Search
× Search
Tuesday, October 4, 2022

Archived Discussions

Recent member discussions

The Algorithmic Traders' Association prides itself on providing a forum for the publication and dissemination of its members' white papers, research, reflections, works in progress, and other contributions. Please Note that archive searches and some of our members' publications are reserved for members only, so please log in or sign up to gain the most from our members' contributions.

Is draw down the wrong way to measure performance?

photo

 Sander B., Independent SAS consultant and Systematic Algorithmic trader (MT4)

 Wednesday, August 6, 2014

Drawdown is often used as a way to measure performance of a strategy that trades in only one instrument. The risk per trade is often based on the expected relative drawdown. But I wonder if that is the right way of doing it? First of all, sometimes it is used in the wrong way. A number of times I’ve seen people mention that they have a 10% drawdown for instance, and then say that they use a constant position size. In that case, RDD doesn’t mean anything. Let’s say someone risks €1000,- on every trade. An account balance of €10.000,- will result in a completely different RDD compared to a €50.000,- account balance. RDD is only meaningful if you risk a percentage of your account balance per trade. But secondly, RDD doesn’t account for randomness. Let’s look at two examples of 9 trades. Every trade either wins or loses 1% of the account balance. Example A has the following pattern: 2 winning trades, 1 losing trade, 2 winning trades, 1 losing trade, 2 winning trades and 1 losing trade. Example B has the following pattern: 5 winning trades, 2 losing trades and 2 winning trades. Example A has a better RDD but Example B has a better end result. Now which of the two has the best performance? I get that looking at RDD is a way to stay in your comfort zone, but it is also rather arbitrary. Eliminating your feelings is one of the reasons to automate trading in the first place.


Print

5 comments on article "Is draw down the wrong way to measure performance?"

photo

 Sander B., Independent SAS consultant and Systematic Algorithmic trader (MT4)

 Tuesday, August 12, 2014



@Guy: So what you are saying is that you are willing to keep on trading with the risk of your account going to zero. If I had taken that approach I would have lost all my money about 18 months ago. Luckily I got out at the 50% RDD and walked away with about 100% profit. Yes, that particular strategy failed big time, and it is only now that it is starting to show some promise again.


photo

 Guy R. Fleury, Independent Computer Software Professional

 Tuesday, August 12, 2014



@Sander, do the math. For over 5 years now (since March 2009), the general US market trend has been up. That's over 1, 300 trading days, where you could have put money on anything reasonable and won. So, why did you stick around so long in a losing proposition?

My trading methods accumulate shares for the long term, therefore, it is built in the methodology that I will stick around with some positions still opened during down trends. As a matter of fact, almost all my trading positions will show some red. My programs will restart accumulating shares on rebounds. Having a long term objective (20+ years), coupled with an over diversified portfolio, I know that over the long term the strategy will win as if by default. Maybe in part for the simple reason of just having stuck around. As Mr. Livermore once said: “...It never was my thinking that made the big money for me. It always was my sitting. ” And I must say that I have acquired, over the years, some skills in designing “sitting”, or better said holding functions.

Now, if one position was down 50%, in the presented tests, this would represent a: 1.56% (0.50*(1/32)) portfolio decline. So, not a big deal...

However, such a stock would be under scrutiny with one big question: will it survive? If the answer is no, then I'm out; otherwise, I'll stick around and maybe acquire more (well, more like letting the trading script do its job). But note that that particular stock might not have been chosen as part of the data set in the first place due to the long term view of things.


photo

 Tibor Komoroczy, CEO & Founder, Skunkworks LLC

 Tuesday, August 12, 2014



I have no idea how everyone invests. I can only speak from reality and from 32 years of market experience. Guy you would be fired at any hedge fund that I have dealt with. No PM is allowed to swing money wildly. I really don't know how other people work. I work for Steve Cohen for 6 years as a consultant to his personal book. And in 2008 he fired 33% of SAC. And it was clearly for losing money. Sander risk management super seeds everything and if you have a book of 100 stock positions, you are very much at the mercy of the market. I have no idea how a person would ride a position 50% against them let alone 10% against them. All I know is I would be fired in the world I live in so would everyone else. You can't be crazy with money because you have a great story the reality check is the P&L. Risk mangers and great hedge fund mangers deal with money management first and stories second. Sizing and risk management mean more than any investment story. Guy what Sandler is interested in is how your theory worked in 2007-2008. Over 20 years out you're playing with fire you will experience massive draw down just in market correction and bear markets. Problem is running money Guy you have to put it to work in all market conditions and what is your man date says you must have x invested at all times or you don't get to run that money. There is so much more to running money than a back test of 20 years of a scaling model that takes large losses. And nothing is none sense for any one here they are all smart folks so please give that language a break. Mr Buffett isn't the same type of manager as mangers who use quantitative methods he is much more dangerous. Patrick you can quantify market participant globally that isn't a hard thing to do. And each investment inside any fund is an investment unto itself. You don't keep a bad positions because you're good ones are offsetting it you kill the bad ones. A loss never bothers me after I take it. I forget it overnight. But being wrong and not taking the loss--that is what does the damage to the pocket book and to the soul. Of all speculative blunders, there are few greater than selling what shows a profit and keeping what shows a loss. Try to be nice here this is about automation not buy and hold that is in some other forum. Gross exposure and net exposure figure out that into any trading model you build.


photo

 Pankaj Sharma, Portfolio Manager at inCurrency Business Solutions

 Wednesday, August 13, 2014



@ Guy, It trivial to gloss over a 50% drawdown in a 50 year test, you have the benefit of sitting in the future!! On real money, in real life, as you don't have the benefit of knowing the future, abnormal drawdowns will not do your mind any good. And as Tibor said, in the real world, your customer will starting asking questions at 10% and move out at 20%!! If you are into value investing or have multi-year holding periods, AND there is conviction on the underlying business, then its acceptable to hold onto higher drawdowns, but not on mathematical models!!

On Sander's original question, i like to allocate a capital to a strategy = (1+RDD)*exposure + some buffer. My daily reporting calculates drawdown, sharpe. If the position is leveraged, the drawdowns are higher, but then one would also expect higher period returns. I find drawdown to be a useful measure.


photo

 Guy R. Fleury, Independent Computer Software Professional

 Wednesday, August 13, 2014



@Tibor, @Pankaj, I don't want to enter into a debate of what is an acceptable drawdown as this would vary and depend entirely on each individual's taste, resources, trading methodology and downright quasi randomness of stock prices. For sure, no one wants to suffer any drawdown at all. You are not playing the game to lose.



@Tibor, SAC had its own solution to this. It made a lot of money over the past 20 years using an undisclosed technique known as insider trading; very little downside to this one. It's “a” trading method as long as you don't get caught. Of note, SAC paid 1.2B in fines and penalties. It was also forced to go private, which can explain a lot of the layoffs.



A small portfolio can be flipped almost on a daily basis for a profit or a loss. But it is another matter altogether when a portfolio grows in size. And it will have a chance to grow if you give it time. It is not instantaneous gratification.



But no matter what trading methods one uses, he/she will face drawdowns on almost every trade made. The size of which can't easily be forecasted, otherwise why buy before your forecasted low?



So drawdowns you will see. The point is: how much is acceptable to you? Some of those major drawdowns can happen in a single day... often overnight... Five 10% stop losses in a row is less than a 50% clip, but nonetheless, it will reduce a portfolio to about 59% of its previous value.



A bag holder, just as a trader, had to survive the Dot.com bubble and the financial crisis. One by holding on, and the other by trying to compensate on a short term basis. But in both cases, on average, they had a 50% haircut during the financial crisis; and I do mean on average. At least, that is what you will find just by reading the major market indexes. Even Mr. Buffett had a 50% drawdown during that period. At the end of the day, all shares are in someone's hands.



To me, it is not if 50% drawdowns will happen, they just will. And as said, it's your trading strategy that has to determine prior to its execution what you will do with these inevitable drawdowns. It is your automated trading strategy that has to deal with the problem. It's therefore one's responsibility to program their trading scripts to do what they want them to do.



I make my automated trading strategies obey the following payoff matrix equation:



A(t) = A(0) + Σ(H(1 + r + g + T)^t.*ΔP)



which translates to: accumulate shares over the long term (20+ years) and trade over the process. This will build a portfolio of stocks over the long term and use the trading profits to accumulate even more shares. So the shares on hand will follow to the penny the price movements of each individual stock. And therefore, drawdowns there will be.



Maybe a more sobering note would be the following:



http://blogs.marketwatch.com/thetell/2014/08/13/1-chart-shows-just-how-badly-average-investor-lags-even-cash/



which shows the performance of the average trader over the last 20 years.

Please login or register to post comments.

TRADING FUTURES AND OPTIONS INVOLVES SUBSTANTIAL RISK OF LOSS AND IS NOT SUITABLE FOR ALL INVESTORS
Terms Of UsePrivacy StatementCopyright 2018 Algorithmic Traders Association