Here’s this week’s positions for the option volatility model. I choose to use a time spread for the QQQQ’s and a diagonal spread for the S&P500 this week because both models predicted a rise in volatility for the underlying assets. Supposedly, these spreads should protect me from downside risk of an unhedged positions, only time will tell if this works out this week.
I would’ve expected differently but the trend in Real Estate bubble talk in blogosphere seems to be remaining constant. I suspect this is due to a lot of “I told you so” and sub prime related posts. What’s interesting is the nearly same volume of Stock Market bubble posts with crazy spikes. Is it possible that we might see a change in the quantity of stock market bubble posts in the future as the Real Estate bubble blogs close down and open Stock Market bubble blogs?
I started fooling around more with Rapidminer’s web mining abilities and managed to crawl my site and some other popular trading blogs. I’m still a long way from fully integrating this plugin into my data mining activities, but its been fun nonetheless.
Playing with this software renewed my interest in understanding web connections, so I visited a neat web applet I found a few years ago while surfing. This applet lets you see the web connections for a URL and can be found on the “websites as a graph” website. I typed in TraderMike’s web address and from the results, it looks like he’s pretty well connected! Note: The black dot is his root URL.
I was inspired by Taleb’s simulations on the hidden risks of “blowing up” as a Wall Street trader, so I built a
Here’s how I did it, please follow along and provide some critique, I’m not 100% sure if I did this correctly.
First I generated a random % return per year over a 5 year period, I did this for 1000 hypothetical traders and assigned a point value system to these random returns; 0 meant that you blew up before the first year was over (drawdown > 20%), 3 meant you were profitable for 3 years before blowing up, and 5 meant you made it to 5 years of profitability.
Then I calculated the mean and standard deviation for the point totals across the 1000 traders, modeled them over 10,000 observations, and produced a normal distribution curve.
The results were astounding!
You have roughly 1.42% chance, or 70:1 odds, of surviving 5 years as a Wall Street trader! Wow!
I really like the new poll widget I installed in the side bar, its been a fun interactive way to learn what my readers like! Now, I’d like to share my results for last week’s question, “What’s Your Favorite Post Topic on Neural Market Trends?“ The winning answer was: “Neural Nets” followed by “Yale/Rapidminer Tutorials.” Thanks to all who participated!
This week’s question is, “Do you think the Stock Market is in another Bubble?“Â Please take a moment to visit the sidebar and cast your vote!
I had a way better results in my virtual options account this week. My S&P500 short straddle worked out nicely and my overall account gained +0.5%. However, the QQQQ volatility model seems to be wrong more than its right so I’ll have think about how to tighten that up.
The 10 Year T-note yield closed at 4.95% on Friday, way below its previous high of 5.31% set on July 6th. Right after the 4th of July holiday, I posted a chart about the USDJPY carry trade. I created the chart after I modeled USDJPY, 10 Yr T-note yield, S&P500, and CRB index data using a Multilayer Preceptron neural net leaner. Later, I analyzed the data again using a Gaussian Regression learner and created a different chart altogether.
It looks like the action in the currency pair this past week, and the two weeks prior, seem to confirm the Gaussian Regression learner chart. As 10 year T-note yields retreat, so does the USDJPY.
Note: 10 year T-note yields began to rise sharply between July 1st and July 6th and marks the last USDJPY high approximately on July 6th. Previously, the last 10 year T-note yield high, of 5.24% on June 12th approximately coincided with the last USDJPY high set on June 15th-ish. The reason why the USDJPY highs sometimes do not coincide exactly with bond yield highs is that the analysis is a multi-variate model and other inputs such as the CRB index and S&P500, can limit exact coincidence.
Despite this, the model is considered to be valid because of the correctly modeling the relationships between the data and observing the outcomes in the real trading environment.
There’s a pretty good stock strategy out there that says you should buy stocks at all time highs because the odds of them going higher is greater. Traders and investors say the reason behind its upward momentum is that the stock has entered new territory and the bears don’t know where to stop it. A poster boy for this phenomenon is Apple Inc. (AAPL).
Sometimes though this strategy doesn’t work, as in the case with ABM Industries (ABM). ABM has retreated off its all time highs in May 2007.
I left wondering, what are the statistics out there that support this strategy? Has they ever been calculated?
Good Friday Morning! Welcome to another edition of linkfest!
Options expire tomorrow and if all goes well it will have been a profitable week, a lot better than last week’s option strangle debacle.
Here’s a quick snapshot of my positions, the short straddle is working out nicely but its very dangerous position. Short straddles, if unhedged, open you up to a lot of risk. Maybe next week I’ll try a diagonal position.
Another important building block of any neural net model is the creation of training and validation data sets for your model. The data you feed your neural net model is typically called “training data” and you use it to train the neural net model to learn the relationships from this data. The question then arises is, how do you know if the neural net is being trained correctly? Is it learning the right data relationships?
The way to overcome this problem and test the model as the neural net learns is to introduce something called validation data sets. A validation data set is just a random sample from your training data that is taken and then applied to the model. Once the validation data is applied to the model, the model calculates a predicted value. This predicted value is then compared to the actual data value and the error between them is determined. The neural net does this for every validation data point, adjusting the weights (more on this later) in the model each time to minimize the validation error. When the errors converge or can’t be minimized any more, the model has been trained.
YALE has a great operator called the Cross Validation operator that creates a validation data set on the fly for you. The Cross Validation operators allows you to tell just how much of your training data should be used for validation data and if you should use all the data (training and validation) to rebuild your final model.
Tip: YALE has a few other validation operators, explore them when you have time. This particular operator is useful if you want to check performance measures, which I suggest you do always.
The way to use it is to load it right before you place your neural net operators (more on this later) and after your data loading operator. Once you have it in your experiment tree, you can then tweak the parameters to your liking. Some important parameters for this operator is the “Create Complete Model”, “Number of Validations”, and “Sampling Type.”
The Create Complete Model parameter tells the Cross Validation operator to create a validation data set for testing and simultaneously use it to build the model. If left unchecked, the operator will only use the validation data for testing and essentially remove it from the training data set.
The Number of Validations parameter is just the quantity of data points you want to use to test your model with. If you have 100 data points, I suggest using 10%, or 10 data points for validation. If you have 10,000 data points, maybe 20%, it all depends on your comfort level and the complexity of the data your modeling. The last important parameter is the Sampling Type, this pull down menu allows you to choose how to sample your training data for validation data points. You have three choices: linear, shuffled, and stratified sampling (more on this later).
That’s it, another important building block explained. I hope that these smaller, but more detailed, tutorials are helpful to you. If they are, how about subscribing to my feed? As always, feel free to drop me an email or comment if you have questions.
What ever happened to Chipotle Mexican Grill (CMG) after I featured back in May 2007? It looks like it went to another all time high but has since retreated it a bit. I went there recently with my wife for the first time and she liked it. This surprised me because she barely tolerates Mexican food. If they can win my wife over, then they’re doing something right!
Howard figured out long time ago that CMG was doing something right and I know he’s tickled pink riding its trend up.
