This blog post includes updates on my completed weight loss bet, my new six pack bet, school (check out the Mario project!), and poker. Comments are always more than welcome!
Fitness Bets Update:
If you read my previous blog entry (February) then you know that I had two 3-month weight loss bets with some friends. The first was with a friend who bet me that I couldn’t lose 15 lbs in three months; I crushed that bet, losing over 15 lbs in the first month alone! The second bet was more competitive, I had to lose more weight than another friend over the three months. We eventually decided that if both of us lost 25 lbs or more, we’d call it a draw. That’s exactly what happened, I lost about 27 lbs. Here is a before/after pic for the weight loss bet:
Check out the previous blog post to see most of the adjustments I was making. Still not fully satisfied with where I was — and, more importantly, not wanting to revert to old slobbish habits — I joined in a summer six pack abs bet with several friends. Whoever ends up with the “most improved” abs (determined by a panel of judges) after four months wins.
I’m really happy with the progress I’ve made and it’s nice to see that the graph still seems to be decreasing linearly (and not decaying exponentially). I know that I’ll eventually get diminishing returns but I’d be quite happy if I can continue to lose a pound per week ;). So far, I’ve seen quite the improvements:
- I’ve lost about 39 pounds
- I’ve dropped from pant size 38 to 34 (have to throw out over 11 pairs of pants!)
- I’ve dropped from M-L shirts to S-M.
- I’ve put on a lot more muscle mass and have become much stronger
- I’m not tired during the day and actually have much more energy
- I can think much clearer
- I don’t get heart-burn
- My cardio and stamina are much better
- I sweat a lot less
- I’m much more confident and happy with myself
Here’s a 5-month before/after pic:
Being someone who needs to see progress, often in some quantitative form, I decided to get a body composition test done for every two months of the new abs bet (at the beginning, middle, and end). I just completed the second test the other day and was very pleased with my progress: I lost 8.5 lbs (about a 1 lb/week), 2 inches from my waist (1 in./month), and 2.5-3% body fat. I also got four body fat estimates (from several formulas using different skinfold sites): 8.5%, 9.0%, 10.5% and 14.9%. They also give a rating for waist circumference and “sum of five skinfolds”, I improved from 4/4 and 3.6/4 to 4/4 and 4/4, respectively. I also lost fat from every skinfold site except for my chest (the girl performing the test didn’t know the reason but said the difference wasn’t significant).
For the weight loss bet, I was working out three days per week. For the six pack bet, I’ve switched to 5-6 days per week. Here’s what a typical week looks like:
- Weights – Chest and Biceps
- HIIT – 3 minute warm-up, 10 minutes of intervals (5 high, 5 low), 3 minute cool-down
- Walking – 15 minute walk to University and 15 minute walk back
- Ab Ripper – 240 reps total of 12 different ab exercises (uppers, lowers, and obliques)
- HIIT – 4 minute warm-up, 20 minutes of intervals, 4 minute cool-down
- Walking – 30 min.
- Squash – 1 to 2 hours
- Weights – Shoulders and Triceps
- HIIT – 3, 10, 3 min.
- Walking – 30 min.
- Ab Ripper
- HIIT – 4, 20, 4 min.
- Walking – 30 min.
- Squash – 1 to 2 hours
- Weights – Back and Legs
- Walking – 30 min.
- Ab Ripper
- Cardio – 25 min.
- Swimming (I finally started using the pool in our apartment building)
I learned the “Ab Ripper” workout from the girl who administered the body composition test; she modified it from Tony Horton’s P90X Ab Ripper X routine. I’ve since purchased the P90X program from GNC so that I can try it when Melissa and I move to our next destination (wherever that may be).
For weight-lifting days, we perform 3-4 exercises for each of the two listed muscle groups. For each exercise, we do 3 sets of 12 reps (or until muscular failure) with increased weight on each set. We normally do the high intensity interval training (HIIT) on elliptical trainers or exercise bikes (low intensity of about 65% maximum heart rate and high intensity of about 85-90% of max HR).
Our workout was adopted from the Lean Muscle Plan found on SixPackNow (yes, I signed up to one of these gimmicky-looking websites…), a website which turned out to be pretty legit. They gave me a 200+ page pdf book covering diet, stretching, exercise, weight-lifting, and everything you need. They also email you a personalized nutrition plan a few days after signing up (you give them information about yourself and a current body shot photo).
In addition to the workouts and squash, I’ll also be playing ultimate frisbee on Mondays (if we get the minimum number of girls) and slo-pitch softball on Tuesdays or Wednesdays.
Well, I finally finished taking all of my MSc courses (I just wrapped up “Introduction to Reinforcement Learning” with one of the field’s pioneers, Rich Sutton, as my prof). The course ended up being a lot of fun, it kind of changed the way I think about life (in terms of maximizing future reward, the crux of RL). The main reason for the course being so fun was the option of working on Infinite Mario for the course project(s). I obviously chose to work on it for both course projects (the first on my own, the second with a friend, Rich, from the poker group). It pretty much took over my life near the end of the semester, I was probably spending 8+hours a day on it.
The goal of the Infinite Mario domain is pretty simple: get the Mario agent to learn how to play the game by maximizing his future reward. Mario often gets reward for things like collecting coins, killing enemies, and finishing levels. On the other hand, Mario often gets penalties for things like falling down a pit, getting killed by an enemy, or taking too long to complete a level. Mario has to learn what is good or bad through his experience with the environment. That is, he has to play through the level several times and eventually learns things like “Hey, I should not run right into that nearby Goomba, I usually get penalized or killed for doing so… maybe I’ll try jumping over him”. If you’ve ever played any Super Mario Bros. game (shame on you if you haven’t!), you’ve probably discovered through your own trial-and-error experience that running to the right and jumping often is never that bad of a strategy (at least in the side-scrollers 😀 ). It turns out that Mario also quickly discovers this on his own. Here’s a short video clip of one of our Mario agents in action (apologies for the quality; you can get a little better picture by clicking on the “HQ” button):
Rich (my partner, not our prof) and I decided to submit to the Reinforcement Learning 2009 Competition (open to anyone in the world) and placed second! Rich deserves most of the credit, though, as I was not able to help him out after we finished the course.
Regarding my MSc research, I recently submitted two 3-player limit hold’em bots to the 2009 Computer Poker Competition. It would be really good for my thesis if one of my bots wins. One of the bots was built by playing 43 million poker hands against two other copies of itself (learning from its mistakes over time). The other submission uses the bot just described as a base player when it’s 3-handed but substitutes in some heads-up (i.e. 2-player) experts in some select preflop scenarios. For example, in the situation where the button folds, experts could be substituted in for the small blind or the big blind. Six of these heads-up experts were used and, perhaps somewhat surprisingly, this corresponds to them being used for about 70-90% of all hands. Anyway, we’ll get the results of the competition in about two weeks, I can’t wait!
I received an email the other day from one of my physics supervisors from UPEI, he informed me that some of the work I did in my last year at UPEI has just been accepted as a publication submitted to the Journal of Chemical Physics! Here’s a link to the abstract and here’s a link to the pdf version of the paper.
Now that all of my courses are finished and I’ve created some competitive 3-player bots, I’m just writing my thesis and collecting final results. Unfortunately, all results will likely be empirical. I’d like to have some theory in there but multiplayer theory is quite difficult and my background in the area is minimal.
I have a few other bets that I haven’t mentioned. I have two other $50 six pack bets with other friends (one who is already in our group six pack bet, and the other one being Brodie Champion). I also have a sleep bet where I must get at least 7 hours of sleep every night. This was really tough for me as I was a night owl who would often be up until 1 or 2am. Now, I go to bed at 11:30pm and wake up at 6:30am and then hit the gym at 7am every weekday. Finally, I have a $40 thesis-writing bet with a friend where I have to have my thesis submitted by the third week or so of September. I plan on having it submitted and defended by this date, though, so hopefully everything works out.
I’ve purchased a few Wii games since the last blog post: Boom Blox, Boom Blox 2: Bash Party, EA Sports Grand Slam Tennis, Tiger Woods PGA Tour 10, Mario Power Tennis, and maybe a few others. Most of the games mentioned are made by EA and are great! Grand Slam Tennis and Tiger Woods are especially cool as they use the new Wii Motion Plus controller add-ons. There are still a few bugs to be worked out and the “1-to-1 correspondence” is a bit of a fib, but they are great multiplayer games nonetheless.
When the previous blog was written, I was likely halfway through a UofA CS grad student competition called the “Beard-Off”. I entered the month-long “Most Growth” competition and ended up finishing as the runner-up :-(. I really thought I had that competition in the bag but unfortunately, I’m only the second manliest grad student in the CS department. Oh well…
Melissa and I have had a few visitors here in Edmonton. Melissa’s brother Bobby came out for a week or so and that was really fun. We ended up drinking and dancing (who’da thunk?) and showing him around a bit. I also had Theriault, Arlo, and a friend of theirs, Dave, come out to visit for a weekend. That also involved drinking and dancing (again, who’da thunk?). That was about a month ago and I lost my new jacket at the bar (took it off to dance and forgot to pick it up when we were leaving).
Another exciting note is that my good ol’ pal, and one of my groomsmen, McWado (name rhymes with Mike Wazowski), has since moved out to Edmonton to do his Master’s in Bio (and is living with my awesome workout partner, Adam). It’s been a lot of fun having him around this past month. We’ve been playing squash together on Tuesdays and Thursdays, working out and swimming on Saturdays, and playing video games on other random days. We also started a fun little “tradition” where we’re going to make steak and potatoes on Saturdays after our swim. We’re currently trying different steaks every week; so far we’ve tried New York Strip and Cross Rib. We’re thinking T-Bone next week. Any suggestions (including marinades, basting, grilling, etc.) are welcome!
Finally, Melissa and I have sponsored a very cute, four-and-a-half year old, boy from Nicaragua named Norgen A. Molina Calderon through World Vision. I used to think it was a bad idea to give to these bigger organizations which generally take larger amounts (~18%) out of each donation for fund-raising and administration costs. However, I don’t see it as such anymore. If you look at the total amount (82.1% of all donations) given to programs that help the children, I’m sure it’s much much higher (in terms of absolute dollars) than the organizations that give 95-100%. Without the advertising, many of the donators simply wouldn’t know about such a cause (or at the very least, think about it on their own accord). For instance, Melissa and I were shopping in the West Ed Mall and noticed a table containing portraits of these poor children. Beside the table were two (most likely paid) World Vision representatives who informed us about their program. Had these women not set up camp in the mall (which requires both fund-raising and administrative costs), we would likely have never thought about donating. Now that I have more experience with both poker and Reinforcement Learning, I think much differently about these sorts of scenarios (involving investments or future rewards).
Poker blog begins here… if you came for the fitness update, you can stop reading.
FTPoints: 316,902.4 (surpassed my 300k goal for the year)
Iron Man Medals: 1684
|Full Tilt Poker
I’m pretty disappointed, I’ve been saving up my FTPoints for the last two years and set my goal for 300k points this year: the number of points required for a 30″ Samsung computer monitor. I finally reached my goal (halfway through the year!) and then they took the monitor off their store! Oh well… I still have enough points for a new 13″ MacBook, or 3+ 24″ monitors, or close to enough for some large HDTVs. Also, I believe the value of my Iron Man medals is a little over $300 (they can also be traded for more FTPoints).
Even though I said I wanted to stop withdrawing from my poker bankroll, I ended up taking out over $2k. I got Melissa a really nice 12.2 MP digital SLR camera, the Canon EOS Rebel XSi, and lens for her birthday. I also paid for her trip to NYC (as part of the deal I mentioned in my previous post) where she is currently visiting with friends. I miss her already :’-(.
Since the last blog, I’ve played 48k hands of poker. Here are my overall stats:
I haven’t run very well (0.3 BB/100) over this time period (start of February to end of May). I’m starting to think this might be due to how loose I’ve been playing lately (it’s getting eerily close to 40/30). Although I feel like I have a big edge on most players in my games, I think you really need the right game conditions to be playing that loose; it probably shouldn’t be a default style.
When I get time, I plan on revamping my game (by reducing the number of tables, thinking a bit longer before making decisions, reviewing hands, and studying situations with poker software tools (like Pokerazor, StoxEV, StoxCombo, Pokerstove, Poker Equilator, and Holdem Manager). I’ve already spotted several leaks in my game and am creating some preflop charts (based on empirical thresholds from Stoxtrader’s awesome book).
Notice that I’ve taken some shots at $8/16 and $10/20 and they’ve gone well! I actually made an appearance in one of Bryce’s Stoxpoker videos at $10/20 (called something like “10/20 LHE Adventures”) where I proceeded to crack his pocket aces ;). I have enough of a bankroll now to play $10/20 (using the 1000 big bet rule) so I’m quite the nit to be still mostly playing $5/10. I’m okay with that though as the money at $5/10 is still pretty good, lower risk, and lets me work out some of the leaks I’ve discovered.
Here are my session stats over the same time period:
It took about 154 hours to play those 48k hands. Even though I didn’t run very well (in terms of BB/100), I still did well in terms of an hourly: $50US/hr ($27/hr from play + $23/hr from rakeback). Rakeback is now being paid to my FTP account on a weekly basis instead of to PokerSource. I’m so glad since I’ve had so many problems with PS; the only downfall is not being able to easily get Amazon gift certificates to buy books, games, and DVDs.
Here are my position stats filtered for 5-6 players:
My 6max stats are 38/28/1.9, I’m becoming such a LAGtard :-o. I didn’t notice until today, but it looks like my big blind play is getting worse (in terms of winrate). I’m probably defending a bit too much right now (at least for these high-rake games) so I might tighten up a bit on my defense. I have created some defense charts so I’ll have to organize those a bit and start following them. Notice that I’m running over 1 BB/100 when filtered for 5- to 6-handed (my goal for the year, a pretty respectable winrate in today’s games). This suggests that I’ve got some short-handed and HU leaks. I’ll have to look into that.
Here’s a graph for the hands I’ve played (in BBs):
Anyway, congratulations if you made it this far and thanks for reading! As mentioned, comments are always welcome!