Well all the news about the PIIGS taking wing in Europe have rattled the EURUSD pair.  I think it might be heading down to test the $1.25 level and I'll be keeping a close eye on it, since I'm about to start trading Forex in earnest again.

I have my models all rebuilt and my system ready to go, now its just a matter of moving to a new broker and setting up Metatrader 4.  I'm going to start fooling around with Metatrader as well as testing some expert advisors because rules based trading appeals to me and my neural nets.

I'll be starting a NEW $100 Forex Experiment and seeing how quickly I can build it up or blow out!

The results are in this morning and the prediction was wrong this time!  Last Friday’s historical volatility close was 0.1444, DOWN from 0.214090  the Friday before that.  For the last two predictions the model made, I had one correct and one wrong.

For this week’s prediction, you have to read my S&P500 Option Volatility tweet (you can follow me on Twitter here), I will continue to keep score on Monday mornings to see if this model truly is 60% accurate for predicting the direction of volatility.

As I wrote in my last S&P500 Option Volatility Report, I predicted that volatility in S&P500 options would close higher on this past Friday than the Friday before that.  The results are in this morning and sure enough the prediction was correct!  Last Friday’s historical volatility close was 0.214090, UP from 0.213821 the Friday before that. 

For this week’s prediction, you have to read my S&P500 Option Volatility tweet (you can follow me on Twitter here), I will be keeping score on Monday mornings to see if this model truly is 60% accurate for predicting the direction of volatility.

I know that this is late but I just dusted off my S&P500 option volatility prediction model.  For this week only, I will be posting my S&P500 volatility predictions on my blog, after that I will be posting the predictions on Monday mornings via my Twitter account. 

You can follow me on Twitter here. 

FYI, this model is correct roughly 60% for the direction of volatility only, its no good for actual magnitude. 

Some strategies to use in cases where the index option volatility is predicted to increase would either be a long straddle or long strangle.  In cases where the volatility is predicted to decrease you’d consider using a short straddle or short strangle instead.  60% odds are very good and with good money management you have a great edge!

Here is the prediction: For the week ending 7/17/09, a historical volatility (HV) predicted close for the S&P500 is 0.2736, UP from last week’s close of 0.2183. Remember, the direction is what counts in this model (UP or DOWN), not the actual HV number.

I wanted to take a moment and say thanks to Tibor, one of my 12 readers, for forwarding me some really interesting research papers on modeling NFL and NBA games with neural nets. There are some really good nuggets of information in those papers, especially the discussion on setting the right momentum and learning rates.

I used that information to fiddle around with my neural net model and I’m posting some recent results from my DRAFT NFL neural net point spread model.  As you can see, the model is pretty good at determining if the home or visiting team will win (a negative sign means the home team wins) but the predicted spreads are way off relative to actual spreads.

This leads me back to developing some sort of ranking system to feed the model, which I wrote about in my "Thoughts on Ranking Football Teams" post.  The good news is that the research papers that Tibor sent me allude to a type of football match system where the model learns the results of previous games and then applies its statistical analysis to new match ups.  Despite this good nugget of information, I feel that I have a long way to go to get something solid before the season starts.

In the interest of science, and because I love my 12 readers, I’m uploading my EasyNN Plus data file for this particular model.  However, you’ll have to have the full version of EasyNN Plus to use this file because the model uses 980+ example rows and the test version only allows you 100 rows.  If you follow the link above and buy EasyNN Plus from there, I will get a small commission from Steve.

I’ve made some serious headway last week in analyzing NFL football data to model game point spreads.  I was able to determine with great accuracy (84%), using backfitting, what team would win a matchup. I did this by building a backpropogation model from 2007-2008 data with about 20 matchups as the prediction set.  The only bad thing I discovered was that the point spread prediction was way off.

As I wrote before, backfitting is NOT the way to go for building this type of  model and the fact that the magnitude of the actual point spread was off bore that out.  You might be asking yourself right about now, "if this is not the way to go, why did Tom do this?" 

I did this because I’m still in my data discovery phase looking for relationships and tinkering around.  Now that I understand various data relationships and can detemine the winners relatively well, the next step, and perhaps the hardest now, is determining the ranking system.  I suspect that the ranking system will help get the game spreads closer to what they should be for future games.  If all goes well I should be making spread calls on this blog for the next season as a way to determine if my model is worth a sh*t.

I believe that we are fast approaching a bottom based on my market timing model and simulations.  I know I’ve said that before but these levels of panic can’t sustain themselves for long without some sort of capitulation.  Below is a superimposed weekly chart (from 1990 through last week) of the my market timing indicator relative to the S&P500.

It’s a pretty good indication that some sort of top (not necessarily thee top) was made if a rapid rate of change in the indicator occurs while simultaneously making a new high.  Usually at these times I like to skim some profits and put them in bonds or cash.

Likewise, if the indicator experiences a massive rate of change while simultaneously making a new low, you have a pretty good idea that you’ve made a new bottom of sorts.  The timing model can’t tell me if I’ve made the BIG top or bottom, so I have to rely on other models to tell what that probability is and gauge how much money to take out or put into the markets.

Granted, we are seeing things in this market that are unprecedented but my model seems to be coping farily well.  The scary thing is I just watch my indicator blow through the 3.0, then 4.0, and now approach the 5.0 level without blinking.  These markets are in a free fall and I’m feeling the pain!

So how far will she go?

A hastily drawn yellow trend line on the chart shows the S&P500 finding support around the 1000 level first and the model issued a major BUY signal at the level.  Unfortunately we’ve already blown through 1000 so the next level of support, from inspecting the previous BIG bottom in 2003 is around the 800 level.  Then using my Genetic Algorithm model and Monte Carlo simulations, I come up with my most probable BIG bottom.  That’s the 700 level, more than 50% off the all time market high.

Could it go lower than 700, you bet!  But I’m willing to dump more money into my 401k at that level and then hold on tight.

I’ve been busy rebuilding my Gold neural net model over the past few days with new data and inputs using the new version of Rapidminer.  It’s the same model I used to write my published Futures Magazine article and tinkering with it should give me new insight into the current Gold trend.  This is particularly important to me because I want to know if any future pull backs in Gold is a buying opportunity or something else.  I predominately use my trend models to find buying opportunities and this works great in Forex and Metals.

Many people I correspond with thought the Gold market was crashing when the price broke through $750 recently.  I seem to remember a lot option acitivity around the $600 level as well.  My old Gold model told me otherwise and I paid a visit to my Gold dealer earlier this month.  I’m curious to see if it will tell me to visit my Gold dealear again in October.

The one nice thing about Genetic Algorithmic stock market models is that they evolve with new market data. The bad thing about them is that they’re a pain in the ass to create.  My current GA model is surprisingly simple and works in conjunction with my S&P500 Volatility Timing Model.

Although its highly secret, what I can tell you is that I use the Genetik Solver Excel Addin.  I specifically use it because its free and integrates with MS Excel.  Its a bit clunky to use but if you read my tutorial on it, you should be able to learn it in no time.

I take a snapshot of market data in real time using the input variables from my volatility model and use the Genetik Solver to analyze those data points.  I make sure to grab the max and min data points for each variable, load it into the Solver, enter my population size, cross over probability, mutation, and then run it.

The answers I get from running this analysis helps me understand at what levels my input variables (to the timing model) should rise or fall to for things to return to a state of calm.  From this I can better understand the current market internals and  what things to look for as the market evolves in real time.

Right now my Genetic Algorithmic stock market model tells me that in order for the markets to return to a calmer state,  Gold has to drop (no-brainer) and T-bills have to increase in yield to the 2% level.  What wasn’t so obvious was that fact the technology sector must remain fairly flat and the Nikkei 225 has to rally like crazy.  If Gold sits tight, T-Bill yields weaken, and you have a rally in the tech sector, watch out!  At least for for this week.

:)

I went long the AUDUSD pair last night with the same intent to build a longer term position in it. My belief is that this pair will go to parity with the USD this year or early 2009. I’m still long EURUSD and sitting on a nice profit; I’ll be adding to it on pullbacks, which it looks like might be today.

For all my neural net aficionados, the trend remains up for both these pairs.

Here’s the current score:

• EURUSD +124 pips
• AUDUSD +7 pips

Trolling through the Internet last night I discovered that my Futures Magazine article from last year, Forex and Treasuries Provide Clue for Gold, has been republished over at Traderslog. Make sure you read it quick before it gets taken down! :).

My market timing model was in full swing today and it flashed a BUY. Today was a lot of carnage and I suspect that we could go even lower from here. I hope everyone has their stops in place, :).

Normally I wouldn’t move any money in the market until the low at 1276 on the S&P500 is taken out. Still, the market timing model is a nice short term trading model too.

If you went long when the model issued a BUY signal and then closed the position and simultaneously went short with the SELL signal flashed, you’d be sitting on some decent short term gains as my little chart shows. The danger in doing this is that the model was designed to issue BUY/SELL signals for a longer time frame and you could easily get a BUY signal and end up with negative returns in the short term.