how to improve my system

[soylent]

Here's some more from today...

I find NQ is getting crappier and crappier to trade. So many fake moves and overlap on bars.

But...

[mateusz]

your method looks nice, clear and effective as you say. If you feel comfortable with this then its great. Now, you have to say that on dow it was easy day to trade, nasdaq was harder - euro which Im trading is more like nasdaq - sometimes even much harder. There are many shakes of the market - people take position and then because of their tight stops are shaked of the market, then they take oposite position because they want to cover their loss and in this moment it occures that they were on a good side with their first position but they are discouraged to do this because they had two losses and they stopped beliving themselves or their methods. Normal thing someone would say.

Now, as I look at your method, dont you interpretate volume? I think it is very very important thing which shouldnt be missed in any system. Actually I think this is the only true thing we get on screen. Indicators, price, avereges - all great but without volume... it is dangerouse to trust only in price.

[mateusz]

one more thing: for me interpretation of a market is most important - if you interpretate the market good then your trades will be good. Now, is it better to interpretate the last 14 or 20 or even 40 bars on 5min chart than to interpretate the whole situation which you wont see on that chart?

[soylent]

Hi Mat --

Like I said: If your system works, stick w/ it.

I've been trading for 10 months. I went through all sorts of combobulations trying to figure it out.

To me it comes down to:
a) Long-term (3+ days hold time): Look @ multiple timeframes, fundamentals, etc.

b) Intraday: Find a good indicator and use just that one.

I can use things like the 50-day MA as a 'target' on my intraday chart. That does make it a little more efficient. But, I'm looking @ 5 indexes + the euro. I just let CCI tell me 'where the 50-day is'. CCI starts behaving differently around major support/resistance.

What I find is the longer-term indicators give me a hindrance because they bias my belief. For instance, yesterday, I was biased on the NQ because I didn't think we could break the low from last week.

I used to scan 10 different indicators: Trin, tick, ad line, etc... I used to cross-reference markets. I don't find they add value. (OK, I do still look @ them but I'm not sure they are adding value)

About the ONLY trick I use now: I look @ all the indexes. If they are all showing they will break to the upside and one starts ahead of the others, I'll go-long the laggard.

As for volume: Big bar=big volume. Small bar=small volume. Intraday... On 'daily' I think volume is more important. But, even then, w/ derivatives being what they are, I think volume only increases your odds 1-2%. And, aside from that, I want to be in BEFORE the big volume hits -- not after. So, I'm exiting / fading a bar or two after the volume shows up.

So...

Back to volume -- because on a really slow market like EURO Futures it can add a lot of extra info -- what I use is a TICK chart. By using a 133T, I get a bar for every 133 'sales' on the euro. This makes my chart move more quickly during heavy volume / 'sharpens' my indicators blade so I can really see what's happening...

Here's how I trade the euro... I only trade from 8am until 1pm.

-c

[mateusz]

I am testing CCI on 5 min chart - mostly works like stochastic 8.6.3 but i think im getting started to like it better. Faster and more precise. This was really helpful Soylent. Thanks.

As to volume: its not about big bars and big volumes. Volume for me is interesting to exit and enter position because of for example lower low and lower volume which indicates that the down trend has no power - thats ofcourse simple example but when you get to know the instrument you trade better you will see some very interesting volume patterns - it is all about to catch true signals and avoid fakes, wright? And volume can show that better than any indicator which counts only price relations. Because the price is the thing that can be faked most.

[soylent]

volume: That's true - it shows the interest but you're still dealing w/ day-trading mentality. Remember most everyone you compete w/ is a fellow daytrader. There is some actual arbitrage/spreading of orders over longer timeframes that come into play. But, lets assume they are dumping $1M every 15mins. Well, you'll see a spike down then 10mins sideways, spike, 10, spike, 10... Until it levels out.

Some swingers will be exiting @ different times. Some will look @ the effect on 5min, others on 60min. Others @ tick.

So, if you are trading @ the 1min level or tick level, it's just as well to measure actual bar momentum vs. actual volume momentum. Because a) you have a stop loss and b) you have a target profit.

The problem w/ volume is: it can dry up or continue @ anytime. So, I try not to worry about it.

Though, I DO look @ the DOM (depth of market) to see if anyone is tossing orders for 250 contracts @ a certain #. This will show actual buying support/strength @ a target number. A buyer that has a sell/buy order for an actual investor vs. us traders. Often, that person will keep showing up @ various places. I try to look for them.

Again: If you find volume helps -- don't listent to me! I -do- use $ADD to show overall market direction. But, again, I don't know that it helps me.

FINALLY: On CCI: Be sure to set it to HLC/3 as your setting for calculations. It works better / shows a truer picture.

-c

[bassocat5]

Programming?

Speaking purely from a programming standpoint, I feel you can improve things by creating an automated filtering/alerting process. I have created systems that use multiple chart timeframes. Each chart sends mini-alerts to a master chart/efs (probably your 1 minute chart that you are using to "trigger") via global variables. The master then evaluates all the global data and creates the "real" alert. This can all be tested using eSignal's neat tick replay mode. This isn't as fast or as robust as the "one chart" backtesting and strategy evaluation that is normally available in eSignal, but it does give you the ability to easily try out a multi-chart strategy on a few weeks of data. I'd offer my assistance for the experience of seeing a high probability strategy at work while also "playing" with Forex (the no-commission/ pip arrangement is seductive ).

Rich

[mooseman]

bassocat5,
How hard would it be for a newbie at EFS, no ActiveX or DDE experience to adopt your methods mentioned below? Would you make some code and documentation available? What you are doing appeals to me but I fear the hours to make it happen absed on my current EFS experience level. Having something working with some indicators would be a great starting point from which I believe I could alter the indicators to my choosing.
Dave

[tedk]

bassocat5
Interesting. I did something similar and wonder how you handle the "slowlyness" of
the higher timeframes.
For instance, in a 15-min timeframe it takes the full 15 min to
form the whole price bar and only after the last tick the bar is complete and indicator
information becomes is available. And only then a signal of the 15-min TF can be send to the lower
timeframe/master signal generator. But looking at it visually you can often see the most probable direction
and go for it even before the final signal of the higher timeframe is actually given.

[bassocat5]

"Slowlyness" and other stuff

TedK,

Thanks for your post, it's sick, but it's comforting to know that someone else has experienced some of the growing pains that I went through

The indicator information is always available (unless you are using "compute on close" in your efs). The question is: do you want to use the data from the previously closed bar, the current "unfinished" bar, or wait until the close of the current bar. There are a lot of variables involved with creating one of these multichart strategies.

Everything depends on the strategy to be enacted. In "real time", does your strategy require that you wait for the 15 minute bar to close? (If it does, by the way, then the 15 minute chart becomes the trigger chart. )

Not all indicators have the same volatility (obviously).But what I'm getting at is, some indicators can swing madly in and out of a user defined threshhold during the creation of just one bar. Just watch a short term stochastic sometimes. On the other hand, an OBV indicator is less likely to "swing".

I have set things up so that my global variables are arrays, passing lots of info, the indicator's data, the cpu clock timestamp and a timebar stamp (sounds redundant but believe me, those stamps come in handy).

In your post you say: "looking at it visually you can often see the most probable direction and go for it even before the final signal ".
In the 1st minute of a 5 minute bar, most signals are almost worthless, but as it gets closer to the close of bar, it becomes more "reliable". So maybe "probable" can be defined by inserting exception rules like: use previous closed bar unless current bar is 75% completed (where I might use my timestamps in a calculation). Obviously, these types of coded strategies require a different level of programming expertise.

I treat these types of strategies like software projects, by spending a large chunk of time on the defining of terms and strategy before any coding is done. Sometimes during this definition stage, I might find that the strategy is flawed or that it relies too much on "seat of the pants feel".

I'm not the "king of programming", I'm just trying to give people a heads up as to what is involved with testing and executing these types of automated strategies.

Rich

[bassocat5]

mooseman

Dave,

If you look at my post to tedk, I think you might seel that these types of strategies are difficult undertakings for beginning programmers. The thought of properly documenting one of these strategies with associated testing procedures gives me a head ache. If you decide to try one for these multichart strategies for yourself, I believe my previous post has a lot of usable info to start off. If you want to start your own thread on the topic (I haven't looked, there might already be one), I'll be willing to add to the discussion (and I'm sure many more experienced programmers on this board would too).

Rich

[tedk]

bassocat
Thanks for your reply. Indeed, it depends very much of the type of indicator whether the "running" value can be used or the final value only.
An alternative approach might be to covert higher TF info to lower TF. eg MA5 of the 15-min TF is the "same" as a MA75 of the 1-min.

Obviously, a simplistic view. Bear in mind the following.
Assuming simple method on both MAs 1 Close of the 15-min is replaced by 15 Closes on the 1-min. But actually 15 Closes
(simple) of the 1-min is closer to 1 weighted bar of the 15-min instead of 1 Close simple. And even that is not true.
Whatever is done the result will not be exactly the same. Look at it, MA5 is rather choppy whereas MA75 is very smooth.
Particularly in a high volatility market the difference is obvious.
Nevertheless, this simple approach might be an easy way out to implement some parts of an multi-TF tradingsystem.