Bought Phillip Morris International

Picked up shares of Phillip Morris International (PM) at 47.57. Looks like I jumped the gun with the afterhours Research In Motion (RIMM) purchase, as that's down another $5. Oh well.

RIMM Purchased

Research In Motion (RIMM) came out with estimate beating earnings, but their revenue forecast for the next quarter was shy of estimates ($3.6B-3.85B vs. $3.9B). Nevertheless, I am sticking with my original buy target as I used more conservative numbers to begin with.

Bought shares of RIMM in the afterhours market at 74.82.

APEI Purchased

Picked up some shares in American Public Education, Inc. (APEI) at 34.93.

Adding Few More Buy Candidates

Adding a few more names to the short list of buy candidates.

American Public Education Inc. (APEI) - ($35-36)
Charles Schwab (SCHW) - ($17-18)
Coinstar (CSTR) - ($30-31)
Devry (DV) - (sub-$53)
Edwards Lifesciences - (sub-$65)
Fortune Brands - ($40-42)
Lorillard (LO) - ($70-73)

Partial Sale of the Seal

Sold a block of Wet Seal (WTSLA) at the close at 4.00 even. Still plan on selling some more, but at the moment, there's no need as I haven't done anything aside from the small Arena Pharmaceuticals (ARNA) speculative purchase. So, this should be good for the time being.

Again, I want to make it clear that I have no current intent to sell my core position in WTSLA.

Speculative Purchase

Following up with my previous post... picked up a small position in Arena Pharmaceuticals (ARNA) at 4.90. It's a speculative position, so its size will likely be small in comparison with other purchases. Don't have much time right now to talk about that and others on my (soon to be growing) short list, but hopefully, I'll get around to it.

Haven't yet sold any of the Wet Seal (WTSLA) yet, but looking to sell at least a small piece today.

New Cash For Deployment

I recently added substantially to my largest position, Wet Seal (WTSLA), in the 3.20-3.35 range (ya, ya, I know... after-the-fact reporting, but whatever, everyone knows how large of a position it has been and continues to be).

As most of you know, over the last couple of years, I've held a large core position and have only traded in and out of a smaller trading position (hell, I started the Wet Seal Fan Club on Facebook a while back -- its administration has since been transferred to Wet Seal, and it has flourished with over 25K fans now). It looks like it's time for me to sell off the trading position and diversify my portfolio some. Well, the run up in its share price to over $4 makes it pretty obviously imprudent for me not to sell off a part of the position. However, it will remain my largest holding.

I know I haven't talked too much about stocks... I've still been very active, but I just blogged less about it since people seemed to be less and less interested. But, based on what a few friends have talked to me about, looks like interest is starting to pick back up. So, maybe I'll have some time to blog about stocks and the markets again.

I've updated the my portfolio holdings on the Yahoo ticker widget. It's a shame I sold off that Kongzhong (KONG) so cheaply way back when. Heh.

Anyway, here's a short list with a rough target buy-in price that I'm still working on... I will probably choose mostly from this list when it comes time to deploy the WTSLA cash.

Amylin Pharmaceuticals (AMLN) - Speculative - (sub-$13)
Arena Pharmaceuticals (ARNA) - Speculative - (sub-$5)
Gulf Offshore (GLF) - ($31)
Linear Technology (LLTC) - ($25-26)
CVS Caremark (CVS) - (sub-$34)
Gamestop (GME) - (sub-$25)
Phillip Morris International (PM) - ($47-48)
Chevron (CVX) - ($68-70)
Research In Motion (RIMM) - ($76-78)

I'll update if and when I do sell some of the Wet Seal, and when I do put some of that cash to work.

Any thoughts appreciated, public or private.

Careful with Floating Point Sums

Floating point computations are imperfect in the computing world. Often times we overlook this fact, so it's good to refresh our memories every now and then.

A simple summation loop such as this one can produce results with large errors.

int i;
float fSumFloat = 0.0f;
double fSumDouble = 0.0;
const float fIncrementFloat = 0.0001f;
const double fIncrementDouble = 0.0001;
const int nIterations = 25000000;

for( i = 0; i < nIterations; i++ )
{
   fSumFloat += fIncrementFloat;
   fSumDouble += fIncrementDouble;
}

You would expect that summing 25 million 0.0001's would result in 2500, but depending on your machine the results will vary. Here are results of running the above on my PC (with all optimizations turned off). You'll see that there is some error even when using double precision floating point.

Sum (float) = 2048.0000000000
Sum (double) = 2500.0000004718
Expected Sum = 2500.0000000000

Now that I have your attention, there are ways to combat the roundoff errors that ultimately kill us. One popular method is the Kahan Summation method. Another method is known as pairwise summation, where you continually sum pairs of terms until all you have left is the final sum. For example, start with 1 + 2 + 3 + 4. The first pass yields you (1+2) = 3 and (3+4) = 7. The second pass gives you your desired sum of (3+7) = 10.

Both of these summation methods are slower than naive summation, but they are available if needed.

One thing that I found a bit interesting is that the results of these two methods were different on my PC as compared to a Linux box. I'm pretty sure that it has nothing to do with the OS, instead it's probably due to CPU and compiler differences.

On my computer at home:
Sum (float) = 2048.0000000000
Sum (double) = 2500.0000004718
Actual Kahan Sum (float) = 2048.0000000000
Pairwise Sum (float) = 2499.9998779297
Expected Sum = 2500.0000000000

On a Linux box:
Actual Sum (float) = 2048.0000000000
Actual Sum (double) = 2500.0000004718
Actual Kahan Sum (float) = 2500.0000000000
Pairwise Sum (float) = 2500.0000000000

I also noticed that when I had compiler optimizations turned on, the results coming from my PC were definitely better.

Actual Sum (float) = 2499.9999368447 [Compiler Optimizations Turned On]

Anyway, the bottom line is if you're going to be dealing with floating point calculations (and, summations in particular), be careful and run lots of experiments. Even when the code you've written is highly portable, you will want to make sure that results will be as you expect on your target environments.

San Francisco, Executions, and Creola

On a whim, we headed out to San Francisco for the day. We started off at Sophie's Crepes in Japantown. When we arrived at around 11:30, it had not yet opened. This was strange, so we poked around Kunokuniya, the Japanese bookstore nearby, for a bit. Eventually, the crepe place opened its doors about half an hour later, and we munched on some of their goods.

I then had a craving for some seafood cocktails, so off we went to Fisherman's Wharf. After having some crab and shrimp cocktail from one of the many seafood stands, we checked out Musée Mécanique, which was this odd museum of old school coin-op machines. It was here that I ran into an old high school friend who just happened to be visiting the area... crazy randomness, really.

Anyway, the museum was pretty strange. One thing I noticed in particular was that there were three different coin-op machines that showed off executions; two were British, and one was French. I was curious enough to pay a quarter to watch the French Execution. Let's just say that it was not worth it at all. Here's a picture of the three execution machines, and a video of the French execution. I mean, seriously, couldn't they have extended the show a bit?




After the very short execution, we hit up a bar called Ireland's 32, and it was nice. Since it was mid-afternoon, the place was a bit empty, but the bartender was good and it seemed like it would be a pretty happening place during prime time.

Of course, we could not miss out on some O'Keeffe's, which I posted about in the past. That was a good time, just like the last time. We shot some pool, played Liar's Dice with the bartender, and had a few drinks. Definitely a place to check out if you're ever in the area. Too bad we live nowhere near.

We then tried to get some Korean food over at the well-known and respected Brothers restaurant, but when we got there the wait was too much for us to handle. So, we opted to go to a place I had been in the past out in San Carlos called CreoLa. The food was as good as I had remembered. I felt that my dish could have used a few additional garnishes to make it more pleasing to the eye, but as far as taste goes, there was no complaint. One thing though was that the kitchen was a bit slow, but at least the service was good.

I'll stop here, but before I close this out, here are some pictures from dinner.

Fried Green Tomatoes with Gulf Shrimp

Crab Cake

Blackened Redfish with Dijon Mustard

Pork Tenderloin

Peanut Butter Mousse and Lite Chocolate Mousse on a Vanilla Wafer Crust

Filipino Food

I went out for Filipino food a while back, and I never got around to putting up any quick post about it. Anyway, we went to Pinoy Lechon BBQ & Grill in Morgan Hill after reading good Yelp reviews. I personally like Filipino food, and I liked the two pork dishes we ordered, Lechon Kawali and Pork Sisig. The food's not the healthiest, but it was all very tasty and satisfied my desire to have something a bit greasy that afternoon.

If you ever have any craving for Filipino pork dishes and you're near Morgan Hill, I'd say give them a try. I'll definitely go back and try some of their other dishes. Also, the prices were very reasonable, so that's a plus.

Here are a couple pics.

Pork Sisig

Lechon Kawali