Random Entry & Perception

[meanreversion]

DT, you're missing the point slightly. Your argument against the possibility of a profitable random system is elegant but flawed (if a random system can't exist because everyone would use it and then the market would change --- you could apply this argument to any trading system to "prove" it doesn't work).

However, whether your argument is specious or not is irrelevant. The "random" system they have tested is not really that random, as I pointed out in my previous post.

The point is this --- random ENTRY coupled with specific sizing and specific exits can be profitable, THUS sizing and exits are important, far more so than people realise. That is the point of the exercise.

 

[bevok]

I see your point DionysusToast and you're absolutely right, although to be fair they only claimed to be testing random ENTRIES, not a random strategy to the markets per say. The 'trading volatility' element came from the volatility based exit i.e. ATR based. The same is true for exits as well of course. You can test some entries with a random exit and make money as well. In this case you're exploiting things like prior technical indications of momentum, rather than the fact that random exits are profitable.

It is true that Van Tharp's final conclusion (that exits are most important and that entries don't really matter) is incorrect, and shows why we should never accept others opinions or interpretations uncritically. His experimental evidence was skewed because he did not reverse his hypothesis (i.e. test random exits).

 

[BSD]

Hi Markus,
To accuse DT of being substance free is unfair and untrue in the extreme. To suggest he's clueless really does you no favours at all

Hi Tim,

sorry but I totally disagree.

Him claiming TA doesn't work, or that "pros" don't use charts is complete, total and utter BS, particularly as I keep posting evidence to the contrary.

Have you ever seen him post a positive contribution, ie not his constant unfounded this doesn't work, that doesn't work as per nothing else than his unproven opinions ?

I haven't, any more than the Expert whom he admired so much.

And that really says it all imo, falling for that prize clown.

And as I keep saying also I back up my stuff with objective hard facts as anything else is really just comedy central which unlike say forexfactory is occurring here much too much imo.

 

[BSD]

Ok now I know I've made it, I've been quoted on t2w. Happy days!

I just brought up my "random" system and it's still producing some nice results. But there are some things I need to make clear

1. The level of randomness is limited, as it applies only to the entry. Adding a trend filter significantly increases profitability, BUT of course makes the system less "random".

2. The objective of coding up this type of system (trend following but with coin flip entry) is really to explore the value of position sizing and exits. I've read through the posts on this thread and there is discussion about what constitues "random" and other (quite deep) mathematical discussions. I am not particularly interested in that.

For me, it's more this ----- position sizing and exits are incredibly important, BUT are the least talked about. My "random" model enters on a coin flip, but then scales (pyramids) in in a specific fashion, and exits in a specific fashion. Most threads about trading are on where to buy (e.g. "buy BP here?") but position sizing and exits are rarely discussed.. why is that?

If 90 pct+ of retail traders fail (surely the numbers should be 50:50 or 40:60 if you allow for commission, bid/ask etc., but certainly not 10:90 or 5:95) is there a reason for this? Is it because of general lack of awareness of position sizing (which also incorporates an element of money management) and exits?

Good stuff mate

 

[BSD]

Like I said, nothing but comedy central:

Toast says:

Regular/Regurgitated Technical Analysis

You know MACD divergence, MA xovers, Stochastics - all of those common entry techniques often discussed in books on trading despite the fact they don't work.

I think this covers anything you read or learn in 99% of books & courses.

When the reality is:

Commerzbank have a fund that has all of one parameter for entering and exiting trades...

The trusted MACD indicator.

And the fund is profitable.

"Deshalb hat die Capital-Redaktion
einmal nachgerechnet und verschiedene Timing-Strategien
über einen Zeitraum von 13 Jahren gegen den DAX ins Ren-
nen geschickt.Am besten schnitt dabei dieMACD-Strategie
ab.Wer die Signale dieses Indikators konsequent umsetzte,
konnte in den vergangenen 13 Jahren mehr als das Sechs-
fache*aus seinem Geld machen."

http://www.comdirect.de/pbl/static/pdf/corp0075.pdf

Grew your money six-fold over the last 13 years which ain't bad for a public fund.

Course MACD don't work tho according to the knowledgable Toast right !

Good for the guys who grew their money 6 fold that they didn't read Toasts musings.

"Superfund's origin dates back to 1991, when Christian Halper and Christian Baha developed a software system for the technical analysis of financial data. Today, the group has more than 1.5 billion dollars under management from more than 55,000 retail and institutional investors."
http://www.superfund.com/
Technical analysis:
Trading without emotions
Sound managed futures funds like Superfund funds are based on proprietary, fully automated technical trading systems. A vast range of technical indicators and historical prices are analyzed by the computerized trading systems to automatically generate buy and sell signals.
http://www.superfund.com/HP07/Superfund_Trading System.aspx

Or the simple trend following approach that has made Billions for it's propagators with really simple systems:

http://www.turtletrader.com/it.html

Presumable grownups pretending at having insights into the world of professional trading when nothing could be further removed from the truth while being in complete denial about obvious, perfectly undeniable objective hard facts have no place on a trading board that wants to attract and retain pros imo !

 

[DionysusToast]

DT, you're missing the point slightly. Your argument against the possibility of a profitable random system is elegant but flawed (if a random system can't exist because everyone would use it and then the market would change --- you could apply this argument to any trading system to "prove" it doesn't work).

However, whether your argument is specious or not is irrelevant. The "random" system they have tested is not really that random, as I pointed out in my previous post.

The point is this --- random ENTRY coupled with specific sizing and specific exits can be profitable, THUS sizing and exits are important, far more so than people realise. That is the point of the exercise.

Good point - all I wanted to show was that the system was exploiting a certain profile of volatility and was not random.

Actually - I would change one of your statements...

"random ENTRY coupled with specific sizing and specific exits can be profitable, "

to

"random ENTRY coupled with specific sizing and specific exits to exploit certain market conditions can be profitable whilst those market conditions are in play"

It is the markets they chose thet caused the profit every bit as much as the techniquesthey applied after the exit. Whether they realise that or not we'll probably never know.

 

[DionysusToast]

BSD - may I humbly suggest that if you have a point to make, then you start a thread about it.

This conversation is very much about randomness and perception. Your posts are about your opinions of me. Whilst I am most flattered that you take so much time thinking about me, why not start your own thread where you can give this matter the attention you think it deserves ?

You can, if you wish, start a poll about banning me from the forum. I am sure this would be most entertaining.

 

[BSD]

Lol toast what you're still not getting or pretending to is that Meanreversion ran NEW backtests to clarify the point to himself, hence he'd hardly be fiddling the data lol !

Same like this guy who ran NEW tests that were profitable 100% of the time:

"As Anthony suggested I run 100 tests of the random entries-trailing stops at 10 ATR from 1982 until September 2007 on a portfolio of 69 futures with $100 for commission and slippage and the system made money 100% of the time as you can see in the following figure."

http://www.tradingblox.com/forum/viewtopic.php?t=3637&postdays=0&postorder=asc&start=0

This is only about your twisting facts to fit your fictionalized version of reality and doing my part to rebut the patent falsehoods you post such as making a generalized claim that TA don't work Bro just coz it don't for you.

This is a traders board and one really can't let such a wilful distortion of reality pass unchallenged.

 

[bevok]

I think we're talking at crosspurposes here. I think DionysusToast has acknowledged that the various tests quoted were not 'proving' that random trading worked, but rather random entries. What made those random entries work was having an effective exit. We could also run a similar test with a good entry (we could try a MACD ) and a random exit - these also can work with an effective entry. The key is the 'effective' entry or exit. The example from Trader's Roundtable works because he is using a 10ATR stop exit. The original test with a 3 x ATR stop doesn't work now.

The conclusions:
- systematic trading can work, based on sound market principles.
- markets do change - at the moment the 1992 random entry experiment exit isn't profitable.
- entries (and exits, and position sizing, and trading psychology etc etc) on their own aren't everything and don't deserve the disproportionate attention they receive

 

[scose-no-doubt]

IMO this whole discussion is pointless. It's descending into personal attacks etc and tbh it's just silly. Your whole perspective on the market depends on the approach you take. You can't tell a quant that the markets aren't a collection of random variables any more than you can tell Raj Rajaratnam that he was just having a mad punt on the stocks. If you're approach is making you money then your approach and therefore your hypothesis and perception is right. If we can all agree that the markets don't operate on a random walk basis on the longer term (LTCM! lol) the only thing that should be considered is how right you are. This can be measured by hit rate and profit per contract.

 

[MrGecko]

There is a reason that random strategies don't make money - it's simply because they can't. When I say they can't - I mean it is against the laws of mathematics and in a way physics.

You are right DT, but for the wrong reasons IMO. I shall try to explain what I mean through a little thought experiment...

Let us suppose we have a trading strategy that goes Long on a head and Short on a tail. Trades are left until either stop or limit is taken out, when the coin is thrown again and a new position is taken accordingly. Stops and Targets are set equidistant from the entry point (i.e. R:R = 1)

Can this system make money?

Yes, of course it can. If the strategy is deployed where one outcome is more likely than the other (i.e. the probability of an up move is greater than the probability of a down move in a bull market), then the strategy has a positive expectancy. This, however, is because the population set (the market in a bull run) also has a positive expectancy. The edge in the strategy is merely transposed from the edge in the market.

The edge in the market exists because asset prices are limitied by the boundary zero. If you don't lose leverage, you can't lose more than you put in, but of course you can double your money and more.

The point here is that the trading strategy doesn't add any alpha of its own*

What about exit critera?

Well, lets consider another thought experiment.

Same coin tossing rules, and firstly, I set the R:R at 1:1 (portfolio A)

We can expect Portfolio A to have 50% longs 50% shorts. If the market has the property that an upmove is more likey (say, p(up) = 0.6) then I can expect to make money with this system (with the caveats from before). For this portfolio the expectancy is +0.1R.

Now, can I make this system more profitable by changing the R:R?

I set the R:R to 1.5:1 (portfolio B) and run the same system on the same data as portfolio A. It also hase 50% longs and 50% shorts. It MUST be true that the probability of a winner falls to match the expectancy of the first strategy (in this case P(up) = 0.44).

Why?

Because otherwise I could run the two strategies alongside each other except in opposite directions, and recieve a risk free profit for my efforts. No arbitrage, innit.

 

[BSD]

Unfortunately just one lil problem - well 4 actually, lol -, with whatever all that is supposed to mean or show let alone prove is that it completely ignores this lil ol objective, hard and backtested fact here.

Same like this guy who ran NEW tests that were profitable 100% of the time:

"As Anthony suggested I run 100 tests of the random entries-trailing stops at 10 ATR from 1982 until September 2007 on a portfolio of 69 futures with $100 for commission and slippage and the system made money 100% of the time as you can see in the following figure."

http://www.tradingblox.com/forum/viewtopic.php?t=3637&postdays=0&postorder=asc&start=0

I guess why all these trend following Billionaires got as rich as they are is coz they learned to keep it simple, and to accept that a duck is a duck when it looks, talks and walks like one without needing written confirmation from the great spagheti monster in the sky.

While people are wasting time coming up with ever more arcane, osbcure and complex beyond belief theories as to why MACD or TA or random trading or whatever doesn't work cause that's just a feeling they have others have garnered together a couple of facts based on simple insights on what's truly success relevant, done a couple of backtests, gotten on with it, and started making money.

Any one who tells me trading ain't simple - if not clearly always easy - really doesn't get the success relevant factors and is also in denial of what provably works for a great number of successful market participants.

 

[montmorencyt2w]

Markus,

How about (probably in another thread) you doing a little test:


See if you can grow some arbitrary starting amount sixfold using only entries and exits based on MACD.

You can choose the starting amount, but it should be a real money account.


Fair enough?

 

[BSD]

Can provide you with that right away here:

"Wer die Signale dieses Indikators konsequent umsetzte,
konnte in den vergangenen 13 Jahren mehr als das Sechs-
fache*aus seinem Geld machen."
http://www.comdirect.de/pbl/static/pdf/corp0075.pdf

That's a Commerzbank fund btw, Germanys second largest bank after Deutsche.

Oh and no I wouldn't trade using only the MACD, most importantly because I believe that it only works on long aka daily time frames and above.

 

[MrGecko]

Can this system make money?

Yes, of course it can. If the strategy is deployed where one outcome is more likely than the other (i.e. the probability of an up move is greater than the probability of a down move in a bull market), then the strategy has a positive expectancy. This, however, is because the population set (the market in a bull run) also has a positive expectancy. The edge in the strategy is merely transposed from the edge in the market.

...

We can expect Portfolio A to have 50% longs 50% shorts. If the market has the property that an upmove is more likey (say, p(up) = 0.6) then I can expect to make money with this system (with the caveats from before). For this portfolio the expectancy is +0.1R.

Sorry but this argument here is a bit of a brain fart.

Of course the expectancy of a 50% long 50% short strategy with R:R = 1 is zero.

No arbitrage argument still applies though.


EDIT: In order to arrive at the first situation (i.e. +ve expectancy), you would need to introduce some NON-random filter into the system. For example, if you knew that P(up)>(Pdown), you could amend the heads = long, tails = short strategy to a long only strategy, which is why I make the point about asset prices being bounded by zero.

 

[BSD]

"Wer die Signale dieses Indikators konsequent umsetzte,
konnte in den vergangenen 13 Jahren mehr als das Sechs-
fache*aus seinem Geld machen."
http://www.comdirect.de/pbl/static/pdf/corp0075.pdf

That's a Commerzbank fund btw, Germanys second largest bank after Deutsche.

That Commerzbank fund uses all of one parameter for entries / exits.

Larry Hite is another extremely rich mechanical, technical trend following automated trading CTA who averaged out at 30% p.a for years and years and is the guy who made MAN into the giant - biggest money manager in the world - they are today... (trader Daily, october 2008)

Ditto for fellow CTA Bill Dunn, who has all of two (!!!) parameters driving his mechanical, computerized trading model. (Trader Daily, june / july issue 2006)

So it really can be that simple, decades upon decades of success aren't a fluke in the real world of hard facts, and where people have an interest in making money, and not sounding of on message boards trying to appear clever by posting unproven, complex and obtuse assertions that fly completely counter to what the world shows to those willing to open their eyes and minds.

 

[Hilarymannah]

"where people have an interest in making money, and not sounding of on message boards trying to appear clever by posting unproven, complex and obtuse assertions that fly completely counter to what the world shows to those willing to open their eyes and minds."

BSD,..Well said that man !,...

 

[meanreversion]

Wie komme ich am besten zum bahnhof? That's the only German I remember.

This thread is skirting around something fairly fundamental here. I posed the question earlier, but no-one really responded.

They say that "90 pct of retail traders lose money". How is this possible? If the bid/ask on everything was zero, you would imagine 50 pct would make money, 50 pct would lose, right?

Given we have commission and bid ask, maybe 40 pct make, 60 pct lose? Possibly 35/65? But no, it's 10:90 or even 5:95.. how can this be?? It simply doesn't sound credible. This means if that everyone just sat there tossing a coin (what a bunch of t*ssers lol), the success rate would be HIGHER than what we currently have, which is 90 pct of people losing money ----- doesn't anyone find this astonishing? In fact it is hard to imagine a scenario in which more people lose money than they already do.

So ---------- why does this happen. I think (just my opinion of course) that the desire to want to be right is one of the main issues. Traders want at the minimum a 50 pct success rate or even higher, when this is not necessary to make money.

The desire to be right also leads to taking profits quickly, and running losses (can't admit to being wrong). This is fatal, I believe.

There is also desire for excitement, and this leads to focus on the entry.. after all, nothing more exciting than researching a stock and then buying it. (In addition, the more trading, the more fun, BUT of course you're giving back more in bid/ask and commissions)

In short, I think we are hard-wired to be poor traders, and I know I'm sh*t at discretionary trading, which is why I've ended up developing a mechanical trend system.

Ok, OF COURSE there are times when trend models get chopped up. BUT there are times when discretionary traders get chopped up too.

Forget win rate. Forget picking highs/lows (another desire to be smarter than the market - but it means trading against the trend, statistically likely to lead to losses). All that matters is developing a system with positive expectancy (this is why the "Holy Grail" thread of medbs is instructive - his only interest is 100 pct winning trades, and he has blown his account a couple of times over).

Trend systems depend on kurtosis (this is your line BSD). It depends on trends developing, which a normal bell shape distribution does not indicate, or "fat tails" as they are called. This in turn depends on human nature remaining more or less constant.

You can go back hundreds, even thousands of years and find big trends, whether in real estate (Pompeii had a massive property bubble before the volcano blew, I am not kidding) or flowers (Black Tulips) or computers (dot com boom of 2000).

When human nature changes and we all become sane and rational, trend following will cease to work, as will random entry systems.

 

[bevok]

Yes the main thing discovered in various investigations of traders results was that they don't close out their losing trades, and trade against the trend (see The Futures Game p 309 for example). Witness the number of people in different threads on this site taking long trades in the Euro. Talk about shovelling sh*t uphill!

Of course if random entry did work, 90% of people would give up after five losers anyway - a simple system like that could never make money, could it? ;-)

 

[MrGecko]

MeanReversion,

What I think has happened on this thread is that two arguments are being held in the same discussion.

Your argument seems to stem from the premise that most traders make irrational decisions (cutting winners, keeping losers). This is a valid argument (from a valid premise) to make, but IMO is not the most appropriate here.

I feel that a more appropriate argument w.r.t the OP is the logical inference of why a random trading system will only ever yield random results. To understand the argument, One will need to have an understanding of elementary calculus/maths/probability, but it is not so tough for average intelligence buggers like me to get. This is the argument that I have tried to articulate, clearly I have not done very well. That does not however jeapordise the legitimacy of the argument.

Moreover, the discussion (IMHO) has been disrupted by posters such as BSD, who draw conclusions from evidence based arguments that are far from concrete (quoting Market Wizards is not a panacea for reason), and then resort to ad hominem arguments when these original arguments are challenged.

Much Love

G

P.S. I believe it is not kurtosis but skew that trend following systems rely on, and the success rate of "coin tossing traders" would infact be 100% including costs.