Primer on the Formulation of an Index of Weighted Biases

[UrmaBlume 10]

The idea is to calculated an index of weighted biases from higher time/volume frames, store/map them as global variables and then access them from a lower time/volume frame where structure allows more precise entries and closer stops.

 

A bias can be as simple as a faster moving average being above or below a slower moving average.

 

In the case demonstrated below we take 24 inputs from higher time frames such as Jurik smoothed measure of price, net commercial trade and the balance of trade along with MACD's and rates of change on those same data series and several others.

 

We optimize the settings on each of these indicators and then asign a bias of +1 or -1 to each of the 24 inputs. We then run further optimizations to find the most suitable weight for each of the 24 inputs. The sum of those weights for an 8k chart in the ES is shown below.

 

This is one of the very few places were traditional processing such as MACE and Rate of Change can be of use and where there are many, many combinations of biases and time frames than can produce profitable results in almost any market.

 

This chart shows the first 4 1/2 hours of trading today 7/16/2009.

 

[N57H 10]

Could you post the indicator please?

[dandxg 10]

Thank you UB for sharing you wisdom and insight.

[ramora 10]

Great post and screen shot!

 

Since you are open to questions, what are the 24 inputs that created that screen?

 

 

Thank you again for all of your posts.

[UrmaBlume 10]

Great post and screen shot!

Since you are open to questions, what are the 24 inputs that created that screen? Thank you again for all of your posts.

 

Thank you very much for the kind words.

 

Each of the 24 inputs was either precisely described or strongly hinted in the OP.

 

The idea is that there are many, many ways to read higher time frame bias and bring the power of that bias down to where you execute.

 

My 24 inputs are but some of the infinite number of combinations of inputs, settings, weights and time frames that will produce profits from the basic formulation protocol defined in the OP.

 

It's the best combination treasure hunt and video game around.

 

Enjoy, and thanks again for the kind words.

[atto 20]

I don't use indicators in my own trading, but come from a Math background, so this stuff is interesting. What was your sample size of the data you optimized for?

 

Mind posting today's as well? As you know, it's quite easy to cherry pick any technical indicator/setup so it looks amazing (not saying you did). Thanks.

[insideday 10]

@Urma

 

What is the highest timeframe for that calculation of the GV?

[UrmaBlume 10]

I don't use indicators in my own trading, but come from a Math background, so this stuff is interesting. What was your sample size of the data you optimized for? Mind posting today's as well? As you know, it's quite easy to cherry pick any technical indicator/setup so it looks amazing (not saying you did). Thanks.

 

1. Sample size was the most recent 50,000 8k day session bars.

 

2. Today's charts are posted in 3 different time frames, not as smoth as yesterday - but still plenty of room to make a buck.

 

3. I didn't cherry pick, I only posted the most recent data as you can see from the time stamps on both the post and the chart, I am PST.

 

cheers

 

 

PS - these are the charts we use for training - not for trading.

 

 

 

 

 

 

 

 

 

[Truckload 10]

Wow, this is great work UrmaBlume. I just found this forum today and this is my very first thread I read and BAM! I am completely blown away. I also use Tradestation and have never seen anything like this on their forum. My thanks for your sharing these ideas!

[UrmaBlume 10]

Wow, this is great work UrmaBlume. I just found this forum today and this is my very first thread I read and BAM! I am completely blown away. I also use Tradestation and have never seen anything like this on their forum. My thanks for your sharing these ideas!

 

Thanks for the very kind words. You will find posts on the TS forums from UrmaBlume1-4 bitching about the granularity of their time stamp.

 

I have been buying tech products from Bill Cruz since long before Trade Station, Omega Research or Super Charts.

 

They will get their time stamp together and it will continue to be the package of choice. In the meantime thank goodness for dll's & the window kernel.

 

cheers

Edited July 17, 2009 by UrmaBlume

[ramora 10]

The charts look great. They do what you describe very well.

 

However, one question does come to mind. You are using a very good moving average, possibly Jurik or T3, in your ES chart. Going to the work of calculating 24 different bias does not give you much of an advantage over simply using your tight moving average crossover. In cases where your bias indicator would make you money you could find a case where the tight Moving Average performed better in a trade.

 

This is a question I have to ask over and over to myself. Complexity for the sake of complexity, or does the complexity help in the trading?

 

I am sure you understand this question. The charts are impressive when posted. However, if you broke out a Jurik MA in a separate panel it would also appear very similar.

 

So, my question, "does the 24 bias indicator perform better than a tight fitting MA such as Jurik or T3 (not SMA or EMA)"?

 

I am asking because there are few people who would share their thoughts as you have.

 

Thanks again,

[UrmaBlume 10]

The charts look great. They do what you describe very well. However, one question does come to mind. You are using a very good moving average, possibly Jurik or T3, in your ES chart. Going to the work of calculating 24 different bias does not give you much of an advantage over simply using your tight moving average crossover. In cases where your bias indicator would make you money you could find a case where the tight Moving Average performed better in a trade.This is a question I have to ask over and over to myself. Complexity for the sake of complexity, or does the complexity help in the trading? I am sure you understand this question. The charts are impressive when posted. However, if you broke out a Jurik MA in a separate panel it would also appear very similar. So, my question, "does the 24 bias indicator perform better than a tight fitting MA such as Jurik or T3 (not SMA or EMA)"? I am asking because there are few people who would share their thoughts as you have. Thanks again,

 

 

The quick answer is no.

 

It depends on the data stream that is the subject of the function. In the OP I mentioned price, net new commercial trade and the balance of trade. I also mentioned MACD, ROC and the weighting of inputs from different time frames.

 

There are many different inputs to these indicators and most of them are not based on price so a quick Jurik is not the answer.

 

Weights, pre-processed inputs, multiple time-frames can't be answered with a speed to a tight Jurik of price only.

 

One point of note from the charts is that there are Jurik averages of price on all the upper graphs and none of them turn before price turns. If you look at the indicators below price on all three charts you will see that oftentimes the indicators do, indeed, turn before price.

 

Plus it's casual Friday and I am a bit loaded so - hasta manana.

 

cheers

Edited July 17, 2009 by UrmaBlume

[trade-samarai 10]

Yes, I noticed that sometimes your indicator leads price. It sounds like many man-hours went into building this indicator. Very interesting.

[UrmaBlume 10]

Yes, I noticed that sometimes your indicator leads price. It sounds like many man-hours went into building this indicator. Very interesting.

 

Thank you for the kind words.

 

Below is a shot of the last 3 hours of today's (08/03) trade on a 5k ES chart with 7 trade signals. Times are PST.

 

This graph is mostly a period of chop and yet still well defines several points of opportrunity. This is the result of a feed-back loop that adjusts to different levels of bias and trader commitment.

 

We don't believe that the passage of time motivates changes in price but rather it is an imbalance in commercial order flow that produces changes in price and these weighted indexes of biases are one way to measure that force as it is applied to price.

 

[trade-samarai 10]

"We don't believe that the passage of time..."

 

Yes, I noticed that you're using 5k share volume bars. It also frees space on your chart since no volume bars are needed.

[Frank 10]

hi,

 

just checked out this thread and had somewhat related question to the idea above. sounds like your index is reading order flow that is weighted for different (higher) timeframes so that you are coming at the market from multiple different angles despite showing just 1 chart at a time. so in the case where you show different lower timeframe charts -- would 'the index' then reflect higher timeframes relative to the chosen lower timeframe or are the higher timeframe global variables locked on absolute levels ? hope that makes sense.

 

I guess what I am really getting at is --- what are your primary higher timeframe intervals that really drive the model?? ie, if you were to rank them -- which would be the top few? ie, 25k volume - 50k ??

 

We don't believe that the passage of time motivates changes in price but rather it is an imbalance in commercial order flow that produces changes in price and these weighted indexes of biases are one way to measure that force as it is applied to price.

 

 

one other question- I agree that the 'passage' of time is not so important as the time of day function -- which you pointed out in another thread. so I am thinking that you either have to anticipate index trade signals as the more active time of day period approaches --- or does the index already reflect that and you still just wait for signals?

 

thx for any commments.

[UrmaBlume 10]

All of the traders in our small operation trade on workstations that drive 6 - 8 monitors such as the one shown below.

 

Bloomberg made a breadthrough with his screen when he combined different data streams onto the 1 screen. The multiple screens present higher time frame data for trade information and lower time frame data for trade execution - more precise entries and stops bases on lower time frame structures.

 

The weighted indexes are presenting information collected and procedssed from that and higher time frames. The chart shown is an 8k contract chart which is roughly equivalent to a 90 second chart. Some of our traders execute from that chart and our real speed demons, over 100 traders per session, execute from a 1k, 15 second chart.

 

We collect and process this information from these different time frames down to the execution chart to allow the trader to have one screen for information and one, the trade matrix, for execution. This way a trader can participate in 3 or 4 markets in the very short time frame all from the same workstation.

 

The bottom pic is of the last 5 hours of trade in the ES on Friday - times are PST.

 

 

 

 

Our "slower" traders execute of this 8k chart and our younger, mostly European, super stars execute from a 1k chart with the information parsed down to that, 15 second, time frame.

 

 

[N57H 10]

Hi

 

where can i find this indicator or workspace?

 

Thanks

[UrmaBlume 10]

Hi where can i find this indicator or workspace? Thanks

 

If you will note the title of this thread - it is designed as a primer so that you can write your own.

 

All of the concepts are here - all that is left is a bit of work on your part.

 

cheers

[Frank 10]

Starting point N75H,

 

This is the well-known 3/10/16 Oscillator 'Slow Line' run on a 7000 volume chart --- the time of the turns are each within 1-2 mins of the other chart.

 

Inputs: Price(Close), Avg1(3), Avg2(10), Avg3(16), StdDsp(1);

 

IF Currentbar>30 Then Begin

IF StdDsp=1 Then

Value1=Average(Price,Avg1)-Average(Price,Avg2);

IF StdDsp<>1 Then

Value1=Average((Price-(Average(Price,3)[3])), 2);

 

Value2=Average(Value1,Avg3);

 

Plot1(Average(Average(Price,Avg1)-Average(Price,Avg2),Avg3),"16Line");

Plot2(0,"ZeroLine");

End;

 

[N57H 10]

thanks

 

when do you know where to buy pr sell?

 

you need a cross over of the lines, isn't it?

[UrmaBlume 10]

Starting point N75H,This is the well-known 3/10/16 Oscillator 'Slow Line' run on a 7000 volume chart --- the time of the turns are each within 1-2 mins of the other chart. Inputs: Price(Close), Avg1(3), Avg2(10), Avg3(16), StdDsp(1);

 

Good indicators lead price not the other way around.

 

While it is true that the lines on price are zero phase implementations of Jurik's adaptive moving average, the index itself takes 24 inputs from 3 different time frames and most of those inputs have nothing to do w/price. We feel that neither time or price motivate or are predictive of future prices but rather it is the multidimensional imbalance in order flow that predicts price.

 

When you look at the chart with the moving averages you will note that price is to the left of the lines which means price is leading the indicator and not the other way around as a good indicator should.

 

Here is the same chart withe the index imposed over price and the MAs and you can see that the indicator does indeed lead price as it should. No averages of price can ever lead the market becasue changes have to occur in price BEFORE they can occur in the indicator.

 

Note that the index is to the left of price and the averages are to the right. I know of no price based indcator that LEADS price. This index does and thus is not made up of price based biases.

 

 

[N57H 10]

Could you please post this workspace with the indicators?

 

Thanks

[trade-samarai 10]

QUOTE: "We collect and process this information from these different time frames down to the execution chart to allow the trader to have one screen for information..."

 

So you you must be using a global variable (dll) since EL can't share data from different charts, correct? (Unless you've hidden the other time frames on your main chart.)

[UrmaBlume 10]

QUOTE: So you you must be using a global variable (dll) since EL can't share data from different charts, correct? (Unless you've hidden the other time frames on your main chart.)

 

There are more than a couple of ways that you can use EL and readily available free functions and dlls to share data from different charts.

 

As you begin to lean more about the functionality of TS & EL - I suggest you investigate the ADE - All Data Everywher ELD & dll as well as the EL Collections ELD & dll. While these routines are somewhat lacking in time granularity they will at least get you started in the right direction with their list & map functions.

 

While these collections are a bit slow and lack certain capabilites for our work they are free and available to the public and make a good starting point for those beginning to explore this area of opportunity.