3 Trading systems on the Q's
Posted by Rogerdodger, Feb 21 2006, 10:38 PM
LINK TO STOCKCHARTS I've picked these up (stolen) from others.
Two use the RSI primarily and the other uses STO and CCI.
Use them at your own risk.
I find them useful when used with other TA.
They show Q's but try them on other stuff.
RSI 2
BUY: 1. Above 200ma
2. 2-period RSI falls 3 days in a row
3. 1st day must be below 60
4. 2-period RSI is below 10 = BUY
5. Buy an additional unit if the 2-period RSI falls 5 days in a row
6. Exit at the close when the 2-period RSI closes above 75
SELL: 1. below 200ma
2. 2-period RSI rises 3 days in a row.
3.The first rising day (day #1) of the 2-period RSI must be above 40.
4. Today the 2-period RSI is above 90: SELL SHORT
5. Sell an additional unit if the 2-period RSI rises 5 days in a row
6. Exit at the close when the 2-period RSI closes below 25.
RSI 5
Buy the Nasdaq 100 Trust (QQQQ) when the 5-day Relative Strength Index (RSI) closes below 30.0.
Sell the Nasdaq 100 Trust (QQQQ) when the 5-day Relative Strength Index (RSI) closes above 50.0.
The Nasdaq 100 Trust is purchased during after-hours trading on the day the RSI buy signal is generated.
The Nasdaq 100 Trust is sold during after-hours trading on the day the RSI sell signal is generated.
The 5-day RSI strategy has a tendency to underperform the buy-and-hold strategy when the market is strong and outperform when the market is weak.
1997 to 2005 Results: NDX "Buy & Hold"up 76.2%, "5 day RSI" up 349.7%
1997 to 2006 Results: NDX "Buy & Hold"up 108.3%, "5 day RSI" up 381.8%
[url]http://www.vtoreport.com/rsi.htm[/url] ([url]http://www.vtoreport.com/rsi.htm[/url]) (Site now closed)
RSI 12 (Sorry, I can only show 2 indicator windows)
Bull Market Signal:
Buy #1: STO & CCI both turn up, unless RSI >70.
Buy #2: CCI turns up and STO below 20.
Sell: STO & CCI both turn down.
Bear Market Signal:
Sell #1: STO & CCI both turn down, Unless RSI<30.
Sell #2: CCI turns down and STO above 80.
Buy: STO & CCI both turn up.
NOTE RSI 12, CCI 10, STO 5,5
Dear Friend,
If you're wanting to be a successful trader let me impress upon you the importance of back testing your system. I believe it to be one of the most critical steps in designing any trading system and here's why.
Dr. Van Tharp in "Trade Your Way To Financial Freedom" describes the 3 components of successful trading to be: the system, money management & your psychology. The only way I know to improve all 3 components is through backtesting.
David Jenyns Live
Here's how backtesting directly impacts each component:
1. The System: Backtesting allows you to find out if and how profitable your system truly is, even before you risk a cent. You can also test different variables to identify the most profitable combinations.
2. Money Managment: Backtesting allows you to discover which money management model works best for your system, further demonstrating the importance of having these rules in place.
3. Psychology: Backtesting is guaranteed to increase the confidence you have in your own system. Not only will you prove you're trading a winning system (making it easier to follow) but you'll have the metrics to back it up.
Backtesting really is what gives professional traders their edge! Sadly, even though it's most important, it's one of the least understood areas of trading.
Worse still, as I illustrated in my article "Upgrading Metastock", MetaStock just isn't up to the task of back testing your trading systems. Watch the video below to see what I mean.
With all that said and done, there is a solution and it is by far the best backtesting system I have ever seen. TradeSim is the first realistic true portfolio trading simulator and back tester available for Metastock.
I highly recommend you read all about it below!
All the best,
David Jenyns | Professional Trader, Author & Coach!
Co-author of the MetaStock Programming Study Guide
Back Testing Your System Is Critical!
By David Samborsky | Director of Compuvision - makers of TradeSim.
Whatever technical analysis criterion you use to trade with, be it moving averages, candle sticks, volatility breakouts, fibonacci retracements or any other trading system you have devised you're going to need to back test your trading system thoroughly and objectively in order to remove any possible doubt about it's capability.
To remove any self-doubt you need to thoroughly back test or simulate your trading system in such a way that it matches the conditions under which it will be traded. Once you have established that you have a reliable and robust trading system, only then will you be confident in trading your system.
The Importance of Back Testing
When trading what is the question in most traders’ mind? To answer this question I shall quote the introduction from Chapter 8 “Back Testing” of Mark Jurik’s book "Computerized Trading".
"Will my trading strategy be profitable? After having gone through the arduous process of crafting a trading strategy, this is the question you must ask yourself."
The ability to answer that question is the great promise that back testing holds out for all traders. A successful back testing procedure will greatly reduce the probability that you will begin trading with either an unprofitable strategy or one that does not meet your expectations. By adopting a sound and rigorous back testing approach, you will: -
Pinpoint which approaches to the market that are likely to be successful and which ones are not.
Generate good estimates of future performance for each trading strategy you test.
Create a record of your trading strategy's historical trading performance.
Produce data necessary for other components of your trading approach such as your asset allocation strategy.
Important Trading System Criteria
Profitability is not the only criteria by which a trading system should be evaluated. Drawdown and stress should equally be considered as well... for example, before you open a trading account:
- Are you satisfied that your system is reliably profitable?
- Will drawdowns wipe out your account?
- Is your system trading in a way you can tolerate?
- Can you tolerate long periods of no trading or too much trading?
- Can you tolerate a large string of losses?
The only way to answer these questions is to subject your trading system to extensive back testing.
Discover The Flaw In MetaStock’s Enhanced System Tester
Unfortunately MetaStock’s system tester leaves a lot to be desired when it comes to realistically and objectively testing the performance of a trading system.
Even though the Enhanced System Tester tests multiple securities in one batch, Metastock treats each security independently of the other. In other words, when Metastock tests the first security, it uses your predefined float and takes the trade over the test period selected. Once completed, it repeats the same process for the second security, using the same initial float, with no reference to the first securities traded.
The consequence of this type of testing is that you receive the same result as you would if you simply tested each security individually and added the results together. Not only is this process dreadfully slow, but the question of the performance of the trading system still goes unanswered. The trader then has to trade the system with a lack of confidence and a certain amount of self-doubt.
Lack of Confidence
Lack of confidence usually forces traders to question their own trading systems with a temptation to continually modify it with devastating consequences. This temptation is usually spawned by a string of losing trades or an opportunity to replace their trading system with a whiz bang indicator that was talked about on a traders chat forum which seems like the answer to all traders prayers.
Anything that sounds too good to be true will attract the attention of a trader who is not satisfied with their own trading system simply because they have not properly tested their system in the first place and have not built up the necessary confidence needed to successfully trade it.
In the end these negative subconscious thoughts will only hinder and destroy your ability to trade successfully. To improve your confidence in your trading system you need to thoroughly and objectively back test it - simple as that! Only then will you be confident enough to commit time and money to it!
The Trader's Dilemma
How can we test how a trading system will perform over a period of time when trading an arbitrary group of securities?
Introducing TradeSim...
TradeSim is the first realistic true trading simulator/analyzer for Metastock that can quickly back-test and evaluate a trading system across a portfolio of securities. With its powerful data processing capabilities TradeSim can evaluate the historic performance of a given trading system within a matter of minutes and do it with a realistic representation of a real-life trading scenario. Whether a single security or a multiple security portfolio, TradeSim answers the simple question: -
"What would have happened if this system had been traded in the past using an arbitrary portfolio of securities?"
Sounds simple - but it is extremely complex if not impossible to do with Metastock as it stands. However, with TradeSim it is just a simple matter of running a Metastock exploration on a portfolio of securities using your own set of indicators. When the exploration has finished you just simply run TradeSim and analyse the resulting trade data.
...
If you're interested in purchasing TradeSim now Click Here
a minha sugestao para a maioria de voces e' que desistam do trading e que se concentrem em tactical asset allocation / dynamic asset allocation
fazem 15% ao ano e gastam melhor o tempo
eu proprio vou acabar por fazer isso pois da mais liberdade e eu quero viajar
a minha sugestao para a maioria de voces e' que desistam do trading e que se concentrem em tactical asset allocation / dynamic asset allocation
fazem 15% ao ano e gastam melhor o tempo
eu proprio vou acabar por fazer isso pois da mais liberdade e eu quero viajar
a minha sugestao para a maioria de voces e' que desistam do trading e que se concentrem em tactical asset allocation / dynamic asset allocation
fazem 15% ao ano e gastam melhor o tempo
eu proprio vou acabar por fazer isso pois da mais liberdade e eu quero viajar
de um livro de investimento que li diz o autor que em dezenas de anos de vida nos mercados nunca viu ninguém enriquecer de market timing
cumprimentos
Estive a pesquisar por alguns livros mais básicos, etc... e encontrei esta série:
"Algorithmic Trading - Algorithmic Trading Strategies"
[url]http://www.amazon.com/Algorithmic-Trading-Strategies-Bootstrapping-ebook/dp/B007C4IAAY[/url] ([url]http://www.amazon.com/Algorithmic-Trading-Strategies-Bootstrapping-ebook/dp/B007C4IAAY[/url])
Parece-me ter aquilo que se precisa: do simples ao mais complexo.
Alguém conhece, recomende?
Alguém consegue ... por exemplo o Volume 32, etc...
System Rules:
1. Buy Signals are generated by a cross over the 5 day simple MA.
2. Sell Signals are generated by a cross below the 8 day simple MA.
3. Do not short GDX if the price is < 3% above the 200 day MA, Instead go to cash and wait for the next buy signal.
4. All new Trades (Longs/Shorts) are entered at the close of the day (true end of day swing system), except for Stops which can occur intra day.
5. Exit Longs and Short when $BPGDM goes above 86%
6. Exit Longs and Short when $BPGDM goes above 83.8% with a gap up >1% in the GDX ETF.
7. Set an initial Stop loss of 8% from the entry price Note: this is NOT a trailing Stop and can be hit any time during the day.
8. Do Not Take shorts when the BPGDM is 12% or below.
9. Time Stop - If you are in a position for 20 days, set a stop loss of +0.50%. This will ensure that you can't take a loss now
10. Exit trades and go to cash once a Profit Target of 13.5% has been met.
11. ATR Trailing Stop, Custom Tradestation Indicator
12. RSI Filter
Há algum 'software' que se distinga para o teste de sistemas? O 'Enhanced System Tester' do Metastock é muito fácil de programar, embora já tenha ouvido falar bem e mal dele, além de se tratar de um programa que usa princípios de uma década atrás, entretanto houve evoluções que penso não terem sido incorporadas no MetaStock.
Da minha pouca experiência com ele, pareceu-me ter observado algumas inconsistências, que se notam ao desenhar os sinais de entrada, saída, stop no gráfico. Aparentemente sinais que deveriam estar lá, não estão, mas pode ser falha minha, ou de configuração do 'tester'.
O Metatrader poderia ser usado, mesmo para simular sistemas não automáticos, o problema é que tem um interface de programação complicadíssimo e pobre. Tem muito a evoluir caso seja destinado ao retalho, para a autoprogramação, terá de evoluir para algo mais prático de programar. Dá a entender que o programa não foi foi feito para isso. Hoje há a possibilidade da programação por objectos, que muitos fornecedores de 'software' usam, e isso até as crianças programam. O compilador do Metatrader é de 'baixo nível' equivalente aos compiladores que se usavam na década de 1980!
Não conhecem nada com a simplicidade do MS mas eventualmente mais eficaz?
Neste verão fiz um robot que negociava Head and Shoulders. Fiz uns testes com ele com resultados fraquinhos, depois nunca mais peguei nele nem acho que vá alguma vez tentar usá-lo outra vez. Acham que há alguma possibilidade de eu o vender? Qual a melhor maneira de o fazer? O código tem que seguir alguma nomenclatura/procedimento standard?
Eu fiz uns backtests... Tenho que experimentá-lo numa demo também. Quais são os sites mais frequentados de compra e venda de robôs, sabes-me dizer?
Eu fiz uns backtests... Tenho que experimentá-lo numa demo também. Quais são os sites mais frequentados de compra e venda de robôs, sabes-me dizer?
Bom dia a todos, eu pessoalmente uso metatrader e para me auxiliar na escolha de estratégias e combinação das mesmas, para efetuar backtest uso este software gratis [url]http://www.mqlsoft.com/download/reportmanager.[/url] ([url]http://www.mqlsoft.com/download/reportmanager.[/url])
Assim consigo organizar-me e saber o comportamento do bolo de estratégias, podendo inclusivamente adicionar o nr de estrategias e nr de pares.
Pessoalmente combino estrategia de break out 2 variações e estratégia de retracement.
Um exemplo de 12 anos e qualquer coisa.
Initial deposit: 5000.00€ Total net profit: 9288.78€
Maximal drawdown: -391.94€ Total trades: 10898
Short positions (won %)5174 (45.419%) Long positions (won %)5724 (46.506%)
Profit trades (% of total)5012 (45.99%)Loss trades (% of total)5886 (54.01%)
Maximum consecutive wins (profit in money) 285.34€
Maximum consecutive losses (loss in money )-96.90€
Maximal consecutive profit (count of wins) 484.22 €
Maximal consecutive loss (count of losses) -143.64 €
Estes valores são sem gestão do risco, lote fixo 0.01lote.
Bons trades a todos.
o connorsrsi nao eh la muito bom, mais vale um simples williams %rQQ dia vou testar uma adaptação dele:
testei o bicho e desisti, parece-me uma manobra de marketing
%R = [(Highest high over 10 bars - today's close) /
(highest high over 30 bars - lowest low over 10)] * -100
Eu experimentei a demo do Multicharts e é excelente, mas o preço faz com que tenha ficado de parte.Só para uma nota.
Para coisas muito básicas uso o Meta, mas ainda hoje vou muitas vezes buscar o velhinho e descontinuado Omega Prosuite porque a Easy language é realmente muito fácil de programar e bastante mais versátil do que o Metastock. O Omega entretanto foi substituído pela plataforma do Tradestation que o Inc referiu, mas não sei se pode ser utilizado por quem não tenha conta.
Há também quem use o ProRealtime que é gratuito e a linguagem de programação é parecida à Easy Language.
o que sao multi-testes?
o amibroker tem walk-forward testing
muitos dos programas aqui referidos não têm built-in a walk forward analysys
já ouvi falar que o traderstudio possui este atributo/função e talvez a tradingblox
alguém sabe de alguma plataforma que não custe os olhos da cara e que produza gráficos decentes, execute multi testes em multi mercados e multi time frames e execute de forma correcta uma walk forward analysis ?
o Metastock, que eu uso e vou provavelmente largar, não tem a não ser via add-in e os bons são caros
desde já o meu obrigado
o que sao multi-testes?
o amibroker tem walk-forward testing
eu queria dizer multi sistemas...erro meu
já realizaste algum walk forward no amibroker ? achas que é fiavel ? é que há algumas aplicações de walk forward que não são grande coisa
que tal é o amibroker em termos de linguagem de programação ?
muitos dos programas aqui referidos não têm built-in a walk forward analysys
já ouvi falar que o traderstudio possui este atributo/função e talvez a tradingblox
alguém sabe de alguma plataforma que não custe os olhos da cara e que produza gráficos decentes, execute multi testes em multi mercados e multi time frames e execute de forma correcta uma walk forward analysis ?
o Metastock, que eu uso e vou provavelmente largar, não tem a não ser via add-in e os bons são caros
desde já o meu obrigado
Penso que o Multicharts ([url]http://www.multicharts.com/[/url]) faz o que queres e usa a Easy Language. Eu em tempos experimentei a trial e gostei bastante (mas também é verdade que vinha habituado ao velhinho Omegatrade Suite 2000).
O preço é que é capaz de estar no intervalo dos olhos da cara. :D
A linguagem de programação é parecida com alguma outra ?
Nunca experimentei o Tradingblox mas tem uma funcionalidade interessante que é a simulação Montecarlo (pelo menos tenho quase a certeza de que foi do Tradingblox que um trader me falou em tempos).mas ja viste o preco?
mas ja viste o preco?Já, por isso é que nunca o experimentei, sequer.
penso que o tradersstudio e a trading blox tambêm têm multiple frame
bem...tenho agora além dos dois referidos o amibroker e o quant share graças ao Neo
obrigado "capitalist pig" :D
agora é cruzar os atributos (onde as análises/simulações multi e a linguagem de programação têm ponderação acrescida) com os preços, pesquisar atentamente umas reviews (parece-me haver umas boas no elite trader) e apurar o vencedor
o amibroker tem money management totalmente programável ?
Essa é a diferença entre um sistema automático e um discricionário, e só em situações muito raras ou anormais é que se justifica a intervenção humana num sistema automatizado.tudo bem...até posso concordar contigo no que respeita aos sinais mas e a colocação de stops, por exemplo, perto duma resistência ou suporte ou outra acção que implique reconhecimento de padrões gráficos....aqui uma mente humana experiente não poderá fazer melhor trabalho que uma máquina ?
I am all for the machine... let it do the dull routine! :)
O mundo da realidade tem limites. O mundo da imaginação não tem fronteiras. - Jean-Jacques Rousseau
o quantshare tem montecarlo, muito fixeHoje decidi fazer o trial.
eh sem duvida o melhor negocio que ja vi, comprei hoje
Kaeppel's Corner: The Greatest Gold Stock System You'll Probably Never UseTestem lá...
By Jay Kaeppel, Optionetics.com | Wed November 28, 2012 12:27PM PT
One of the great paradoxes of trading success is:
A) the need to develop a “method” (roughly defined as an entry and exit timing technique and some attendant money management and risk control measures in place) that has a realistic probability of making money in real time over the long run.
B) the need to avoid messing up the result of A by endlessly “tinkering” with it.
For many individuals this roughly equates analogy wise to either:
A) Wandering in the desert for 40 years.
B) Swerving endlessly from one lane to the next to try to get ahead.
For even once a trader ultimately settles on a winning approach there is always this haunting feeling that there is “something better” out there somewhere. So there is a tendency to not “leave well enough alone.” As a person who has unofficially tested approximately 10,362,123 trading systems (although that number seems a little low) all I can say is “trust me on this one.” This by the way is a mistake. More on this in a moment.
Logic Mistakes
For those new to the method development game there are also the inevitable “logic mistakes”. Typically this is uncovered by the fact that you run a simulation of your newly devised trading system and it makes more money than you could ever dream of making. After you allow the initial rush of euphoria to wash over you for just a few seconds you eventually find yourself saying something that involves “bad words.” For you come to the grim realization that there is some flaw in your trading logic that improves results on paper that you could not actually experience in the real world (I hate that part).
Logic mistakes take many forms but as a youngster my favorite was writing a system so that it would allow you to buy or sell at yesterdays open after you see what happens in the market today. Works fantastically well in back-testing. Not so much in walk forward.
So what follows is my advice in terms of devising and settling on a trading “method.”
The Proper Way to Update an Existing Method
Step #1. Do whatever work you have to do to settle on something that you are comfortable using (i.e., when it gives a signal you actually follow it) and start putting it to use.
Step #2. DO NOT TINKER!! As time goes by and you learn new things you will be tempted to retest things and see how your newfound knowledge might help or hinder performance. That’s fine. Set up a set interval for considering changes to your existing method.
For example, once a month or once a quarter, etc. But – repeating now – DO NOT TINKER with your method day in and day out. If you are constantly tinkering with a method then you will never quite feel settled with it and sooner or later you will start to second guess things when you should simply be placing orders.
A Method for Trading GDX – Maybe?
So in the title of this week’s piece I mentioned a “great system” for trading ticker GDX, but also suggested that you will probably never use it. What the heck is that all about? Well, I read something recently by Laurence Connors and he pointed out the fact that over some given period of time the S&P 500 has gained far more from each daily close to the next day’s open than from each day’s open to that day’s close.
To state it more clearly, if you bought at the close every single trading day and sold at the next trading days open you would have made more than if every single trading day you bought at the open and sold at the close. This of course seems somewhat counterintuitive since many traders are sort of preprogrammed to believe that holding a position “overnight” is far riskier than hold a position “intraday” and “exiting by the close.”
So somehow – must have been the end of a month – I tested this idea using ticker GDX, the exchange-traded fund that tracks the price of gold stocks.
The results were, well, “interesting”.
The GDX LC Method
Since I got the idea from Laurence Connors I’ll use his initials here in naming this method (although in all candor if it somehow becomes super popular I will undoubtedly change it from LC back to JK – sorry, it’s just my nature). Ticker GDX started trading on 5/22/2006.
Test #1:
Buy 100 shares of GDX at the close of trading each and every single trading day. Sell those shares at the open of the following trading day.
Test #2:
Buy 100 shares of GDX at the open each and every single trading day. Sell those shares at the close of trading the same day.
The results appear in Figure 1.
Figure 1 - $’s gained using Test #1 (blue line) versus $’s lost using Test #2 (May 2006-Present)
The blue line in Figure 1 displays the dollars gained by buying GDX at the close each day and selling it at the following open.
The red line in Figure 2 displays the dollars lost by buying GDX at the open each day and selling at the close of that day.
For the record:
-The method of buying at the close and selling at the next open gained $12,154.
-The method of buying at the open and selling at the same day’s close lost -$11,131.
So to go one step further, Test #3 works as follows:
Step #1 - Day #1: Buy 100 shares of GDX each day at the close.
Step #2 - Every subsequent day: Sell short 200 shares of GDX each day at the open and then buy 200 shares of GDX at the close.
Repeat Step #2 ad infinitum.
The results of this method (long GDX overnight and short GDX intraday) appear in Figure 2.
Figure 2 – Long GDX at the close, short GDX at the open, repeat (100 shares since May 2006)
Summary
So who’s ready to sign on and start trading? That’s kind off what I figured. There are a few things to ponder here. First off is the obvious fact that the results shown here do not take slippage or commissions into consideration. Obviously I you are paying commissions of any meaningful amount the results would deteriorate quickly due to the frequency of trading.
Secondly – and repeating now – this method is counterintuitive to what most traders are taught to think – that “overnight risk” is “bad” but intraday risk is OK.
So despite the compelling results as a standalone method this is something of “You First” method, as in, “you try it first and let me know how it works.”
The real lessons here may be:
-There can be an advantage to looking at things that seem counterintuitive for they represent ideas that most people are not looking at or considering. If you are looking for an “edge” does it make more sense to look, a) where most people are looking, or b) not.
-The particular “raw” system that I detailed here may be a good starting point for some further tinkering.
Of course, I’ll have to wait until the end of the month. My “No Tinkering Today” sign is up.
Jay Kaeppel
Staff Writer and Trading Strategist
Optionetics.com ~ Your Options Education Site
This is a variation of Jay Kaeppel's post at optionetics but using some extra ETFs that have exhibited the same kind of behavior. Original article: [url]http://www.optionetics.com/market/articles/2012/11/28/kaeppels-corner-the-greatest-gold-stock-system-youll-probably-never-use[/url] ([url]http://www.optionetics.com/market/articles/2012/11/28/kaeppels-corner-the-greatest-gold-stock-system-youll-probably-never-use[/url])
It buys equal amounts of each of these 6 etfs: WOOD,GDX,EPU,IDX,PALL,JJG.
It buys on the close of the U.S.session, holds overnight and sells on the open the next day.
That's it.
Would you invest in this strategy?
parece-me algo terrivelmente exposto a slippage e o slippage é algo que embora incontornável eu fugirei sempre como o diabo da cruz porque muito dificilmente o consegues quantificar para o incorporares nas contas dum sistema. o slippage é um dos motivos porque muito dificilmente eu algum dia terei um sistema que me dê como resultado muitos trades pois é em sistemas de daytrading ou de time frames muito curtos que estás mais exposto a slippageCitarKaeppel's Corner: The Greatest Gold Stock System You'll Probably Never UseTestem lá...
By Jay Kaeppel, Optionetics.com | Wed November 28, 2012 12:27PM PT
One of the great paradoxes of trading success is:
A) the need to develop a “method” (roughly defined as an entry and exit timing technique and some attendant money management and risk control measures in place) that has a realistic probability of making money in real time over the long run.
B) the need to avoid messing up the result of A by endlessly “tinkering” with it.
For many individuals this roughly equates analogy wise to either:
A) Wandering in the desert for 40 years.
B) Swerving endlessly from one lane to the next to try to get ahead.
For even once a trader ultimately settles on a winning approach there is always this haunting feeling that there is “something better” out there somewhere. So there is a tendency to not “leave well enough alone.” As a person who has unofficially tested approximately 10,362,123 trading systems (although that number seems a little low) all I can say is “trust me on this one.” This by the way is a mistake. More on this in a moment.
Logic Mistakes
For those new to the method development game there are also the inevitable “logic mistakes”. Typically this is uncovered by the fact that you run a simulation of your newly devised trading system and it makes more money than you could ever dream of making. After you allow the initial rush of euphoria to wash over you for just a few seconds you eventually find yourself saying something that involves “bad words.” For you come to the grim realization that there is some flaw in your trading logic that improves results on paper that you could not actually experience in the real world (I hate that part).
Logic mistakes take many forms but as a youngster my favorite was writing a system so that it would allow you to buy or sell at yesterdays open after you see what happens in the market today. Works fantastically well in back-testing. Not so much in walk forward.
So what follows is my advice in terms of devising and settling on a trading “method.”
The Proper Way to Update an Existing Method
Step #1. Do whatever work you have to do to settle on something that you are comfortable using (i.e., when it gives a signal you actually follow it) and start putting it to use.
Step #2. DO NOT TINKER!! As time goes by and you learn new things you will be tempted to retest things and see how your newfound knowledge might help or hinder performance. That’s fine. Set up a set interval for considering changes to your existing method.
For example, once a month or once a quarter, etc. But – repeating now – DO NOT TINKER with your method day in and day out. If you are constantly tinkering with a method then you will never quite feel settled with it and sooner or later you will start to second guess things when you should simply be placing orders.
A Method for Trading GDX – Maybe?
So in the title of this week’s piece I mentioned a “great system” for trading ticker GDX, but also suggested that you will probably never use it. What the heck is that all about? Well, I read something recently by Laurence Connors and he pointed out the fact that over some given period of time the S&P 500 has gained far more from each daily close to the next day’s open than from each day’s open to that day’s close.
To state it more clearly, if you bought at the close every single trading day and sold at the next trading days open you would have made more than if every single trading day you bought at the open and sold at the close. This of course seems somewhat counterintuitive since many traders are sort of preprogrammed to believe that holding a position “overnight” is far riskier than hold a position “intraday” and “exiting by the close.”
So somehow – must have been the end of a month – I tested this idea using ticker GDX, the exchange-traded fund that tracks the price of gold stocks.
The results were, well, “interesting”.
The GDX LC Method
Since I got the idea from Laurence Connors I’ll use his initials here in naming this method (although in all candor if it somehow becomes super popular I will undoubtedly change it from LC back to JK – sorry, it’s just my nature). Ticker GDX started trading on 5/22/2006.
Test #1:
Buy 100 shares of GDX at the close of trading each and every single trading day. Sell those shares at the open of the following trading day.
Test #2:
Buy 100 shares of GDX at the open each and every single trading day. Sell those shares at the close of trading the same day.
The results appear in Figure 1.
Figure 1 - $’s gained using Test #1 (blue line) versus $’s lost using Test #2 (May 2006-Present)
The blue line in Figure 1 displays the dollars gained by buying GDX at the close each day and selling it at the following open.
The red line in Figure 2 displays the dollars lost by buying GDX at the open each day and selling at the close of that day.
For the record:
-The method of buying at the close and selling at the next open gained $12,154.
-The method of buying at the open and selling at the same day’s close lost -$11,131.
So to go one step further, Test #3 works as follows:
Step #1 - Day #1: Buy 100 shares of GDX each day at the close.
Step #2 - Every subsequent day: Sell short 200 shares of GDX each day at the open and then buy 200 shares of GDX at the close.
Repeat Step #2 ad infinitum.
The results of this method (long GDX overnight and short GDX intraday) appear in Figure 2.
Figure 2 – Long GDX at the close, short GDX at the open, repeat (100 shares since May 2006)
Summary
So who’s ready to sign on and start trading? That’s kind off what I figured. There are a few things to ponder here. First off is the obvious fact that the results shown here do not take slippage or commissions into consideration. Obviously I you are paying commissions of any meaningful amount the results would deteriorate quickly due to the frequency of trading.
Secondly – and repeating now – this method is counterintuitive to what most traders are taught to think – that “overnight risk” is “bad” but intraday risk is OK.
So despite the compelling results as a standalone method this is something of “You First” method, as in, “you try it first and let me know how it works.”
The real lessons here may be:
-There can be an advantage to looking at things that seem counterintuitive for they represent ideas that most people are not looking at or considering. If you are looking for an “edge” does it make more sense to look, a) where most people are looking, or b) not.
-The particular “raw” system that I detailed here may be a good starting point for some further tinkering.
Of course, I’ll have to wait until the end of the month. My “No Tinkering Today” sign is up.
Jay Kaeppel
Staff Writer and Trading Strategist
Optionetics.com ~ Your Options Education Site
Que raio de ideia!!
E adaptar para outros...
alguem experimentou o quantshare? o que acharam ?ainda não tive tempo para me debruçar sobre isso
alguem experimentou o quantshare? o que acharam ?
alguem experimentou o quantshare? o que acharam ?
Fiz o trial. Fiquei bem impressionado. Mas ainda não mudei. Requer um consideravel investimento de tempo, que de momento não tenho.
alguem experimentou o quantshare? o que acharam ?
Fiz o trial. Fiquei bem impressionado. Mas ainda não mudei. Requer um consideravel investimento de tempo, que de momento não tenho.
qual eh o que usas agora?
Sempre longo no overnight e sempre curto no intraday.
Here is a capsule of How to design a Profitable Trading System:
· observe market inefficiency to exploit (lots of reading, research and watching markets/charts)
· quantify how to exploit the inefficiency
· create exact rules for entries and exits
· determine inputs needed and whether they will be fixed, adjust to market, or be re-optimized on a periodic walk-forward basis.
· write the strategy code (and/or indicator for visualization) and apply to a chart of out of sample data. this data could be from a symbol you never plan to trade, or from 15 years ago on the symbol you do want to trade, or for the past 30 days knowing this period will never be valid for testing results (I many times use this period when I'm testing something I've noticed in the charts). All of this is to ensure you have the strategy coded correctly, AND you are not using data you are going to test on.
· determine measure(s) of strategy being successful (net profit, profit/drawdown ratio, expectancy, sharpe ratio, average trade net profit, etc.)
· determine commission and slippage needed for accurate results and be conservative. For example, say your round-turn commission is $4.70 and you expect less than a tick slippage each side (entry and exit), I would use $5 for commission and 1 tick slippage each side).
· test on in-sample data and DO NOT test on out-of-sample data. For example, say you are testing a strategy on the ES (S&P 500 emini futures), you could use 1998-2003 for in-sample. If the strategy fails here YOU ARE DONE with this strategy - move on to the next one. Or maybe adjust the strategy rules a little bit and determine the method for re-optimizing some of the parameters going forward.
· walk-forward test the strategy on a portion of the next data (2004-2006 in our example). If the strategy fails here YOU ARE DONE with this strategy - move on to the next one. Seriously! If you keep tweaking you are just over-fitting and you will not get real-time results that are similar to the backtest.
· if the strategy has passed all measures to this point then you might have something (yes I'm saying there is a chance ;-) the odds are still against us) AND everything about the strategy is finalized (no more tweaks!), then you can walk-forward on your pure out-of-sample data. This is it, the last straw, if it works it works, if it doesn't it doesn't. You can't go back! Put your seat belt on, hope for the best, and be ready to be completely disappointed.
· if this is the 1 of 100 (or 1000) that passes all the testing, then test it live (in simulation or small amount of real money) to make sure code is correct for live (real!) trading.
· once satisfied it is working correctly then determine your position sizing based on your risk control (topic for another day) and start trading with real money.
· monitor your trading based on the measures you originally created and STOP trading the system when it fails to pass these measures.
· now start over with your next strategy....
Bom dia, alguém disponível para testar um sistema para o S&P 500 com dois indicadores e para entrar apenas longo, coisa muito simples, mas com bom aspecto, ou no mínimo fazer um backteste?
Bom dia, alguém disponível para testar um sistema para o S&P 500 com dois indicadores e para entrar apenas longo, coisa muito simples, mas com bom aspecto, ou no mínimo fazer um backteste?
barras diarias? posso fazer isso, manda pm ou coloca aqui as regras
Bom dia, alguém disponível para testar um sistema para o S&P 500 com dois indicadores e para entrar apenas longo, coisa muito simples, mas com bom aspecto, ou no mínimo fazer um backteste?
barras diarias? posso fazer isso, manda pm ou coloca aqui as regras
Estou bloqueado, não aceitas a minha msg privada 8)
Bom dia, alguém disponível para testar um sistema para o S&P 500 com dois indicadores e para entrar apenas longo, coisa muito simples, mas com bom aspecto, ou no mínimo fazer um backteste?
barras diarias? posso fazer isso, manda pm ou coloca aqui as regras
Estou bloqueado, não aceitas a minha msg privada 8)
ja deve funcionar