We have made tremendous headway on True Reckoning for Crude and 6E.

Will keep you posted on our progress…in the meantime, here’s the 8th Wonder of the World.

If you think that you might like to take …sceeto’s True Reckoning for NinjaTrader for a test drive, you can register for a 14 day free trial here.

<p>Please take some time to study and digest the information in this chart.</p><p><a href=”http://sceeto.com/components/com_wordpress/wp/wp-content/uploads/2012/03/03-29-12-Chart-examples-1.png”><img title=”S&amp;P 500 Exploratory Data Analysis | 30-MAR-12 | Courtesy of IOAMT” src=”http://sceeto.com/components/com_wordpress/wp/wp-content/uploads/2012/03/03-29-12-Chart-examples-1-300×242.png” alt=”S&amp;P 500 Exploratory Data Analysis | 30-MAR-12 | Courtesy of IOAMT” width=”300″ height=”242″ /></a></p>

We have launched a new build of …sceeto’s True Reckoning.

If you think that you might like to take …sceeto’s True Reckoning for NinjaTrader for a test drive, you can register for a 14 day free trial here.

The charts that show the Gaussian Filter (recent examples here, here, and here) are generated by Billy Duryea of IOAMT.

Billy does an amazing job of integrating what may look like disparate pieces of information, but are really inter-connected pieces of the same market puzzle.

I reached out to Billy this evening to get his insight into his approach for Gaussian and he popped right back with the following valuable insight:

“Both the Gaussian and the super smother have a look-back window of 9 sample.

The combination of the gaussain and the super smother create the k-mean.

In data mining, k-means clustering is a method of cluster analysis which aims to partition n observations into k clusters in which each observation belongs to the cluster with the nearest mean. This results into a partitioning of the data space into cells.

The problem is computationally difficult), however there are efficient algorithmsthat are commonly employed that converge fast to a local optimum.

These are usually similar to the expectation-maximization algorithm for mixturesof Gaussian distributions via an iterative refinement approach employed by both algorithms. Additionally, they both use cluster centers to model the data, however k-means clustering tends to find clusters of comparable spatial extent, while the expectation-maximization mechanism finds clusters larger data sets

Suggest google’in K-mean mathworks statsistcs toolbox
http://www.mathworks.com/help/toolbox/stats/kmeans.html

Wolfram K-mean clustering algoritym
http://mathworld.wolfram.com/K-MeansClusteringAlgorithm.html“

Thanks Billy…your input and insights are always greatly appreciated.

If you are interested in taking S&P 500 True Reckoning for NinjaTrader out for a spin, try a 14 day free-trial by signing up here.

The challenge with Momentum indicators

Though Momentum indicators can be used to identify new tradable trends, their challenge is that they cannot identify a new trend until that trend is well formed and has run significantly.

This is so because the increased momentum of a new trend can only be measured after an initial reading of momentum has been taken in the direction of the new trend.

This is the basic flaw of looking for increased momentum.  In order to have increased momentum you must first have momentum to compare it to.

In the real trading world, where time is money, by the time increased momentum is realized the trend is already well on its way.

Thus, though increasing momentum does a good job of confirming a trend, it does not do a good job of providing timely entry signals.

To counter this lagging deficiency, some traders look for trends to end by measuring decreasing momentum.

Using decreasing momentum to infer when a trade may be ending is a very handy method, but it also has core flaws.

Specifically, no matter how much momentum has drained from a move, no matter how ‘tired’ a trend is, in today’s electronic markets, where massive explosions of bots can be fired off, which they themselves excite and fire off other bots, all within 24 milliseconds, any ‘dying’ trend can instantly regain new life.

Hence decreasing momentum, though a handy tool in an earlier market age, and still at times viable today, can only tell you what happened.  It cannot tell you what is about to happen.

‘True Reckoning’ changes all momentum indicator’s inherent ‘lagging’ dilemma

True Reckoning eliminates the lagging nature of momentum indicators by doing a couple of things really, really well.

First, True Reckoning looks at more data than the typical Open, High, Low, Close data that a classic momentum indicator does.

True Reckoning analyzes each and every transaction in a particular market (not just the Open, High, Low, and Close of every bar).

True Reckoning also uses an advanced, proprietary ‘noise-to-signal’ filter algorithm tuned specifically for the unique market data that is generated by electronic markets.  This ‘noise-to-signal’ filter reduces the amount of ‘clutter and chatter’ in the market data that it processes and leaves a cleaner stream of information with which to analyze.

Once the market data is scrubbed by the ‘noise-to-signal’ filter, True Reckoning leverages a recursive data processing approach to hunt specifically for, and identify, a probable data ‘footprint’ that is unique to a significant portion of new trends.

When these new trend footprints are identified signal is surfaced.

True Reckoning was developed by Algo Futures’ Lab and is distributed exclusively via …sceeto.

The cobbler’s children have no shoes.

BATS goes Batty:

http://www.businessweek.com/articles/2012-03-23/the-bats-affair-when-machines-humiliate-their-masters

The software business is terribly complex.  Add managing massive amounts of real-time market data to the mix, and you are in an environment where good management controls of the software development life cycle, which includes a massive amount of very expensive and time consuming testing is essential.

Seems like the folks at BATS are not going to get any ‘Software Developers of the Year’ awards.