Entries and Exits Don't Really Matter

[Mathemagician]

Why is it that most traders spend nearly all of their time chasing after a better set of entries and exits? (Heck, most focus squarely on the entries!) In reality, you can use pretty much any modestly stable set of entries and exits and achieve results that would please any trader whose goals are properly aligned.

In order to illustrate this I've attached two equity curves. They both use an embarrassingly simple entry/exit methodology found in a book that was published in 1998 (just a daily breakout). It's literally a two-line system, one for long entry and one for short.

The first EC is just the plain old system applied to a portfolio of 65 global futures markets using 1 contract each. The second EC is the identical set of entries and exits on the identical set of markets but with a halfway decent position sizing method applied. (Costs applied are $115 per round-turn.)

You'll note that the testing period is 25 years long with the last 10 years having occurred after the publication date. Notice there is no degradation in performance after the publication date. With position sizing turned on, this is a 21% compound annual growth rate with a maximum daily peak-to-valley drawdown of just 25%. This result is definitely competitive with the top CTAs and money managers for this period!

Once you outgrow the B.S. presented by the make-1000%-a-month-with-95%-winning-days hypesters and work out what a 21% CAGR means to you, you quickly realize that this type of performance is more than you'll ever need and the drawdowns associated with it are quite comfortable. Why do I say this? Well, here's a VAMI table showing the results of a 21% annual return compounded over a career...

That's right, after 10 years you've got a 670% return on initial capital, and after 20 years you've got a 4500% ROIC. Best of all, during this time the worst drawdown you've endured is 25%.

Why aren't more people trading this way? Why are so many people chasing the perfect entries? It's a mystery to me, but I'd sure love to hear your reasons...

jj

 

[dcraig1]

One reason is that it takes a fair bit of dosh to trade 65 futures markets and even more to trade all those markets with position sizing applied.

Can't do that in a 1000 quid fx account. (not that I'm advocating fx).

 

[dbphoenix]

Also because we're actually trading rather than just running numbers

 

[Mathemagician]

One reason is that it takes a fair bit of dosh to trade 65 futures markets and even more to trade all those markets with position sizing applied.

Can't do that in a 1000 quid fx account. (not that I'm advocating fx).

Fair point, but there really isn't much one can do in a 1000 quid account at all! Oh, one can try, but the reality is that nearly any sensible trading plan will be underfunded due to the size of the stops or swamped by the additional relative costs imposed by those who cater to accounts of this size.

Also, while more markets is better, it's not necessary to hyper-diversify. I've seen this type of model applied to 13 market portfolios with success. Further, this approach works about as well with individual equities and one can trade a basket of equities effectively in a small account.

jj

 

[Mathemagician]

Also because we're actually trading rather than just running numbers

A fair question, indeed, but this is not simply an untested theory. I make a living trading just like this on my own behalf and on behalf of clients, both retail and institutional. This is a talk that I most definitely walk.

jj

 

[Crap Buddist]

Time. Some people want that return in months, weeks, days? Whose to stop them ?

Not Magicians, alchemists. Make David Blane look like he's a thunderbird puppet when pulling his tricks.


Look into my eyes not around my eyes but into my eyes.....

Think of any even number between 50 and 100, numbers have to be greater than 5 & no two numbers can be the same.
































68.

Now think of the market like a trick.....

 

[dcraig1]

Fair point, but there really isn't much one can do in a 1000 quid account at all! Oh, one can try, but the reality is that nearly any sensible trading plan will be underfunded due to the size of the stops or swamped by the additional relative costs imposed by those who cater to accounts of this size.

Also, while more markets is better, it's not necessary to hyper-diversify. I've seen this type of model applied to 13 market portfolios with success. Further, this approach works about as well with individual equities and one can trade a basket of equities effectively in a small account.

jj

I agree with all that and it always suprises me that there is so little discussion of trading a basket of equities which can be done in a fairly small account. Grey1 is really the only poster here that does give it proper attention.

 

[dbphoenix]

A fair question, indeed, but this is not simply an untested theory. I make a living trading just like this on my own behalf and on behalf of clients, both retail and institutional. This is a talk that I most definitely walk.

jj

I'm sure you do. Nonetheless, the various theories and models and so forth that academics come up with have nothing to do with the trading that ordinary traders engage in. Sells books and services, I suppose, but has no relevance to the bottom line.

 

[nine]

Now gentle people ... tolerance to other's ideas please, lets explore rather than challenge (too) hard ... if I can only get one good thing from everything Mathemagician says I'll be happy. And I think I'll end up with more than one.

FWIW, IMO, you need a minimum in the 100k to 200k range to trade an end of day futures portfolio properly. 10-13 contracts will give good diversification and its hard to get much beyond that anyway because thinness, slippage, and in some cases a tendency to nasty locked limit days and rapacious floor brokers (may they all disappear) make them far less profitable than they appear on paper.

 

[dbphoenix]

No arguments, nine. It's just not relevant. That's why I don't pursue this stuff anymore. Had enough of it with Tom and David Gardner back in the late 90s.

A question was posed. I responded. Done.

 

[Mathemagician]

I'm not sure what long/short futures has to do with buy-and-hold-forever equities, but you're obviously entitled to your opinion.

jj

Edit: Besides, this time it's different!

 

[Crap Buddist]

Also, I don't heat soup in a microwave because it impairs the flavour. It matters to me. Ditto baked beans, and scrambled eggs, and rice etc..

Then again some people don't care. Choice or ignorance ?

Hmm there was a quote or sig or here about its just eggs and omelettes or something but some chefs can't cook a simple omelette because the chef counts, its still used as a basic assessment of chefs too I believe.

So its not just eggs and omelettes, cookie counts too..

 

[Mathemagician]

FWIW, IMO, you need a minimum in the 100k to 200k range to trade an end of day futures portfolio properly. 10-13 contracts will give good diversification and its hard to get much beyond that anyway because thinness, slippage, and in some cases a tendency to nasty locked limit days and rapacious floor brokers (may they all disappear) make them far less profitable than they appear on paper.

You're right about the min requirement for a futures system, if not a bit understated. However, the solution to producing a realistic backtest for an EOD futures system is no different from doing it for any other kind of system -- estimate costs accurately and when in doubt overestimate them. Experience has shown that $100/rt for slippage and $15/rt for commissions is an overestimate (rollovers included) and can routinely be beaten in reality. Commissions/fees are lower than ever, floor brokers are gone/irrelevant, and the odds of a purely locked limit day occurring on the same day you have an order to fill are small. For that matter, pure locked limit days are easy to build in when coding your system. If h=l then no executions that day.

jj

P.S. I definitely encourage critical thinking and divergent opinions. All I ask is that those whose opinions diverge maintain the expected level of decorum. FWIW, I've got no complaints so far in this thread whatsoever.

 

[Lightning McQueen]

I agree with all that and it always suprises me that there is so little discussion of trading a basket of equities which can be done in a fairly small account. Grey1 is really the only poster here that does give it proper attention.

I do believe the blades also trades stocks in this way, hopefully if he has time he may stop by this thread, I'm sure it would interest him greatly.

 

[alexander]

The first EC is just the plain old system applied to a portfolio of 65 global futures markets using 1 contract each. The second EC is the identical set of entries and exits on the identical set of markets but with a halfway decent position sizing method applied. (Costs applied are $115 per round-turn.)jj

Very interesting. It could become more interesting if you could, provided of course you have the time, to post the P/L per futures contract at the end of the testing period.

Questions:

(1) Did you take into account limit moves in the testing. For example, did you prevent the system from exiting a long position during a limit down move and a short position during a limit up move?

(2) Did you scale the number of contracts for equivalent risk exposure in the different markets?

(3) What was the maximum margin deposit required and the maximum intraday drawdown of the tested system in all markets combined?

Thanks,

Alex

 

[Mathemagician]

Alex,

I never really concerned myself much with the P/L in each individual market. There are markets that do very well and there are markets that do very poorly (even over the entire 25 year period). No attempt is made to chose the best markets. The position sizing methodology is designed so that the daily return stream from each market is statistically indistinguishable from any other, and it does accomplish this goal regardless of the entry/exit methodology employed. To answer your questions...

1. Yes.
2. Yes.
3. I've no idea what the maximum intraday drawdown is for a system like this, only daily close-to-close. There's really very little point in watching your p/l intraday as you receive no actionable information. Margins change all the time and I am not aware of a good source of margin data. I do know that the present margin required is about 8% of capital.

Hope that helps!
jj

 

[StratOpt]

Even if it were not relative to me personally at this point in my trading career, I would hope that it would affect me at some future point in my trading career and I could then take full advantage of the benefits of doing such a thing. Learn now to profit later. I also know to listen to those who are actual and not all academics. Those who know do, and those who don't teach. We have all heard this before. In the case of Mathemagician you get both the know and the teach in one so you can bet that I am listening and taking notes. I guess I would lean towards the Boy Scout motto of Be Prepared.

 

[alexander]

3. I've no idea what the maximum intraday drawdown is for a system like this, only daily close-to-close. There's really very little point in watching your p/l intraday as you receive no actionable information. Margins change all the time and I am not aware of a good source of margin data. I do know that the present margin required is about 8% of capital. jj

Intraday drawdown is very important because it determines capitalization requirements. Many GBP futures traders got liquidated intraday when the Quantum fund took positions against the Bank of England in the early 90s because they were not properly capitalized.

If I'm not wrong the returns you stated are based on an initial equity of $1,000. If this is the cases, this is nowhere near to the capital levels required to trade a system like that. Note that the formula to calculate the starting equity is given by:

E = maximum margin ever posted + maximum intraday drawdown ever

As an example, if the system had at a given time 10 open positions, the total margin would be anywhere from 20K to 50K (assuming no big S&P trades). Then, if the maximum drawdown was something like 25% you would need a minimum of 25K to maybe 100K to start trading a system like this.

Alex

 

[Mathemagician]

Intraday drawdown is very important because it determines capitalization requirements. Many GBP futures traders got liquidated intraday when the Quantum fund took positions against the Bank of England in the early 90s because they were not properly capitalized.

If I'm not wrong the returns you stated are based on an initial equity of $1,000. If this is the cases, this is nowhere near to the capital levels required to trade a system like that. Note that the formula to calculate the starting equity is given by:

E = maximum margin ever posted + maximum intraday drawdown ever

As an example, if the system had at a given time 10 open positions, the total margin would be anywhere from 20K to 50K (assuming no big S&P trades). Then, if the maximum drawdown was something like 25% you would need a minimum of 25K to maybe 100K to start trading a system like this.

Alex

No, that's my fault. I was unclear and I apologize for the misunderstanding. VAMI is a standard calculation performed on a theoretical indexed starting balance of 1000. Intraday drawdown is irrelevant at the levels of leverage we're discussing here. The margin requirement for this model right now is about 8%. That means if you've got a $1m account you need to post $80k for margin.

jj

 

[Chorlton]

Mathemagician,

Interesting Thread....

...and the odds of a purely locked limit day occurring on the same day you have an order to fill are small. For that matter, pure locked limit days are easy to build in when coding your system. If h=l then no executions that day.

Can you elaborate on what Locked Limit Days are?

Also, what would you recommend as the minimum starting capital to trade such a system as the one you mentioned?

Thanks,

Chorlton