Welcome to the Traders Laboratory.

Money Management Risk and money management related topics.

Like Tree7Likes
Reply
Old 10-28-2015, 08:29 AM   #17

Join Date: Feb 2009
Location: Sheffield
Posts: 124
Ignore this user

Thanks: 6
Thanked 18 Times in 15 Posts

Re: sizing and 'stays'

Quote:
Originally Posted by zdo »
Noobs,
The status quo / the ‘voice of trading’ sugar coats it. They would have you apply very general rules to MM (see http://www.traderslaboratory.com/for...tml#post200085 for example)... and mix the "logic" of sizing and ‘stays’, etc etc. An alternative way to look at it is -
‘Stays’ (where to exit / “stop out” in adversity plus where to exit with profits) are best based on individual system stats.
Sizing is best done with portfolio (of systems) level stats.
Ie ‘stays’ levels are best derived from a different set of calculations from the calculations of optimal ‘sizing’ ------- and yes, that’s even if your portfolio of systems is only one system!

... and fwiw, imo your portfolio of systems should never be just one system
I'm totally in agreement about the distinctions that you make between portfolio and system based stats. Its a very good point, and its very good advice.

The points that I was trying, and probably failing to make were as follows:

Any "statistic" derived from a trading strategy needs to be more along the lines of rough estimates, rather than high precision statistics. I'm sure there are people out there using insanely complcated statistical approaches, but thats probably beyond the realm of most retail traders.

I'd also probably argie that the units used to derive any statistics are probably best based on some sort of market based values rather than points, or dollar amounts.

The main point that I was trying to make is that if you take a simple analysis tool like Software for Position Sizing Optimization : Forex Trading Systems : Stock Trading Online : Adaptrade Software and you attempt to optimize any MM strategy its simply a matter of fact that different strategies are going to perform differently at different times and in different market conditions. Over a diversified portfolio of systems, and enough time, things tend to average out, so you might as well go for an MM strategy thats easy to understand an impliment, and that you can actually trade (my example of optimal f hopefully illustrates the point that the optimum strategy isnt necessarily the best strategy)

The problem of course is that you dont know, what you dont know, and I am prepared to accept that there are systematic traders out their who's level of sophistication in the implimentation of their MM is way beyond mine, and that they are possibly able to dynamically optimise their MM approach based on market conditions, and that such an approach might potentially form a part of their edge.

I'm a system trader these days, and I tend to associate with other system trades who share similar views to my own, which is possibly shortsighted, but it cuts out a lot of extraneous noise. However I started out staring at a screen, and there are times when you are definately in tune and on a roll, and you know it. I never reached the point where I had quite enough confidence to push the gas a little harder during these periods, but I suspect that if I'd continued along that route, eventually I almost certainly would have done, and I know guys who regularly do that based on nothing more than intuition

However, on the other hand, in my limited experience, organisations who employ multiple traders tend to handle the risk aspects on a far higher level of abstraction than that that of an individual trader or individual strategy, and I suspect that they've almost certainly done a lot more work that I have on this stuff, so if they cant do it, who can.

the concepts of minimising the risk of ruin for a single system isnt rocket science, nor is the idea of diversifying risk across a portfolio of systems. What is rocket science, is the implimentation of those methods. I however would argue that the inherent randomness in returns pretty much invalidates all assumptions on which even the most complex of those models are built. Take 1000 guys trading a simple % fixed fraction, and a thousand quants implimenting the most complex of MM strategies, and after 30 years, I suspect the distribution in final PnL from the two populations wont be too disimilar

I'm not attempting to sugar coat anything, I'm saying there is no MM silver bullet, you have to take the rough with the smooth. Im not advocating that people shouldnt reseach these issues, I still spend a fair proportion of time looking at MM, despite previous reseach being completely fruitless, and I suspect that I'll continue to do so

I even make the point that you'll need a different approaches based on system metrics, but getting bogged down in unnecessary optimisation is in my experience not the best use of time
zupcon is offline  
Reply With Quote
The Following User Says Thank You to zupcon For This Useful Post:
zdo (10-28-2015)
Old 10-28-2015, 03:29 PM   #18
zdo

Join Date: Nov 2007
Location: boonies
Posts: 2,763
Ignore this user

Thanks: 422
Thanked 570 Times in 430 Posts

Re: 90% Analysis/10% Money Managment

off topic re 'report':
Well - that “2” was fun... and right back where we started - as usual...




back on topic...

zupcon, make the “final PnL from the two populations” very “dissimilar”

Quote:
... that they are possibly able to dynamically optimise their MM approach based on market conditions, and that such an approach might potentially form a part of their edge.
That is IT! Precisely.
Long ago I was lucky enough to realize that I needed to take a bassakwards philosophy / approach and MarketType (tmz) the auction first prior to any “analysis” and use that framework as a basis to fluctuate relative weights for each system across a portfolio of 12 (most of them very simple) automated systems ... with the intent of sizing each trade to the ‘odds’ of the opportunity.
Even if the exposure to many of the opportunities is relatively small, the ultimate functional purpose is to not miss ANY opportunities... which would be an operational impossibility for me to accomplish manually...

Now turning to ‘risk’ (... and I’m assuming we’re focusing here on the ‘trade itself’ type of risk across a series of trades)
re
Quote:
different strategies are going to perform differently at different times and in different market conditions.
and
Quote:
the optimum MM approach during period X is going to differ from the optimum approach in period Y. This inherant randomness in returns kind of makes any MM optimisation pretty much redundant
zup, while I hear what you’re saying about the ‘limits’ of MM and I understand why you’re saying it, from my experience, I can categorically recommend traders not be discouraged by the realities of “inherant randomness”. Rather, they should meticulously find and tune a system's ranges of “ the optimum MM approach during period X [which] is going to differ from the optimum approach in period Y” (brackets mine) and stay within them...
It would be better for a trader to even randomize his or her ‘MM’ decisions within those ranges than utilize vastly simplified ‘stay’ decisions or - worst - base them on ‘psychological ease’



kwikblurb1...
Again...
Base Sizing on opportunity ... algorithmically derive sizing from portfolio (of systems) level stats.
Base ‘Stay’ (or not - in profit or in loss) on ‘risks’ ... algorithmically derive them from individual system probabilities.

Kwikblurb2...
re“inherant randomness”
Fwiw, It is only randomlike - not really “randomness”. Big difference in the long run.


Kwikblurb3...
for sizing, Optimal f, etc is a starting point - nowhere near the finishing point

blurb4
re “I'd also probably argie that the units used to derive any statistics are probably best based on some sort of market based values rather than points, or dollar amounts.”
NONE of my systems have ever ‘optimized’ stays to dollar targets or to dollar stops. Almost all my systems ‘stay’ decisions are founded on ‘price action’ patterns... with some of them throwing in some ‘order flow’ stuff, some of them throwing in some ‘indicators’ work (WOT?!) (btw, nothing ‘off the shelf’ used... ), some of them throwing in some ‘scaling out’ work... etc etc.

Kwikblurb5...
The higher the hit rate for a system, the more closely ‘stays’ (ie stops and profit taking) must be finely calibrated. etc etc

blurb6
(Very very Generally) --- Re the Drawdowns allowed issue.
True trend systems can be allowed a 60% drawdown before shutdown.
Quasi - trend following systems can be allowed ~ 40% drawdown before shutdown.
Momentum/coattail systems (and the 'equally' successful strat of fading the system's signals ) can be allowed ~ 30% drawdown before shutdown.
Regardless of ‘testing’ results, just about all other systems should never be allowed to run past 20% drawdown before shutdown ... considering most of them are losers to begin with etc...







“And then the day came when the risk it took to remain tight in a bud was more painful than the risk it took to blossom.” Anais Nin
zupcon likes this.
zdo is offline  
Reply With Quote
Old 11-09-2015, 10:46 AM   #19
zdo

Join Date: Nov 2007
Location: boonies
Posts: 2,763
Ignore this user

Thanks: 422
Thanked 570 Times in 430 Posts

more sizing comments... fwiw

"If something is worth doing, it's worth doing badly." Leslie Buckland

Most system developers first turn to ‘ volatility’ as primary driver of ‘condition specific’ sizing... and get stuck there...
(btw, my advice = in sizing work, skip 'volatility’ completely ...you'll naturally be brought back to when and how to include it properly later ) ... and ...

Most of these developers turning to volatility as primary driver of ‘condition specific’ sizing get sucked into using measures of range (ATR, etc.) as a measure of volatility. A better (however more complicated) measure utilizes studies of passage through price brackets/zones ... number brackets traversed, time duration in each, etc...and ... btw

Most developers never realize the high reliability of cyclicity in properly measured ‘volatility’

A $million+ (c)TA offers this: Improving Trading System Performance Using a Meta-Strategy

A $billion+ CTA offers this: "I have a staff of PhD's researching volatility based position sizing. They haven't come up with anything."

" You cannot teach a man anything; you can only help him find himself." Galileo
zupcon likes this.
zdo is offline  
Reply With Quote
Old 11-11-2015, 10:22 AM   #20

Join Date: Dec 2012
Posts: 118
Ignore this user

Thanks: 11
Thanked 11 Times in 11 Posts

Re: 90% Analysis/10% Money Managment

70-30, trade small and more frequently, big bets will get you at the end.
sergso is offline  
Reply With Quote
Old 11-13-2015, 10:53 AM   #21
zdo

Join Date: Nov 2007
Location: boonies
Posts: 2,763
Ignore this user

Thanks: 422
Thanked 570 Times in 430 Posts

Re: 90% Analysis/10% Money Managment

Quote:
Originally Posted by sergso »
70-30, trade small and more frequently, big bets will get you at the end.
sergso,
"70-30, trade small and more frequently,... “ may be perfect advice for anyone trading the exact same system as you, on the same instruments, and with the same capitalization... etc

“70-30” may be correct --- but only for you, your financial state, and your system. There is no way you or anyone else can pigeon hole everyone else into that ratio. The usefulness of that advice is not ubiquitous even to someone trading a single method on a single instrument with single contract size... And past a certain system specific threshold of capitalization, the practicality of such advice goes completely microscopic. Keep going ... and past a certain threshold in the diversity / complexity of one’s portfolio, regardless of the ‘analysis’ methods used, flipping this 'advice' to “30-70” will not be nearly enough to manage dynamic sizing and allocations... especially with leveraging or in opm cases...etc etc. ... and ... At this point I won’t even go into the variability in the depth / time / etc. of ‘analysis’ between traders - besides to say it’s another reason pushing out a generalization about the correct or dominant ratio of analsis and money manglement is pitiful.

“trade small and more frequently” If limiting size and goosing frequency improved the performance of your trading that’s really great - but, come on man! - You can’t deal out advice like that as if it applies to all systems. It is a disservice to your peers ... at the very least a waste of their time. Traders need to trade as closely as possible the optimal size for their system(s). And a trader needs to trade at the exact frequency a system calls for without inserting trades (ie “more frequently”) or omitting trades (ie “I missed that trade” or “I didn’t take that trade”)... randomly / arbitrarily goosing or limiting frequency and / or size is just another loop ride on the loop around Looserville.

“big bets will get you at the end”. You may be shooting for some brevity with this one... so I’ll almost give you a pass -. Almost  ... still , let’s look at a couple of the words you used and some possible ‘presuppositions’
Re: the word “bets” - at a some point in each type of betting game, the “window closes” and from that point forward the results to your equity are completely out of your hands. In the trading game up until the time your broker liquidates your position(s), the ‘window’ stays open...with trades you still have possibility to influence the results to your equity.

And re: “big” The crucial variable here is not “big” or “small” (/ size). The crucial variable is positive expectancy.
so more accurately - and hopefully just as succinctly
If you have a positive expectation, the only thing that will get you is holding losses until they are “big”

I can’t think of a field where ‘general guidelines’ are less applicable than in trading - yet the ‘voice of trading’ continues to deliver ‘truths’ that are actually only true in ~1% of instances ...by the tens of thousands....media, forums, books, articles, seminars, etc etc...

sergso, almost all the good posters are gone now... it’s up to us average posters to do a better job...
zdo is offline  
Reply With Quote
Old 12-16-2015, 12:16 PM   #22

Join Date: Nov 2015
Posts: 6
Ignore this user

Thanks: 0
Thanked 1 Time in 1 Post

Re: 90% Analysis/10% Money Managment

I'd say 90% mental(emotions), 10% technical when live, 90% technical 10% emotional when in sim.
engulf is offline  
Reply With Quote
Old 12-17-2015, 06:32 AM   #23

Join Date: Feb 2009
Location: Sheffield
Posts: 124
Ignore this user

Thanks: 6
Thanked 18 Times in 15 Posts

Re: more sizing comments... fwiw

Quote:
Originally Posted by zdo »

Most system developers first turn to ‘ volatility’ as primary driver of ‘condition specific’ sizing... and get stuck there...
And ironically, many traders employing this approach end up taking their largest positions (and losses) at times of low volatility, and thats generally the time that the types of systems typically traded by the retail crowd tend not to do particularly well

Still trading "educators" will continue to peddle this crap, and people lap it up. You cant blame them really (the vendors that is)
zupcon is offline  
Reply With Quote
Old 05-12-2016, 11:15 AM   #24

Join Date: Apr 2016
Posts: 94
Ignore this user

Thanks: 0
Thanked 3 Times in 3 Posts

Re: 90% Analysis/10% Money Managment

If 90% is for analysis and 10% for money management, then what ratio can be allocated to emotions..?
aimhi is offline  
Reply With Quote

Reply

Thread Tools
Display Modes Help Others By Rating This Thread
Help Others By Rating This Thread:

Similar Threads
Thread Thread Starter Forum Replies Last Post
Follow the Smart Money / COT Analysis Dunstan Futures Trading Laboratory 244 10-17-2017 05:12 AM
What Should I Start with - Technical Analysis or Fundamental Analysis? phaynes1985 Beginners Forum 5 09-23-2014 09:58 AM
Are You Favor Fundamental Analysis or Technical Analysis? I Have Another One…. wmck6167 Trading Psychology 14 06-17-2013 04:50 AM
Fiat Money Tradewinds General Discussion 32 01-25-2013 10:59 AM
If You Didn't Need the Money Would You Still Trade? TradeRunner General Discussion 32 11-15-2012 11:35 AM

All times are GMT -4. The time now is 07:54 PM.
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
CS to VB integration by DeskLancer
©2006-2011 Traders Laboratory, All Rights Reserved.