Money Management - the art of the exit

Nader,

Agree with you 100%.

Great to see you posting again.

Best wishes for 2011,
David

Well so much for no losing trades in 2011....
Or maybe you meant you won't be trading in 2011.


Loose vs Lose
The words loose and lose are mixed up in writing; for some reason, many people write loose when they really mean lose. But there's no reason to lose your mind worrying about this, just lose the extra o!


Loose

Loose is an adjective, the opposite of tight or contained.

My shoes are loose

I have a loose tooth

There's a dog running loose in the street


Lose

Lose is a verb that means to suffer the loss of, to miss.

I win! You lose!

Don't lose your keys

I never lose bets


The Bottom Line

Simple carelessness leads people to write loose when they mean lose. Just remember that lose has one o, and loose has two. Start with loose, lose an o, and what do you get? Lose!

Thanks Hanover Happy 2011 to you as well and all the best.


Regards,

Nader

Good explanation.

It surprises me the number of folk who can't spell 'lose'. I know this is merely a forum, hence correct spelling and grammar really isn't important, and typos are inevitable, but it takes more keystrokes to mis-spell the word (as 'loose').

No big deal; it's just a trend that I've noticed.

Hey Hanover.
Just havin a little fun.... nothing personal, I know your work.
Trade well and a prosperous new year to you.

okay some interesting stuff here for starters :-


Trading IS NOT Random, it's not a 50/50 coin flip, if it is your doing it wrong.


The only way to enter is with the current direction, we'll say UP, no we can't predict when the direction will change to down, but as it's up go Long and wait and see how far it goes.

Therefore your exit should be, when it's no longer going up, doesn't matter if your in profit at this time, if the market turns save your pennys and exit.

It really is that simple, in essence doing above it tricky enough, most people are trying to predict the turn down and going short in a upward market then they wonder why they lose, it looks easier and safer but it's just suicide.

Trading is really very simple, it's the self discipline that's tough.

And your right it is not all that random, price carries momentum and that is our edge.

Take this EUR/USD 50 day moving average vs price cross system for example. Since September of 2005 it has made 5,467 points, and that is without any, none, nada, money management, it is always in the market and simply changes direction when price closes over the MA.

Now look at the number of losses compared to the number of wins. Could you trade a system like this? You can see it is profitable. BUT do YOU have the discipline to take trade after trade with these strings of losses?

51 losses, and 19 wins...... max consecutive losses, 11.... max consecutive wins 2.

Now what sort of MM would you use to increase the profit of this simple system?

Here is the link.
50 day moving average vs price

Hanover, I would like to see some of your thoughts on this one.
I guess you could say this is a very good test to use MM to optimize your exits.

Tradestar,

For me, the key is not whether the system can be improved by adding an exit strategy, but whether the underlying concept is robust. This is a simple trend following system: if it catches enough long trends to offset the frequent losses that come from the sideways moves, it will win; otherwise not. IMO it's as simple as that. So the question for me is whether the behavior of EUR relative to USD from Sep 2005 until now is 'typical', and hence likely to continue into the future. More on this, and also exits, later in this post.

From a purely statistical angle, I would not feel comfortable trying to draw any kind of conclusion from a mere 70 trades; hence the underlying robustness becomes even more important. Nonetheless, if I wanted to stay within the statistical realm, I'd backtest the results over a similar time period, e.g. Jan 1999 thru Sep 2005, and see if the results are approximately similar. If so, then I'd feel a little more comfortable.

Another key is whether the MA period of 50 is the result of curve fitting. So I'd test with different MA values (e.g. 30,40,50,60,70,....) over both time periods (Jan 1999 thru Sept 2005; and Sept 2005 thru to today), and see if any patterns emerged. If, for example, one test showed that SMA(30) was BY FAR the most profitable, then I'd be a little concerned, especially if the other test showed SMA(70) as being by far the most profitable. Ideally, if the underlying concept is robust, there shouldn't be too much dependence on getting the numbers exactly right; but, undermining this conclusion we again have the problem that a (small) sample of 70 trades is unlikely to prove anything conclusive.

I had a quick look at the Jan 1999 thru Sep 2005 EU D1 chart, and there did seem to be some long trends between some of the crossovers, which is encouraging.

Currencies that are negatively correlated trend well. In other words, if XXX's economy tends to be strengthen when YYY's is perceived as being weak, then we have negative correlation and some good strong trends by pairing the two against each other. So the question is: when banks and governments are trying to sell USDs because the US economy is apparently weakening, or for whatever other reason, do they tend to buy EURs, or look elsewhere? And vice versa? What does history tell us in this regard? Is this a phenomenon that's likely to continue moving forward?

Conversely, currencies that are positively correlated tend to move sideways for prolonged periods, making them less suitable for trend following. As an example, I dislike USDJPY for this reason.

OK, now on to exits. I say "exits" rather than "MM", because to me, they are both different things (MM is more related to position sizing, staking systems, capital preservation vs maximizing equity upside, etc). Re exiting of losses, the crossover provides an implicit SL, as the SMA(50) catches up with positions that move against you relatively quickly. You could try testing a X-pip SL to see how effectively it cuts losses even more quickly, but my guess is that it would choke as many winners as it would cut losers (and since the winners are few, and large, this would likely have an overall detrimental effect). And also a 70 trade sample, IMO, is too small to prove anything conclusive. You could try a simple 'price action' type SL, e.g. exit a long position if price falls below a significant support level (and vice versa for shorts). That at least has the backing of a sound TA/orderflow principle, arguably making the small sample size less significant.

Re profit exits: given that the system depends on its few BIG winners, the emphasis must be on letting the winners run, hence I'd be wary of the (popular) idea of using a tighter SMA to exit trades (their rationale is that you give back less profit as price retraces to the SMA). As before, you could test this, but keep in mind the limitations of a 70 trade sample. Again, you might be able to use some simple 'price action' techniques, by looking at candle-implied weakness in a trend, and/or maybe scale out when heavy S/R might curtail a trend. But again, this works against the key idea of letting the winners ride as long as possible. It's my guess that there's no easy answer.

Re "MM", I don't see any advantage in moving beyond something like a 1%-2% fixed frac per trade. I would size conservatively because trend following systems are vulnerable to long losing streaks, and fixed frac makes it more difficult to recover from large drawdowns.

I'm not a mechanical trader, hence I don't spend a lot of time testing, and experimenting with, these kinds of systems. I believe that systems work profitably because of robust underlying principles, rather than an a high dependence on getting the numbers right. However, many quants might rightly disagree with this, and I happily concede, because their knowledge of math/statistical methods/optimizing systems is vastly better than mine. So I am probably not the best person to answer your question; but you did ask me, and I've given it my best shot.

Good luck,
David

I haven't studied the trades you mentioned so this suggestion is just off my head.

I would assume that you are using a fixed position size for all trades. In order to think of optimizing the MM to be used to increase the system profitability, I would assume you will have to think of something logical related to the concept of the system.

What if you try adding more to your position after the price moves X amount of pips in your direction??? Determining the X amount of pips will have to be determined by you. You can even assume 3 or 4 different variations and test them to see if they do have an effect on your system results or not.

Just my 2 cents...


Regards,

Nader

I wish you all a good and profitable () 2011 !!!

As I see it (my point of view) MM can be explained with a simple analogy:

I have several options to go from New York to Miami (yes, a long ride...), but I have only certain amount of gallons in my tank...if I am on the correct highway I can drive economically or speed away (therefore burning more of my precious fuel). How to use my actual fuel is MM.

I know, some will say that this example is wrong, because you multiply your bank account in trading, whilst in my example you only burn fuel (which is what some traders do), but it is just to understand that MM is to treat your money carefully.

I've had a great trade placed for the EURUSD, the direction was right, the timing was ok, but i entered with too many lots and couldn't ride the wave and the retracements (had to exit prematurely). This is like being on the correct highway, but speeding so much that my fuel consumption was too high and couldn't take me to my goal.

I've looked a little more deeply into the equity curve generated by your crossover system. The equity plot is attached below.

The system remains in drawdown for more than 39 months, between Sept 2005 (point 'A') and Dec 2008 (point 'C'). Then it encounters 13 profitable months from point 'C' onward (Dec 2008 thru to today). The final outcome is essentially the result of 5 big winners (> 700 pips each) during the latter period.

Hence, for me, some key questions are:

a) Which of the two periods is likely to represent the more likely scenario for EURUSD henceforth?

b) Could you, from (i) a psychological standpoint, and also (ii) if this crossover system was your primary income source (i.e. trading for a living), survive a period of 39 months with a negative result?

c) Given the statistical limitations of the 70 trade sample, is there enough evidence to discount the 5 big winners from being completely untypical, or 'black swan' type events?

Revisiting your question about MM: I just realized that my suggestion of using fixed frac is unworkable, since there are no fixed SLs. Here is an alternative suggestion; it is a method used by a friend at our local TA group, who is likewise always in the market (long or short), and uses the reverse signal as an implicit loss exit. It's also in keeping with Nader's idea, as it adds to positions that are moving in your favor (as the account balance rises).

1. Calculate the worst maximal drawdown in the testing to date. With your crossover system it is 1460.4 pips, reached at point 'B', from the then previously highest point ('A').

2. Multiply this value by an arbitrary safety factor. Let's say we choose 2. Hence 2 x 1460 = 2920. This is the 'cover' used in subsequent calculations.

3. We divide our account balance by this value, and then by $10 per pip (per full lot traded on EURUSD), to get the number of lots we may trade. Let's say our account balance is $10,000:
10,000 / 2,920 / 10 = 0.34 lots

4. Each day we perform the same calculation, using the equity figure (i.e. including unrealized P/L) to get a new value. Let's say our account balance has risen to $12,000:
12,000 / 2,920 / 10 = 0.41 lots
So now we have enough cover in the account, to add another 0.07 lots to the original 0.34 lots. Open a second position for 0.07 lots
(Hence this is an aggressive pyramiding strategy: as price moves increasingly in our favor, we have enough cover to justify increasing the total position size.)

5. When we get a signal to reverse the position, say from long to short, we simply close our long position(s), and perform the same calculation to size our new short.

6. If the account balance falls, DON'T reduce the position size. That's because there's enough cover in the account to survive 2 x the worst drawdown to date. However, see points 7 and 8.

7. Keep running tabs on the worst maximal drawdown (current low to highest prior peak). If it gets worse than that calculated at point 1, we repeat the calculation to obtain a new cover value, and downsize our position accordingly.

8. If the account balance falls below the original 10,000 divided by the safety factor, i.e. 10,000 / 2 = $5,000, then we have insufficient cover to continue trading, hence we abandon trading the system. In other words, if the safety factor we chose was 2, we are abandoning ship should 50% drawdown be reached.

MM doesn't make you profitable, it keeps you in the game losing at a rate you can afford to top up so constantly bleeding money into the market.

If your system doesn't offer 80%+ Win rate in theory don't bother, as REAL is harder than theory and even 80%+ theory is hard to get 50%+ actual.