If your using EFS, enter the highest return you've had...

[spike500]

Originally posted by gspiker
What is interesting, and may make a nice addition to this thread, is what do you do once you find a system that seems to work for you.


Garth

Once you have a good working system you go LOW PROFILE.
The more you tell your system is terrific the more people will know it, and sooner or later you will get the bill.
In Europe your profits will lead to paying social security (21%) on your profit. After that your profit will be taxed as a professional revenue ( approx 60%). Final conclusion: 100$ profit will give you net 31.6$; all the rest goes to taxes.

Conclusion:Telling how good your system is will bring no profits, only losses (taxes).

And on top of that you will never give details on your system, so no one will ever be able to profit from your knowledge.

Maybe the situation is different in the US, but making money is a crime in Europe.

[gspiker]

Spike500,

Things are a bit different in the US...esp since a good trader will likely incorporate to lower his/her tax base (it's not a crime to make money here, but companies are allowed more freedom to do so than individuals).

But in any case I'm not interested in details. Just as I will not divulge anything in mine that is not already in the public domain, I don't expect others will want to divulge theirs.

Instead I'm interested in METHODOLOGY. Which ultimately I think is the key anyway. There are as many successful system out there as their are traders (well close to anyways) but its the METHODOLOGY of the application and improvement of the system that will turn it from good to great. And I think there are only a limited number of good ways to do this.

I'll give an example. In my early years with my main mechanical system, I ran into a problem of it having either a great % of profitable trades, but with a large drawdown, or I could put some hard rules in place for profit and loss taking and lower my % profitable but also eliminate the huge drawdown. The fix was to use a methodology of finding the closest "strong enough" technical stop (rather than fixed)...and implement a "technical trailing stop" for when the system turned profitable (though I still stuck to taking profits at target levels if a target number was generated).

This is the kind of stuff I think people would be interested in.

Garth

[Gavishti]

Great stuff. Thanks for the posts.

[spike500]

Gspiker

Gspiker,

My system was developed in several steps. In the past years I was convinced at least 5 times that I could not improve my system anymore. But each time I came up with improvements. I think that’s normal because in reality you encounter lots of situations that you analyse much more intensively because things happen in real time, so they happen very slow.
In back testing you have not the same contact with the market as you have whilst trading. Your computer runs through the data and gives you the results. But you don’t see the details and you sometimes have no idea how the computer came to these results.
That’s why I don’t believe that you can develop a good system by writing conditions and running them through your computer with or without optimising them. I wrote my formulas and tested them manually. So I saw for each bar why something happened (or didn’t happen). It appears that I have a very analytical mind (although I don’t have the impression that this is true) that gives me a good insight in the strong and weak points of my formulas.

I never had the problem of huge drawdowns because I trade according a trend following system. The only problem I had was the fact that my profits were small because i traded too trendy in my trend system. Over the years I managed to optimise my entry- and exit points. So my draw down stayed small and my profits increased. The returns of my system went over tenfold of the initial returns without ever increasing my drawdown. And these returns are actually quite high.

[derekg]

Gavishti,

In response to your earlier post, yes the equity curve I posted is based on a system I trade with real money. However, the method I use is a little more complicated than what I can code in EFS as my programming skills aren't at the same level as others here on the board. My purpose in coding was not to find out whether the system was profitable or not, as I have been trading this way for years. Instead, I'm trying to use it as a tool to find other stocks to trade using the same method.

When I first started backtesting what I put together, I was intrigued at the fact the system seemed to be outperforming my real world results even though I entered $.02/sh commissions (a lot more than what I really pay) for backtesting. I figured that was enough to compensate for the occasional slide getting into and out of stocks. I've been watching it kick out signals and there have been many times it has outperformed my judgement remarkably well. Still, I quickly realized the extra commissions weren't nearly enough to account for the slide, especially on NYSE, so I doubled the commission rate yet again and the results are more realistic. You'll see the difference in the chart I'm posting.

The chart I posted early on in the thread was 120 days up to around 5/15/04 - trade #140 on this current 120 day chart. You'll also notice that, ironically, that's the peak of equity. These summer doldrums have been terrible for my style of trading, not just for this stock, but others as well. I should've gone on vacation.

[Gavishti]

derekg,

Great to see real world results. Thanks.


gavishTi

[BakedWafer]

Here is what I got for a full week one:

Trading: NQ 10 Contracts
+6,800 (week one)
+3,100 (week two) Monday and Tuesday

I'm glad I started this thread. It gives us all a good chance to compare systems.

J

[derekg]

Originally posted by BakedWafer
Here is what I got for a full week one:

I'm glad I started this thread. It gives us all a good chance to compare systems.

J

I give it a thumbs-up. And a toothbrush.

[chrisf95003]

Baked wafer thats awesome! How many contracts is that based off of?

With a fully automated system, during the day you just watch the system for bad ticks come up or are you continually doing research and other things in the day also?

Chris

[fflaque]

I would add mine but esignal is horrible with forex quotes

[derekg]

A question for you futures traders and developers. Because of the amount of slippage I had to factor in for equities trades in order to get realistic results, I'm wondering what's best to use when backtesting futures systems.

I've never traded futures, but I understand the ES is extremely liquid and aside from a running market, hitting the ask will get you a fill so fast it's usually confirmed before you release the button (I can only dream of fills this fast.) Is it realistic to code a system to buy ask and sell bid and get realistic results?

How about for the less liquid YM and NQ?

Thanks,
Derek

[BakedWafer]

Originally posted by derekg
A question for you futures traders and developers. Because of the amount of slippage I had to factor in for equities trades in order to get realistic results, I'm wondering what's best to use when backtesting futures systems.

I've never traded futures, but I understand the ES is extremely liquid and aside from a running market, hitting the ask will get you a fill so fast it's usually confirmed before you release the button (I can only dream of fills this fast.) Is it realistic to code a system to buy ask and sell bid and get realistic results?

How about for the less liquid YM and NQ?

Thanks,
Derek

I used slippage of 2.00 when I did my backtesting results for futures with sometimes getting up to 5.00 in order to look for real bad results.

When I started to make a system I realized you must use the premise of 'every trade is a worst case scenario.' By this I mean, when coding, code everything as if you have the worst possible timing.

1. Entering trades at the open of the bar after the signal is generated.

2. Use huge amounts of slippage, since NQ is very liquid you will almost always get the fill price you are looking for but you NEED to be safe.

It is very realistic to use the bid and ask for results for your system. JUST remember to add slippage to your trade, usually a dollar or two to account for anything that may happen during real time trading.

NQ is amazingly liquid if you market your orders, they will usually fill in less than 1/16 of a second. You will not even have enough to time to see it fill. The fluidity of NQ or any trading system for that matter coupled with the speed of the markets makes it a very exciting time for at home on in-house automation.

[BakedWafer]

Q

Trading: NQ (Trading 10 contracts)
+6,800 (week one)
+7,000 (week two) M - Wednesday

[BakedWafer]

Originally posted by chrisf95003
Baked wafer thats awesome! How many contracts is that based off of?

With a fully automated system, during the day you just watch the system for bad ticks come up or are you continually doing research and other things in the day also?

Chris

1. 10 Contracts

2. The system usually only makes one trade a day, sometimes it will make 2. I do watch it during the day, I have a sound file play when a trade comes through and turn my speakers up loud so I can be anywhere in my house, then I just run to the computer and confirm the order. I have yet to get to 'full automation' as I do not trust the computer 100% yet to do my job and make it confirm everything with me first. So to answer your question, I spend a few hours a day researching different methods to squeeze more out of the Futures market. It's tough to make yourself keep looking when you have something that 'works.' But you have to remember some systems work in some markets and some in others, you never know which one you have.

P.S. to E-Signal. This is why we need backtesting for 3-years. GET ON IT!

[derekg]

Originally posted by BakedWafer
I used slippage of 2.00 when I did my backtesting results for futures with sometimes getting up to 5.00 in order to look for real bad results.

Thanks BakedWafer!