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

[spike500]

second chart

[spike500]

1 important rule i forgot to mention:
if i miss a signal (which i sadly enough often do) i will never jump after the trade. If missed i wait for the next signal.
This way of working will make you miss good trades, but i will make you miss lossess as well. And by expiriencxe i know that jumping after a trade generates generally losses and seldom profits.

[soylent]

Spike --

Very interesting stuff!

I'm a bit confused as to your definition of long/short. Especially looking @ the 5min bars -- it's as if you know where the extreme low/high would be ahead of time.

Perhaps you're using bar close above/below on 60m to determine trend then using the extreme point of the 60 as your target entry price. (I'm sure it's more complex than that -- but, looking @ your entry/exit pts makes me wonder how you cross referenced the 60m to the 5m to time the entries).

(I'm mostly refering to your SHORT on 5/24 saying the 60m looks short... While I agree it seemd a bit overbought, I wouldn't want to rush a short on that. But, then again, it seems theres a bunch of folks that play the gap reversals (short bigs gaps up) which would coincide w/ your short).

Using a similar thinking to yours, I'm staring @ NQ right now (10:40am) and thinking: TREND is UP on Daily, UP on 60 (though overbought and falling -- yet in a rather strong up move) and UP on 5m w/ a strong counter move down but trend still up.

So, I'm looking to take this long... But, I don't have your conviction to take a long @ the 'support' so I'm waiting for confirmation of a reversal up via linear regression & cci.

hmmm, lets see how it goes.

-c

[Doji3333]

Thanks for your comments Spike...

Spike,

First, I would really like to thank you for your comments and suggetions. This thread is getting really good and we are discussing many valid topics.

Now, to answer your questions...

1. You have far too many trades. As a daytrader I only have between 1 and 5 trades (exceptionally) a day. Average is between 2 and 3 trades. If you have 10 or more trades a day you generate a lot of commissions and slippage but only small gains.

A. My system is averaging about 7 trades per day and about $5600 per month in profits (on 4 contracts of the EMINI). The commissions of the EMINI (with IB) are only $4.80 per RT trade.

The way I look at this is "different ways to skin a cat".

You are correct in the fact that a variant of my system may reduce the number of trades and thus be a better solution for some traders - but not all. Everyone is different and has different approaches to the markets.

2. This proves that you don’t fully profit from the trend. The system seems to me too mechanical. A system is composed by a set of indicators, not 1 or 2.

A. You may be right, but this system was designed as a type of "scalping system". Thus, it does not really care about long-term trend - just short term trend direction.

I'm not using any indicators - just a trend filter and the entry triggers.

3. You create a new system apparently in “ a few minutes”. I need days to program and test any new element in my “system”.

A. I believe this is because I have 20+ years of programming experience and can often create in a few "hours" what it takes some people MONTHS to create. Remember I mentioned the KISS method - this is a perfect example.

4. I don’t believe in the existing indicators. Everyone has them but 90% of the daytraders can’t make money with them. This confirms the worthlessness of these indicators.

A. I'm not using any of the traditional indicators in this system (not even a MA). Now, I don't agree that the majority of the traditional indicators are "worthless". I think it all boils down to "what benefit the end user is deriving from the use of the indicators" vs. "can they be built into a profitable automated trading system". These are two different discussions (I know), but before I call anything "worthless", I have to look at it from all sides.

5. I cannot back test my system automatically, I have to go through all the trades one by one because the rules of my “system” are impossible to program. My latest back testing took almost 1000 trades and it took me about 2 weeks before everything was checked and calculated.

A. My system is very much "mechanical". Thus, it is simple to test. Your system seems much more complex in nature, thus would require many hours of effort to backtest.

I also have a question for you... You stated "I have to go through all the trades one by one because the rules of my “system” are impossible to program".. Does this mean it is a mix of a system and human action?? If it is "impossible to program", how can it then be a computer driven automated trading system?? Based on this definition alone it would indicate that you (the human) are acting on signals your system is generating - thus not automated or mechanical in nature?

6. You have tested different models . The one with the smallest number of trades but with almost the same return as the others is the best one to trade on. Minimize risc and number of trades and maximize the profits. Don’t try to take all the possible profits, no system can do that. Take the parts with the best risc/reward ratio.

A. I agree with you (in concept). Although (again), how many ways can one skin a cat??

7. To me too many consecutive losses in 1 day means that your trendgiving indicator is not optimal. If your indicator is good you will never reenter three times in a row in the wrong direction.

A. These losses were generated from congestion (trend filter changes in direction). This is something that is inherent in my system and the user would have to be aware of them. There may be a way to resolve this, but I don't have a solution yet (have not really focused on it yet).

8. The drawdown in your equity curve line. I think your biggest drawdown is limited to 25% which is very reasonable.

A. The largest drawdown was about $4000 (with 4 contracts) - or $1000 with one contract - spanning more than one day of trading.

I typically test with 4 contracts and a starting equity of $10K. In real life, 4 contracts would require a minimum of $10K+ just for margin requirements - so $20K might be more realistic (as mentioned here before). With one contract $6++K is fine to start with.

Now, you consider the drawdown, it is rather marginal (over time).

Also, the congestion of the system was "contained". Yes, it did go sideways for a while (entered what I call a "phase"). But did not give too much back and held it's own for that time.

In closing, I want to thank your sharing your thoughts on systems. You have obviously sparked some interesting discussions and your continued ability to share concepts from your system development is prompting myself (and others) to consider different ideas.. Thank you.

B

[spike500]

The most important thing is to find on hourly basis a system that gives you the direction of the trend. (My hourly trend generally takes 2 directions within 1 day. So i generally take 1 long and 1 short trade.) For me the trend has to be conceived in such a way that on hourly charts you take more or less all the moves.

Once you have that you have to find a system to enter the market at extreme points. There are several reasons for this entrypoint. First your risc for losses is smaller because you entered at an excellent quote, secondly because your entering quote is so good you can stay longer in a trade and wait to see what happens. And third if it goes wrong your loss will be smaller in general.
My entrypoints seem to be rather based on a contrarian system, but this isn't true. I enter normally in a move that is the opposite of the trend because there the risc is the smallest.
Only problem is to build a system that lets you do this.

The most important thing to me is to enter there where you never have an open loss. That's more important than taking tops and bottoms because from the moment you enter you never have a big loss.

[shogun]

spike

on the 25/05 what kept you in the trade when there were two pullbacks one at 4 ish and the second at 8 ish.

[spike500]

Doji 3333

I agree with your comments. We compare 2 different approaches.

About the possibility to program my system:
I have a number of fixed rules. But these rules are always linked to a number of fixed scenario's. All this together gives a very complex model. I also use 4 different timeframes simultaneous.

I can manage the whole thing well, but it took me a long time to master the whole thing in a fast way.

[spike500]

Originally posted by shogun
spike

on the 25/05 what kept you in the trade when there were two pullbacks one at 4 ish and the second at 8 ish.

I know that it may sound crazy but i can see in the (near) future.
I told that i have a number of scenario's. The hourly scenario told me that i had to stay long until 21.30 (time on the chart). In these scenario's i can "predict" fairly acurate what will happen in the next 1 or 2 hours. So i can always anticipate on what's coming.
These predictins on hourly charts are correct in more than 90% of the cases, and even if they are not correct i will have plenty of time to get out before losing money.

As long as i can try again for free (without losing money) i don't care.

[gspiker]

shogun,

Here is the link to the group:


http://finance.groups.yahoo.com/group/enthios/

As you can see its a yahoo group, so if you aren't signed up for that you will have to do so. Make sure you check all the privacy boxes or you will be flooded with emails from yahoo and SPAMers.

The group is low volume (usually) 3-4 posts a day and usually deals with S&P emini futures. The thread in question talks about mechanical systems on the emini.


Perhaps the most interesting thing in the thread was a pointer to
this tool on Michael's web site (Michael is the group moderator)
http://www.enthios.com/simulator.htm

It it used to project backtest results into the future to show the possible worst drawdowns you will see with a system. The idea is that even a fairly good system will see a point in time where it will hit multiple bad trades in a row. So you take some of the numbers from the backtests you have run, put them into this tool and it shows you what the worst could look like.


Garth

[shogun]

thanks garth,

[BakedWafer]

System

Been working on a system very similar to Dojo. Here is some preliminary data: RT starts tomorrow:

[BakedWafer]

More;

Here is a little more data

[spike500]

Bakedwafer,

just a few questions regarding the results:

do you trade a fix number of contracts or do you take a number of contracts according to the amount of money youhave at that moment?

over how many months did you do this test?

i ask you this because i have had bad experiences with backtesting and the way the computer calculates the returns.

[BakedWafer]

Let me address some of your questions:

1. This is based on a fixed number of contracts, the contract amount is 4. I am trading NQ (Nasdaq Futures) ESig:NQ M4

2. This is tested over 120 days. I have a much more aggressive system with higher returns, but to be honest I want to keep it to 1 - 2 trades a day as automation on any system is still medicore at best.

3. It is set to calculate comissions for the round trip discounted for share amounts > 300, so commissions are less than the typical 2.40 (4.80 RT). Slippage is set to 100 or 1.00 per trade.

J

Originally posted by spike500
Bakedwafer,

just a few questions regarding the results:

do you trade a fix number of contracts or do you take a number of contracts according to the amount of money youhave at that moment?

over how many months did you do this test?

i ask you this because i have had bad experiences with backtesting and the way the computer calculates the returns.

[Doji3333]

Baked...

Make sure you test the NQ #F contract for ALL LONG BACK TESTS..

The M4 contract may not have valid data going back as far as you are testing, thus skewing your results.

B