[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[amibroker] Re: Do all trading systems stop working? - Howard Bandy's book



PureBytes Links

Trading Reference Links

Sidhartha,

FWIW

I can give you confirmation on a couple of points.

Some experienced traders can, and do, move away from purely algorithmic trading ... IMO it is a natural progression, but I think it is very dependent on our temperament e.g. I actually quite enjoy programming but I do tire of the continual challenge of solving the logic/syntax and wrestling with the rulebook, so I end up not programming every little variation, and nuance, or even every idea.

In fact, while I like programming, I don't love it ... there is a difference.

I can give you a specific example on how our temperament shapes our natural style .... Z and I recently discussed writing some code for Median() .... as per my usual style I took another route and started the journey to Median via RalphVinces equations (based on Pearsons Skew) and I set out to do it via array processing ... I got stuck on variance because that requires looping .... I need an array method of calculating variance ... Tomasz lent a hand by reminding me that I could read into the code, behind the functions, using TradingRecipes (it is online for several languages) ... I appreciate the source, in case I need it, but I didn't read it .... I just set out to make up my own 'easy' method ... doing the thinking in Excel ... last time I looked I seem be close to a solution....

.... but I don't recommend my style to everyone.


You can see a working example of different temperaments at work in this discussion ... Howard used the BT and opt to analyse MA(C,10) and I used conceptual analysis (I looked at charts, with an MA(C,10) plot on it for a few hours ... I haven't tested an MA for 5 years ... didn't really need to because Faber et al have done that for us ... if anything I would only need to test some variations on the theme ..... while there was some difference in the detail we both came to the same conclusion (from a traders perspective) i.e. that buying straw hats in winter pays off.

So that there is no misunderstanding:

- I do a lot of thinking prior to implementing a system design
- I attempt to reduce things down to the simplest and most likely explanation (Occams razor?)
- I carry out extensive and repeated objective tests .. often I repeat the tests over and over or do the same test in a slightly different way (I just don't do many of them in the BT) ... my testing is mainly used to elucidate the philosophy that is behind the trade
- I only use what I have confirmed by testing
- I ended up with a basic model of Core Market Behaviour and a relatively simple approach, which tends to produce simple systems
- because they are simple I can trade them without a computer (except for the brokers platform)
- I continually practise trading my system(s) == mental rehearsal (like practising a golf swing) and before a session I will 'grove the swing'
- I can make 'discretionary' adjustments during the game, however, they are minor adjustments... in golfing terms I don't change my swing but if the wind is blowing a bit harder than usual I might change clubs.
- at the pointy end I believe in KISS and rote learning, which helps keep the focus.


Re random walks:

I can confirm I have done a lot of work studying random behaviour, in general, and the random nature of the markets in particular.

Yes, I do use it in trading and yes, systems that assume the market is random can produce results (I call myself a random walker with a limp!)




.

--- In amibroker@xxxxxxxxxxxxxxx, "sidhartha70" <sidhartha70@xxx> wrote:
>
> I've moved away over recent years from purely mechanical systems, to more rule based systems with a discretionary element. I've alluded to why in previous posts... mainly that I simply don't believe some of these systems can be programmed effectively (or more precisely it's not worth the time required to do so and they would be prohibitively slow once programmed in their entirety)...
> 
> It's my firm belief over what I've learned the past few years, that a good trader with the required psychological attributes to be a good trader (i.e. he has spent time ironing out the kinks & weakness in his psychological makeup as it relates to trading), with a sensible post entry plan in terms of managing his losses & profits can make money over time from purely random entries...
> 
> There has been some work done on this - confirming it - (not by me) but I believe Van Tharp talks about it in 'Trade your way to financial freedom'....
> 
> In a nutshell purely mechanical systems are very vulnerable to any change in the 'expectancy' of the system.  Take the expectancy towards 50% and way before it gets there (due to bid offer spread & commissions) it will lose money consistently.
> 
> A good trader (one of the 10% that actually makes money rather than losing it) is far less vulnerable to changes in the expectancy of his system because it could approach 50% and he would still likely be able to make money.
> 
> 
> 
> --- In amibroker@xxxxxxxxxxxxxxx, "wavemechanic" <fimdot@> wrote:
> >
> > The comment referred to a mechanical system so the "WHY" is a nonissue.
> > 
> > ----- Original Message ----- 
> >   From: brian_z111 
> >   To: amibroker@xxxxxxxxxxxxxxx 
> >   Sent: June 07, 2009 9:20 AM
> >   Subject: [amibroker] Re: Do all trading systems stop working? - Howard Bandy's book
> > 
> > 
> >   >Extraordinary claims require extraordinary evidence".  Where is the published evidence? 
> > 
> >   There will be no proof that "trading the edge erodes the edge" forthcoming because there is no central record of WHY any recorded market transaction is made.
> > 
> > 
> >   --- In amibroker@xxxxxxxxxxxxxxx, "wavemechanic" <fimdot@> wrote:
> >   >
> >   > Without getting into endless mental gymnastics, consider a simple mechanical system - crossing of moving averages.  This system has been known and used for "ages".  How well does it work?  Perhaps as well as it did on day one or perhaps not.  In either case, a critical statistical analysis for a reasonable number of stocks over various test periods is needed to prove or disprove.  Has this been done?  As Sagan famously said, "Extraordinary claims require extraordinary evidence".  Where is the published evidence?  Absent such evidence, everybody will do what works, or feels OK, or etc. and to a first approximation no amount of discussion will change a single mind.
> >   > 
> >   > Bill
> >   >   ----- Original Message ----- 
> >   >   From: Dennis Brown 
> >   >   To: amibroker@xxxxxxxxxxxxxxx 
> >   >   Sent: June 05, 2009 11:02 PM
> >   >   Subject: Re: [amibroker] Re: Do all trading systems stop working? - Howard Bandy's book
> >   > 
> >   > 
> >   >   Yuki,
> >   > 
> >   >   You are spot on.  I know that traders who have consistent winning  
> >   >   strategies are tracked by their brokers, and in some cases the brokers  
> >   >   shadow their trades.  They do not even have to know their algorithm,  
> >   >   just place the same orders in a shadow account.  All brokerages that I  
> >   >   know of have the ability to make these shadow accounts.
> >   > 
> >   >   However, that does not mean that all will do it.  Some brokers pride  
> >   >   themselves in not taking the other side of their customers trades, or  
> >   >   doing anything that could be considered a conflict of interest.  They  
> >   >   are known as fill and bill brokers.
> >   > 
> >   >   One trader I know has been contacted regularly by his brokerage house,  
> >   >   asking for his methods.  Many trading houses look for consistent  
> >   >   winners and offer to let them trade the house money for 50% of the  
> >   >   profits -- and the trader is not responsible for the loses --except he  
> >   >   would lose his job.
> >   > 
> >   >   However, I believe the biggest threat to the "edge" will come from  
> >   >   machines that can out pattern recognize, out compute odds, and run  
> >   >   emotion free.  The machines are getting there, and I don't want to  
> >   >   take the other side of their trades when they do.
> >   > 
> >   >   Also any good algorithm that becomes public will be put into a  
> >   >   machine, and as long as it works, it will drain the profits out of the  
> >   >   trade.  It does not even require a "Big Boy", just a bunch of little  
> >   >   traders will kill it since the machine trading will be additive across  
> >   >   machines.  As long as the machine is making money, who would unplug it?
> >   > 
> >   >   In the mean time, trade on.  Life is short, and we might not live to  
> >   >   see that day anyway.
> >   > 
> >   >   Best regards,
> >   >   Dennis
> >   > 
> >   > 
> >   >   On Jun 5, 2009, at 10:16 PM, Yuki Taga wrote:
> >   > 
> >   >   > KM> Why would it be discovered?
> >   >   >
> >   >   > I would be inclined to believe that any system that is employed for
> >   >   > any reasonably lengthy period of time will be discovered.  I think
> >   >   > this is particularly true now in the data processing age.  Human
> >   >   > beings are, after all, human beings.  And behind all the machines,
> >   >   > there are human beings.  You can't trade without exposing yourself to
> >   >   > the machines (which "remember" all your trades forever) and, very
> >   >   > importantly, to the people who have access to the machines, or who
> >   >   > control the people with access.
> >   >   >
> >   >   > I don't know where this might be illegal or legal, and I'm sure it is
> >   >   > in some places and maybe isn't in others, but if I was a ranking
> >   >   > officer in a brokerage firm, you can be absolutely sure that I would
> >   >   > know exactly who my most profitable clients were over time -- using a
> >   >   > basket of metrics to look for outstanding performance that fell
> >   >   > within allowable risk parameters.  And you can also be sure that I
> >   >   > would spend no small amount of time and effort trying to ascertain
> >   >   > how any sustained profitability that was in the bounds of my metrics
> >   >   > was being generated.  I'd be running the data periodically.  Need I
> >   >   > say more?
> >   >   >
> >   >   > If you are siphoning money out of the market on a consistent basis,
> >   >   > and doing it better than almost anyone else (basis simple RoR, better
> >   >   > risk-adjusted numbers, some the combination of the two, or whatever
> >   >   > measures you happen to be looking for), it is going to be noticed.
> >   >   > There is almost no way to get around this.  Your identity can be
> >   >   > cloaked without too much trouble, but cloaking your play is much more
> >   >   > difficult -- because you have to play.  Conceivably, you could break
> >   >   > your play up among several sets of machines, but if you are
> >   >   > successful enough I think your play is going to be detected.
> >   >   >
> >   >   > If you are small potatoes, you have less of a problem I'm sure.
> >   >   > Almost no problem.  But if you have a system good enough to interest
> >   >   > someone else, you aren't going to remain small potatoes very long.
> >   >   > And in the meantime, you are going to be putting up some trade
> >   >   > statistics that should attract someone's attention.  Let me change
> >   >   > that to *will* attract someone's attention.
> >   >   >
> >   >   > It's called the smell of money.  And one of humanity's most powerful
> >   >   > olfactory capabilities is detecting that odor.
> >   >   >
> >   >   > Yuki
> >   >   >
> >   >   > Saturday, June 6, 2009, 10:32:32 AM, you wrote:
> >   >   >
> >   >   > KM> The statement, "they will be discovered and traded", contains two
> >   >   > KM> assumptions, which I find difficult to accept.
> >   >   >
> >   >   > KM> First, addressed by Brian below, it will be discovered only if  
> >   >   > it is
> >   >   > KM> used to an extreme extent.  The system may, for example, just  
> >   >   > trade
> >   >   > KM> relatively small lots in large and universally held equities.   
> >   >   > One could
> >   >   > KM> possibly make millions from futures and forex without effecting  
> >   >   > the
> >   >   > KM> markets one iota.  Why would it be discovered?
> >   >   >
> >   >   > KM> Second, even if it were discovered and even became widely  
> >   >   > publicized, it
> >   >   > KM> still might not be traded sufficiently by others to have any  
> >   >   > effect on
> >   >   > KM> its success.  The system might, for example, require considerable
> >   >   > KM> patience by the trader, so much so that only a very small number  
> >   >   > of
> >   >   > KM> traders would be willing to use it.  Or it could be based on  
> >   >   > some theory
> >   >   > KM> that all but a few would reject, despite its effectiveness.
> >   >   >
> >   >   > KM> It's believed by many, including yours truly, the the most  
> >   >   > effective,
> >   >   > KM> low risk/reward, way to make money from the stock markets, is to  
> >   >   > write
> >   >   > KM> books and give lectures about how to make money in the stock  
> >   >   > market.
> >   >   > KM> This system has been going on for years, is well known, and so far
> >   >   > KM> appears to be quite profitable.  I doubt that it will ever stop  
> >   >   > working.
> >   >   >
> >   >   > KM> -- Keith
> >   >   >
> >   >   >
> >   >   > KM> brian_z111 wrote:
> >   >   >>>
> >   >   >>>
> >   >   >>> <snip> I find the statement that all trading systems stop working
> >   >   >>> eventually to be too vague.<snip>
> >   >   >>>
> >   >   >>> Howard has provided supportive arguments, to this theory, at various
> >   >   >>> times, and we can not accuse Howard of being vague or equivocating
> >   >   >>> when it comes to trading (I thank him for that).
> >   >   >>>
> >   >   >>> As I recall the basis of his view is:
> >   >   >>>
> >   >   >>> - all systems will fail eventually
> >   >   >>> - they will be discovered and traded
> >   >   >>> - trading the edge erodes the edge
> >   >   >>>
> >   >   >>> By 'erodes the edge' Howard means that if, for example, I am  
> >   >   >>> trading a
> >   >   >>> system and buy, at the entry signal of 100.00,, and sell on the exit
> >   >   >>> signal of 103.00, I have made a profit of 3%.
> >   >   >>>
> >   >   >>> If a lot of people start trading the same system (same market/
> >   >   >>> timeframe etc) then the second person in will have to buy at, say
> >   >   >>> 100.01 and sell at 102.99 (because my action in buying/selling  
> >   >   >>> before
> >   >   >>> them moved the bid/ask (theoretically trader 2 ends up with a profit
> >   >   >>> of 2.98% , calculated on a commission free basis and so on, down the
> >   >   >>> food chain).
> >   >   >>>
> >   >   >>> According to this theory, the efficiency of the trade has been
> >   >   >>> diminished i.e. what was a 3% trade has been reduced to a <3%  
> >   >   >>> trade(on
> >   >   >>> average) due to other traders piling in to the trade.
> >   >   >>>
> >   >   >>> My critique of that argument is:
> >   >   >>>
> >   >   >>> - the reason why any trade (tick) is made (appears on the tape) is
> >   >   >>> unknown to us (except for our own trade)
> >   >   >>> - all ticks, other than those that are trading our system, are noise
> >   >   >>> (to us) and therefore random
> >   >   >>> - ticks associated with our trade, that are not placed by us, will  
> >   >   >>> be
> >   >   >>> dispersed in time, (due to the various trading time delays  
> >   >   >>> experienced
> >   >   >>> by individual traders).... so they will be interposed by random  
> >   >   >>> ticks
> >   >   >>> - in a pure market (no commissions and no manipulation of the trades
> >   >   >>> by insiders) there is a 50/50 chance that my tick (if I take the
> >   >   >>> market price) will be less than the midprice of the bid/ask when the
> >   >   >>> signal was generated at the exchange.
> >   >   >>> - my price could move away from the original midprice substantially,
> >   >   >>> in a fast market, but no one can know the reason for the fast  
> >   >   >>> trading
> >   >   >>> or attribute it to our system (my system only produces a buy signal
> >   >   >>> once every 2-3 days on average - fast markets happen all of the  
> >   >   >>> time,
> >   >   >>> when I am not trading my system, and presumably slippage is still
> >   >   >>> occurring, in other transactions, so the evidence is against the  
> >   >   >>> fact
> >   >   >>> that my system is the cause of slippage and fast markets).
> >   >   >>>
> >   >   >>> The exception to that is if a 'player' with a big account,  
> >   >   >>> relative to
> >   >   >>> the liquidity of the instrument, is also playing the same system, at
> >   >   >>> the same time, in the same market/instrument/timeframe.
> >   >   >>>
> >   >   >>> So the question is:
> >   >   >>>
> >   >   >>> - to what extent are 'big players' trading a system, in a highly
> >   >   >>> liquid instrument, with enough clout to move the market?
> >   >   >>>
> >   >   >>> - IF big players are system trading what type of system would they  
> >   >   >>> be
> >   >   >>> likely to play and what% of the total funds they are controlling are
> >   >   >>> they likely to risk on any single system?
> >   >   >>>
> >   >   >>> - are they likely to play with large enough sums of money to erode  
> >   >   >>> the
> >   >   >>> efficiency of the system they are trading?
> >   >   >>>
> >   >   >>> - IF they are playing a system, with large amounts of money, is it
> >   >   >>> likely that their system would involve entering all of that money at
> >   >   >>> the same time i.e. they would trade in such a way that they would  
> >   >   >>> make
> >   >   >>> an intraday splash OR are they more likely to trade systematically
> >   >   >>> over longer timeframes (that might be a reason that intraday sytems
> >   >   >>> don't get eroded as often as EOD systems ... if that claim, made by
> >   >   >>> some, is true).
> >   >   >>> - IF big players do trade in such a way that they are 'moving the
> >   >   >>> market' do you think they would be so naive that they are unaware of
> >   >   >>> this and haven't factored that in to their strategy..... if 'moving
> >   >   >>> the market' is negative to their strategy would they do that ...if
> >   >   >>> 'moving the market' is positive to their strategy are they more  
> >   >   >>> likely
> >   >   >>> to implement that strategy in illiquid instruments/small  
> >   >   >>> timeframes OR
> >   >   >>> the reverse?
> >   >   >>>
> >   >   >>> But all of that is just a nice theory.
> >   >   >>>
> >   >   >>> The best argument against any theory is evidence.
> >   >   >>>
> >   >   >>> Some forum members have listed some example trading systems that  
> >   >   >>> have
> >   >   >>> been published for decades AND they are still going strong AND their
> >   >   >>> performance has not 'faded in and out'.
> >   >   >>>
> >   >   >>> Anyone who wants to defend the 'trading the edge erodes the edge'
> >   >   >>> argument now needs to prove that these systems were never published
> >   >   >>> AND that after they were published they ceased to work.
> >   >   >>>
> >   >   >>> That won't be an easy task because Samantha's unequivocal example (a
> >   >   >>> 10 bar SMA on monthly data) is based on a trading idea (MA  
> >   >   >>> crossovers)
> >   >   >>> that has been around forever (Tomasz even ships AB with a example  
> >   >   >>> code
> >   >   >>> in his formula folder and the manual) and there are published  
> >   >   >>> studies
> >   >   >>> on the net (rigorous studies at that) that are relatively current.
> >   >   >>>
> >   >   >>> However, the more imporanat question seems to be, if these systems  
> >   >   >>> did
> >   >   >>> not fail, due to being published and/or traded, why didn't they?
> >   >   >>>
> >   >   >>>
> >   >   >>> --- In amibroker@xxxxxxxxxxxxxxx <mailto:amibroker 
> >   >   >>> %40yahoogroups.com>,
> >   >   >>> "Leading Edge Systems" <rdcpa@> wrote:
> >   >   >>>>
> >   >   >>>> I am new to Amibroker and I have been using Howard's which I find  
> >   >   >>>> to
> >   >   >>> be excellent, as a guide to learing AB.
> >   >   >>>>
> >   >   >>>> I find the statement that all trading systems stop working
> >   >   >>> eventually to be too vague. First "stop working" is a relative term
> >   >   >>> and would have a different meaning for each of us. Also I think
> >   >   >>> inefficiencies can come and go in cycles based on the popularity  
> >   >   >>> of a
> >   >   >>> particular type of trading. Once an inefficiency has been traded  
> >   >   >>> away
> >   >   >>> due to over-popularity, it probably will go out of fashion and then
> >   >   >>> become an inefficiency again some time in the future. All this  
> >   >   >>> depends
> >   >   >>> on the specifics of what we mean by "stop working" and "a system".
> >   >   >>>>
> >   >   >>>> Rich
> >   >   >>>>
> >   >   >>>>
> >   >   >>>>
> >   >   >>>> --- In amibroker@xxxxxxxxxxxxxxx
> >   >   >>> <mailto:amibroker%40yahoogroups.com>, "samu_trading" <samu_trading@>
> >   >   >>> wrote:
> >   >   >>>>>
> >   >   >>>>> All,
> >   >   >>>>>
> >   >   >>>>> In his really good book Quantitative Trading Systems, Howard
> >   >   >>> states that all trading systems will stop working forever at some
> >   >   >>> point (because the inefficiency in the market they exploit will be
> >   >   >>> killed by everybody jumping on board).
> >   >   >>>>>
> >   >   >>>>> On the other hand you have momentum / ROC based systems working
> >   >   >>> forever now, same for trend following MA crossover systems like The
> >   >   >>> one propagated by Mebane Faber. Momentum and MA rossover
> >   >   >>> trendfollowing does seem to work "forever".
> >   >   >>>>>
> >   >   >>>>> Any comments from the gurus here?
> >   >   >>>>>
> >   >   >>>>> Thanks, Samantha
> >   >   >>>>>
> >   >   >>>>
> >   >   >>>
> >   >   >>>
> >   >   >
> >   >   >
> >   >   >
> >   >   >
> >   >   >
> >   >   >
> >   >   > ------------------------------------
> >   >   >
> >   >   > **** IMPORTANT PLEASE READ ****
> >   >   > This group is for the discussion between users only.
> >   >   > This is *NOT* technical support channel.
> >   >   >
> >   >   > TO GET TECHNICAL SUPPORT send an e-mail directly to
> >   >   > SUPPORT {at} amibroker.com
> >   >   >
> >   >   > TO SUBMIT SUGGESTIONS please use FEEDBACK CENTER at
> >   >   > http://www.amibroker.com/feedback/
> >   >   > (submissions sent via other channels won't be considered)
> >   >   >
> >   >   > For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
> >   >   > http://www.amibroker.com/devlog/
> >   >   >
> >   >   > Yahoo! Groups Links
> >   >   >
> >   >   >
> >   >   >
> >   > 
> >   > 
> >   > 
> >   >   ------------------------------------
> >   > 
> >   >   **** IMPORTANT PLEASE READ ****
> >   >   This group is for the discussion between users only.
> >   >   This is *NOT* technical support channel.
> >   > 
> >   >   TO GET TECHNICAL SUPPORT send an e-mail directly to 
> >   >   SUPPORT {at} amibroker.com
> >   > 
> >   >   TO SUBMIT SUGGESTIONS please use FEEDBACK CENTER at
> >   >   http://www.amibroker.com/feedback/
> >   >   (submissions sent via other channels won't be considered)
> >   > 
> >   >   For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
> >   >   http://www.amibroker.com/devlog/
> >   > 
> >   >   Yahoo! Groups Links
> >   >
> > 
> > 
> > 
> > 
> >   ------------------------------------
> > 
> >   **** IMPORTANT PLEASE READ ****
> >   This group is for the discussion between users only.
> >   This is *NOT* technical support channel.
> > 
> >   TO GET TECHNICAL SUPPORT send an e-mail directly to 
> >   SUPPORT {at} amibroker.com
> > 
> >   TO SUBMIT SUGGESTIONS please use FEEDBACK CENTER at
> >   http://www.amibroker.com/feedback/
> >   (submissions sent via other channels won't be considered)
> > 
> >   For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
> >   http://www.amibroker.com/devlog/
> > 
> >   Yahoo! Groups Links
> >
>




------------------------------------

**** IMPORTANT PLEASE READ ****
This group is for the discussion between users only.
This is *NOT* technical support channel.

TO GET TECHNICAL SUPPORT send an e-mail directly to 
SUPPORT {at} amibroker.com

TO SUBMIT SUGGESTIONS please use FEEDBACK CENTER at
http://www.amibroker.com/feedback/
(submissions sent via other channels won't be considered)

For NEW RELEASE ANNOUNCEMENTS and other news always check DEVLOG:
http://www.amibroker.com/devlog/

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/amibroker/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/amibroker/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:amibroker-digest@xxxxxxxxxxxxxxx 
    mailto:amibroker-fullfeatured@xxxxxxxxxxxxxxx

<*> To unsubscribe from this group, send an email to:
    amibroker-unsubscribe@xxxxxxxxxxxxxxx

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/