Primer on the Formulation of an Index of Weighted Biases

[Tams 20]

note:

For those who want to dig deeper in programming...

 

Global Variable V2.2

http://www.traderslaboratory.com/forums/f46/global-variable-v2-2-a-6023.html

 

Collections for EasyLanguage

http://www.traderslaboratory.com/forums/f46/collections-easylanguage-5929.html

 

ADE - All Data Everywhere With TypeZero Library! (see post #11)

http://www.traderslaboratory.com/forums/f46/ade-all-data-everywhere-easylanguage-5934.html

[taotree 10]

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

 

I respect your posts a great deal. I don't think the above is quite accurate, depending on how it is interpreted. Yes, you explain a useful method in transferring higher timeframe information onto a lower timeframe chart. However, all the necessary information for a successful trading methodology are not in the original post

 

I would propose that the reason your bias indicator is so effective is because you have done an enormous amount of research, analysis, computation discovering information in the market data. It's that proprietary analysis that is giving you such a wonderfully valuable "indicator" on the lower timeframe.

 

Put another way, it's not that you have taken 24 random indicators from a higher time frame and transferred that to the lower. It's that you have taken 24 *very useful/effective* (and unfortunately for me, unavailable to me until I figure them out for myself) bits of information from the higher timeframes.

 

And I think you make this clear as you posted later in this thread:

 

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.

...

I know of no price based indcator that LEADS price. This index does and thus is not made up of price based biases.

 

So, as you say so yourself, grabbing your typical indicator information from the higher timeframe and applying that to the lower chart is not going to magically make all those lagging indicators any more useful.

 

One still has to dive into some "uncharted" territory (such as "multidimensional imbalance in order flow" as you call it) to continue the journey.

[MtnDog 10]

Thanks UB for sharing your valuable insights. I just recently found this forum, so i have much catching up to do.

 

A question regarding your weighted bias index. Can you give any info about what you are optimizing the weights 'for'? In other words, are you simply optimizing the weights for whatever combo results in the most profit? Or are you possibly using optimized weights that maximize the correlation to a momentum function of price?

 

It also seems to me that this would be best accomplished via some kind of neural net/genetic algo rather than the generic "out-of-the-box" optimizers included in most trading software.

 

Actually, a quick and dirty method would be to simply sum up the weights (-1,+1) of the 24 biases, forming an index that ranges from -24 to +24.

[UrmaBlume 10]

Thanks UB for sharing your valuable insights. I just recently found this forum, so i have much catching up to do. A question regarding your weighted bias index. Can you give any info about what you are optimizing the weights 'for'? In other words, are you simply optimizing the weights for whatever combo results in the most profit? Or are you possibly using optimized weights that maximize the correlation to a momentum function of price? It also seems to me that this would be best accomplished via some kind of neural net/genetic algo rather than the generic "out-of-the-box" optimizers included in most trading software. Actually, a quick and dirty method would be to simply sum up the weights (-1,+1) of the 24 biases, forming an index that ranges from -24 to +24.

 

MtnDog,

 

How very astute. You are absolutely correct.

 

Welcome to TL - smart gets a bias of +1 even before weighting.

 

Before we assign the +1 or -1 we select the inputs. After we select the inputs their parameters must be optimized. We optimize each input's parameters for each of 3 time/volume frames.

 

After the parameters are optimized we assign the +1 or -1 for each input for each bar

 

After the +1s amd -1s have been assigned we begin to optimize the weight of each. Our range of weights is .1 - 2.6. We optimize these weights in more than a few different ways: we optimize for the execution time frame for highest net profit, lowest drawdown, highest percentage of winners and many others including certain correlations.

 

We run these optimzations using both brute force and genetic, survival of the fittest, algorithms.

 

The application of Neural Networks and MARS (Multivariate Adaptive Regression Splines) happens AFTER these optimizations and is the part of our work that you will probably never see posted here. The indicators we have posted are primarily training tools for our in-house traders.

[ZOSO 10]

This is all very fascinating and begs the question, since the entire methodology is computer based: why do you need to train traders or even have any to possibly muck it up with emotion and discretion?

 

In other words, why not simply auto-trade all markets 24/7 or during whatever timeframes are optimal?

 

Lastly, throughout all your most generous posts, there are no indications of what sort of profits or success ratios all this hard work now generates. Care to enlighten us further?

Edited August 31, 2009 by ZOSO
afterthought

[taotree 10]

After the parameters are optimized we assign the +1 or -1 for each input for each bar

 

When you say "for each bar", do you mean that you have a + or - bias on an input for each bar given it's context? So in one context, an input might have a negative bias, but in another it might have a positive bias?

 

After the +1s amd -1s have been assigned we begin to optimize the weight of each. Our range of weights is .1 - 2.6. We optimize these weights in more than a few different ways: we optimize for the execution time frame for highest net profit, lowest drawdown, highest percentage of winners and many others including certain correlations.

 

I created a response function that assigned a value to every tick. It uses future data to determine what level of short/long interest I would want to have at that point in time. The idea is that then I could use that in any sort of supervised learning or fitting algorithm through which I might want to run the data. Basically, if you could have a crystal ball that would tell you one value at any point in time, what would that value be?

 

 

The application of Neural Networks and MARS (Multivariate Adaptive Regression Splines) happens AFTER these optimizations and is the part of our work that you will probably never see posted here. The indicators we have posted are primarily training tools for our in-house traders.

 

I took a glance at MARS. If I understand it correctly, one of the rather interesting things it does as a it finds thresholds or turning points in the value of the data. When working with continuous data based on which one must make a discrete decision, that makes a lot of sense to have something that will find those thresholds.

[zdo 20]

UB, Does your Index of Weighted Biases work across various instruments / markets or is it limited to SP / indexes. Thanks.

[UrmaBlume 10]

UB, Does your Index of Weighted Biases work across various instruments / markets or is it limited to SP / indexes. Thanks.

 

So far we have optimized for S&P, Treasurey Notes, SoyBeans, Crude Oil and certain ETFs with very positive results.

 

The caveat is that it must be re-optimized and reconfigured for each new market and takes about 4 days on 3 dual quad-core workstations to do the optimizations and reconfigs for each new market.

 

cheers

[clmacdougall 11]

This is really great stuff, thanks for even posting Urma. I've read every article and book there is relating to Market Profile and knew that it was just the start of being on the right track. The next step was to throw away most of it because it leads to nowheresville. But what's left is the beginning of market understanding.

Urma, could you please recommend some books or articles for the steps after Market Profile, and for an introduction to understanding the multi timeframe weighted biases approach that you are proficient in.

 

Thanks again

[UrmaBlume 10]

UB, Does your Index of Weighted Biases work across various instruments / markets or is it limited to SP / indexes. Thanks.

 

ZDO,

 

We have had very good luck in moving our work to different markets.

 

While we haven't yet moved these indices to other markets (Certain ETF's are next), our Intensity indicator transfers well to every market we have tried.

 

Here are four spikes taken from trade in yesterday's (09/08) trade in the 10yr Notes and an older pic of SoyBeans:

 

 

 

 

 

[UrmaBlume 10]

This is really great stuff, thanks for even posting Urma. I've read every article and book there is relating to Market Profile and knew that it was just the start of being on the right track. The next step was to throw away most of it because it leads to nowheresville. But what's left is the beginning of market understanding.

 

Urma, could you please recommend some books or articles for the steps after Market Profile, and for an introduction to understanding the multi timeframe weighted biases approach that you are proficient in.Thanks again

 

clmacdougall,

 

How very astute of you and what great words. My feeling about Market Profile and everything I learned from Peter Steidlmayer is exactly the same as yours.

 

Indeed my study of Market Profile formed the foundation of my market understanding and is the basis of almost all of our technical development.

 

Here is a thread that I started to report on and provoke thoughts on - the technical evolution of Market Profile Theory.

 

Also thank you for the kind words. As to book recommendations, I am in the process of finishing 2 that my publisher is nagging me for and I will be happy to recommend once they are done. "Practical Short Term Trading - Techniques & Technologies" will be out next. My first published book was on the odds, proabilities and practical application of game theory in poker. Several on this thread have already outed me as the author.

 

A couple of our in-house traders are old profile buffs and still leave one of thier 8 screens on this shot of 30 minute profiles. This shot is of the first 2 hours of trade in today's (09/09) trade in ES and shows the buying that has propelled price to new highs.

 

The Blue/Red Dots (Blue - Buying, Red - Selling), these are all Blue, report the Net Buying and Selling inside each of these structures as read from the Index on the Left of the chart. This pic is taken from a market report that is updated throughout the session.

 

Edited September 9, 2009 by UrmaBlume