 requirement if it can generate 350 trades in a year and limit max consequtive losses to 20.

Also If one is trading swing/positional or want to keep the total trades as 100 in a year, the same can be achieved by the following systems:
1. R:R=1:2.5 & Accuracy = 70%
2. R:R=1:5 & Accuracy = 40%

To conclude if the system has good expectancy/accuracy and generates enough trades in a year with good R:R then one can easily get good DD:Return ratio.

Hope your doubts are clear now.

 

[headstrong007]

@mindgame, thanks for the name of the book, it's uncommon to me and pretty interesting.

I want to share one uncommon but useful book for trading psychology from my collection as a return.
*********
Below book can improve traders trading psychology. Trading is actually speculation business which has many similarities with the game of poker...

Zen and the Art of Poker: Timeless Secrets to Transform Your Game

It's not costly. Only Rs 442.. Worth reading...

https://www.amazon.in/Zen-Art-Poker-Timeless-Transform-ebook/dp/B000OCXHOK

*********

 

[mindgames]

@madank - thanks for the kind words! you are a major +ve influence on my trading. i am looking forward to the meetings too!

@ncube thank you ncube for the math. i'm embarrassed by my laziness to put forth a sample distribution as an eg. while raising the issue! they should make more people like you and madan

i don't completely disagree with your example. nor do I have a solution to the problem raised.

here are the actual results from what we can say is a good system (ignoring charges though!)....

total trades: 101 - over 18 years. (actually 121 but there were 20 break-even trades. so, let's ignore them)
win:loss count = 61:40 (approx 60%) - this is very good for a trend following system (to be honest i cherry picked this one because i wanted to prove myself wrong)
payoff (avg. profit: avg. loss) = 27,400 : 10,276 (~2.7). (because the break-even trades are ignored, this is low. if break-even trades were to be considered (as amibroker does), the payoff will rise to ~4.
max consecutive losses = 5; max consecutive profits = 6

since this is a good system, let's trade it aggressively with 1 lac --> compounding after every trade so that profits are reinvested eg. invest 10% of total capital (i.e. start with 1o,000 to trade 1st signal. it made a loss of 638 --> next trade will be with 9,362 (10k - 638)....so on and so forth.....

at the end of 121 trades, 1 lac becomes 13.6 lacs --> not bad eh? CAGR = 16%. Max DD? 15% --> CAGR/Max DD = 1.04!!

few observations:

1/ max DD = draw down from peak. however, for simplicity i have computed draw down based on the capital remaining after a trade is closed. possibly the actual draw down will be more than this.

2/ above eg. considers just 121 trades. obviously risk of dd and risk of ruin will shoot up with an increase in the no. of trials.

3/ since we are reinvesting profits, after a point we come close to investing the entire capital. (entire capital = original 1 lac + cumulative profits). this may appear foolhardy but then compounding is what makes such high returns possible in the 1st place.

4/ increasing the original capital (i.e. reducing the fraction of capital invested) will lower the MAR further. OTOH if the fraction is increased MAR will increase. in the current case it increases to a max of 1.4 (31% : 22%).

5/ i'm also attaching an excel file with the trade list so that we can play around with the nos. if required. perhaps i'm making some elementary mistake.

i'm not nitpicking or trying to prove a point - just thinking out loud....

cheers

 

[ncube]

@madank - thanks for the kind words! you are a major +ve influence on my trading. i am looking forward to the meetings too!

@ncube thank you ncube for the math. i'm embarrassed by my laziness to put forth a sample distribution as an eg. while raising the issue! they should make more people like you and madan

i don't completely disagree with your example. nor do I have a solution to the problem raised.

here are the actual results from what we can say is a good system (ignoring charges though!)....

total trades: 101 - over 18 years. (actually 121 but there were 20 break-even trades. so, let's ignore them)
win:loss count = 61:40 (approx 60%) - this is very good for a trend following system (to be honest i cherry picked this one because i wanted to prove myself wrong)
payoff (avg. profit: avg. loss) = 27,400 : 10,276 (~2.7). (because the break-even trades are ignored, this is low. if break-even trades were to be considered (as amibroker does), the payoff will rise to ~4.
max consecutive losses = 5; max consecutive profits = 6

since this is a good system, let's trade it aggressively with 1 lac --> compounding after every trade so that profits are reinvested eg. invest 10% of total capital (i.e. start with 1o,000 to trade 1st signal. it made a loss of 638 --> next trade will be with 9,362 (10k - 638)....so on and so forth.....

at the end of 121 trades, 1 lac becomes 13.6 lacs --> not bad eh? CAGR = 16%. Max DD? 15% --> CAGR/Max DD = 1.04!!

few observations:

1/ max DD = draw down from peak. however, for simplicity i have computed draw down based on the capital remaining after a trade is closed. possibly the actual draw down will be more than this.

2/ above eg. considers just 121 trades. obviously risk of dd and risk of ruin will shoot up with an increase in the no. of trials.

3/ since we are reinvesting profits, after a point we come close to investing the entire capital. (entire capital = original 1 lac + cumulative profits). this may appear foolhardy but then compounding is what makes such high returns possible in the 1st place.

4/ increasing the original capital (i.e. reducing the fraction of capital invested) will lower the MAR further. OTOH if the fraction is increased MAR will increase. in the current case it increases to a max of 1.4 (31% : 22%).

5/ i'm also attaching an excel file with the trade list so that we can play around with the nos. if required. perhaps i'm making some elementary mistake.

i'm not nitpicking or trying to prove a point - just thinking out loud....

cheers

@mindgames , Got it...the problem in your calculation is that you are considering CAGR:MAX DD, which is not correct as CAGR measures the returns on an annual basis. However the ratios mentioned by Madan are for Total Returns: Max DD.

When backtesting we are more interested to measure how the system behaves for the entire test sample. Hence we should measure the Total/Absolute Returns:Max DD.

If we apply this to your example system the ratio comes to around 1360%:15% ~= 90:1 or if we consider 10k as initial capital then its 900:1 ...which is definitely a holy grail

 

[mindgames]

@mindgames , Got it.......the problem in your

hmm.... that explains. comparing lifetime returns with a one time event of max drawdown?! interesting.

 

[ncube]

hmm.... that explains. comparing lifetime returns with a one time event of max drawdown?! interesting.

My friend, max draw-down is not a one time event, it is the worst case scenario (sequence of consecutive losses) for your system as observed in the historical test sample. This will help you make a rough estimation of how your system may perform in future.

I am not saying it should not be done but If you want to use lifetime CAGR return then better compare it with the lifetime average annual draw-down... so that the measurement metrics are at same scale.