Despite success in season long football leagues for the past few years, I was pretty unsuccessful playing DFS. I would deposit a few hundred bucks at the beginning of football, swing up and down a bit, and eventually lose it by the end of the year. I tried out basketball last season, but I knew basically nothing and lost some more money.

Roothlus (the poker player) convinced me that I should give NBA DFS another shot this season. It's more fun, predictable and beatable than other sports. For instance, if Lebron scores 20 points, 4 assists and 4 rebounds thats a pretty bad fantasy game while 30 points, 7 assists and 7 rebounds is a great game. the variance in game to game results is much smaller than other sports like football or baseball.

So at the beginning of the season I threw another $250 on both fanduel and draftkings with the intention of treating it just like poker when i started. I was essentially starting as a .05/.1 NL player and trying to work my way up without ever shooting my roll and using safe BRM.

It turns out DFS is just like poker. You play other people in a variety of cash games and tournaments and the site makes money off rake. The most common cash games are 50/50s (half the field cashes for 1.8x) and HUs (1 on 1 with another player). And tournaments work the same way as in poker, with 10-25% of the field receiving money with top heavy prizepools.

I tried to never put >20% of my roll into play on any given day and split my action between a few lineups to reduce variance. Fanduel went great right from the start. I was winning in cash games and I got 2nd place in a tournament for $1500 early on which allowed me to start playing a bit bigger. Draftkings...not so much. Basically lost every day there until my $250 deposit was gone. I didnt care too much because I could get all the action I wanted on fanduel and i was doing well over there.

Since that initial tournament cash, I have not cashed in another tournament for more than $700. Despite this lack of big scores, I have run up my roll on fanduel despite still not really being an expert on basketball. After the all star break, I redeposited $500 to DK and have run it up there, and started playing on victiv and am winning there.

Fanduel results since early november, provided by a really cool tracking site similar to PT and HEM built by liquidpoker's Panorama:

if you notice, I made a lot of my money in very low buyin 50/50s. its very easy to enter a lineup you like into tons of contests and just let it go to work.

Graph:

The tracking site is a little buggy for draftkings, but i've deposited $750 this season and have my roll up to $4k, and on victiv ive run a $1k deposit up to about $3k in the past week. So overall, including some bonuses and FPPs im up about 26k this season, all started from $500. I've basically grinded up from micros to midstakes in 4 months without any prior knowledge of NBA.

tips if you want to start playing:
-trust your projections. if your optimal lineup is 290 points, and a lineup you personally like better with players you trust more is projected at 275, throw it in the trash. It's ok to audible a pick over machine projections occasionally, but make sure your lineup is as close to the optimal projections as possible. In general I dont like to give up more than 1% in points for lineups i like. Those points add up over the season and they are where you get your edge.

-use good BRM and start smaller. most poker players who come into DFS just shoot it like they are playing the same stakes in poker/DFS, yet their DFS skill is much lower than their poker skill to start.

-play lots of smaller buyins. It's better to enter your 1 lineup into 100 $1 cash games than to enter it into two $50 cash games. players get better at higher stakes and the only downside to playing more lower stakes games is the time it takes to register them.

-a lot of edge in NBA is due to injuries. its a long season and rough sport, there are often tons of late scratches and random rest days. if a starting player goes down, and his cheap backup is slated to get 30+ minutes, he will often do well. minutes = money. stay on top of the injury news, know the depth charts/rotations, and who to plug in when starters miss a game.

-treat it like a job. i work all day on registering, tinkering with lineups, reading injury reports, figuring out where the projections may be off and which players i should trust/not trust, etc. it's a lot of work, but its not quite as intense as poker. from 9am to 2pm I basically spend 5-10 minutes out of every 30 minutes checking news, registering for more contests, tweaking lineups and the rest of the the time i am free to do other things like run errands, walk my dog, beat off, take a shower, whatever. then from 2-4 i am working really hard on lineups since lineup lock happens at 4pm PST, and then from 4-6pm i check if there are any changes i need to make for the sites (draftkings only for me) that allow changes to be made after the first games started. i dont really even have to watch the games but i do because basketball is a fun, high paced game with the most athletic people in the world.

if you want to sign up i suggest playing mostly on fanduel and draftkings. by far the biggest sites with the most players and largest contests offered.

theres a month left in the NBA season, and probably a 1-2 playable weeks of playoffs after that. i will be working hard for that time, and theres still time for everyone else to play too. and if NBA isnt your thing, baseball is starting soon, and there's also NHL, MMA, golf, soccer, and college sports. something for just about everyone.

good luck to anyone who plays and anyone who decides to give it a shot.

Posting this here for 2 reasons. First as a brag, thought it was pretty cool that I came across a problem with some real world applications and was able to improve on the solution for the simplest case. And second cause there are some really smart mathematicians here (catyoul for instance and others) who may want to run with my algorithm and apply it to the more general cases or come up with even better algorithms than I have.

Was doing my normal web surfing last week and came across the secretary problem, something I had never heard of before.

" imagine an administrator willing to hire the best secretary out of n rankable applicants for a position. The applicants are interviewed one-by-one in random order. A decision about each particular applicant is to be made immediately after the interview. Once rejected, an applicant cannot be recalled. During the interview, the administrator can rank the applicant among all applicants interviewed so far, but is unaware of the quality of yet unseen applicants. The question is about the optimal strategy (stopping rule) to maximize the probability of selecting the best applicant. If the decision can be deferred to the end, this can be solved by the simple maximum selection algorithm of tracking the running maximum (and who achieved it), and selecting the overall maximum at the end. The difficulty is that the decision must be made immediately."

When I read the solution I was at first impressed by it's mathematical beauty: Reject the first n/e applicants, choose the next best applicant after that or choose the last one. this gives a probability of 1/e of delivering the best applicant.

But on second thought it seemed rather risky. When the best applicant occurs in the first n/e spots, the last applicant is always the one chosen. Which means ~36.8% of the time you get a random choice between the worst and 2nd best applicant. It's likely that the applicants are normally distributed about an average number, and while the best applicant may be a standard deviation above the 2nd best or next group of best applicants, the worst applicant might be just as far on the other side of the spectrum or worse. This could be really disastrous.

For example, the most likely real world application of this problem is in dating. Sure you dont know exactly what N will be, but you date a certain number of girls and then eventually you choose one to settle down with. If you wait a while and refuse to settle for someone worse than your first set of applicants, you may end up getting stuck with a real shitty girl at an older age.

I thought I could immediately improve the solution to this problem. My solution was fairly simple: iterate this same process. Go through first n/e applicants, write down max, reject those. now there are n - n/e applicants. look through next (n-n/e)/e applicants, if you find a better one in there you choose it, if not you write down the max from that group, etc. iterated on to the end where you just go 1 at a time.

I programmed a simple version of both in python. I ranked secretaries 0-99, randomized their order and used each method to test the % choosing best applicant, the median applicant chosen and the average applicant chosen.

#Original secretary problem solution based on 100 secretaries and just on the ranking of secretaries from 1-100
#want to figure out both average ranking chosen and how often the best candidate is chosen

#function which sorts a list
def sortedRankings(sec):

x=0
while x <99:
import random
from random import randint
a = random.randint(x,99)

temp = sec[x]
sec[x] = sec[a]
sec[a] = temp

x = x+1

return sec

#function which returns a secretary based on original algorithm
def chooseSecretary(sec):
max=sec[0]
x=1
while x<36:
if (sec[x] > max):
max = sec[x]
x=x+1

while x<99:
if (sec[x] > max):
return sec[x]
x=x+1

#sort winners
a=0
while a<1000:
b=a
min=100
minLocation=-1
while b<1000:
if(min > winners[b]):
min = winners[b]
minLocation=b
b=b+1
temp=winners[a]
winners[a]=winners[minLocation]
winners[minLocation]=temp
a=a+1

print winners
print winners[499]

results: Ran the simulation 1000 times. The best secretary was chosen 375 times. The median secretary chosen was the 2nd best. The average choice was the 18th best. And roughly 3.5% of the time a bottom 10 secretary was chosen.

#function which sorts a list
def sortedRankings(sec):

x=0
while x <99:
import random
from random import randint
a = random.randint(x,99)

temp = sec[x]
sec[x] = sec[a]
sec[a] = temp

x = x+1

return sec

#function which returns a secretary based on original algorithm
def chooseSecretary(sec):
max=sec[0]
max2=0
max3=0
max4=0
max5=0
max6=0
max7=0
max8=0
max9=0
max10=0

x=1
while x<36:
if (sec[x] > max):
max = sec[x]
x=x+1

while x<60:
if (sec[x] > max):
return sec[x]
if (sec[x] > max2):
max2 = sec[x]
x=x+1

while x<74:
if (sec[x] > max2):
return sec[x]
if (sec[x] > max3):
max3 = sec[x]
x=x+1

while x<84:
if (sec[x] > max3):
return sec[x]
if (sec[x] > max4):
max4 = sec[x]
x=x+1

while x<90:
if (sec[x] > max4):
return sec[x]
if (sec[x] > max5):
max5 = sec[x]
x=x+1

while x<93:
if (sec[x] > max5):
return sec[x]
if (sec[x] > max6):
max6 = sec[x]
x=x+1

while x<95:
if (sec[x] > max6):
return sec[x]
if (sec[x] > max7):
max7 = sec[x]
x=x+1

while x<97:
if (sec[x] > max7):
return sec[x]
if (sec[x] > max8):
max8 = sec[x]
x=x+1

while x<98:
if (sec[x] > max8):
return sec[x]
if (sec[x] > max9):
max9 = sec[x]
x=x+1

while x<99:
if (sec[x] > max9):
return sec[x]
if (sec[x] > max10):
max10 = sec[x]
x=x+1

#sort winners
a=0
while a<1000:
b=a
min=100
minLocation=-1
while b<1000:
if(min > winners[b]):
min = winners[b]
minLocation=b
b=b+1
temp=winners[a]
winners[a]=winners[minLocation]
winners[minLocation]=temp
a=a+1

print winners
print winners[499]

Results: ran 1000 times. best chosen 263. 3rd best was median choice. average: 4th best. worst chosen: 50th best.

this seems much much better. of course i didnt program it for random values in a normal distribution, nor did i program it for the more likely real world scenario of an unknown N. but this clearly seems like a much better solution despite the fact that the original solution chooses the best secretary ~10% more often.

So, if you guys would like, run with this in the more general forms or try and come up with an even better solution.

I took my first BJJ lessons in the summer of 2008 with nazgul, mig, ket, tjbentham and a few other poker players. We only did 2 or 3 sessions but I really enjoyed it and decided to keep up with the sport after I left vegas. From that summer til probably about 6 months ago I continued training but at a very irregular pace. I would go 10x one month, suffer a small injury, take a few weeks off, then start traveling and somehow miss 8 months. Since then I have lived in new jersey, arizona, california, vancouver canada and spent a lot of time in vegas so it was hard to really get into a rhythm training wise. I would estimate over the 5 year period from september 2008 - september 2013 I probably only took about 200 classes. And since I took so many breaks and had so many months where id show up only 2-3x, I wasnt really getting any better. But the good thing was that I never really regressed.

A few months back I realized that where i live (southern california, specifically irvine in orange county) is basically the mecca for BJJ in the world. Most of the top guys are here and there are tons of great schools everywhere. I dont know how long I will be in california and I am not on the road as much as I once was so why not concentrate and try to put a lot of time and effort into the sport as I can while im in the best place in the world to study it? Especially since I have all the free time Id ever want. So I decided that any week where I was healthy id go in as much as I can.

this past week i finally got to put away my white belt for good.

Blue belt is probably the belt with the largest skill disparities between top to bottom (ignoring complete novice white belts). Some schools promote guys to blue belt super fast. For example, when I was up in vancouver I took a few classes at gracie barra vancouver, and they like to start guys out with quick promotions, so most people get their blue belt after 80 classes in order to keep newer practitioners excited about the sport. Consequently, when I was a 3 stripe white belt I was about equal to most of the 1 stripe blue belts there. Whereas some other schools refuse to promote and try to get their guys to world champion levels before giving them the next belt (mega sandbaggers)

Additionally, blue belt is by far the longest belt. Nobody gets a purple in less than 2 years and most guys need more than 3. So even though I am starting out a higher skill level than most new blue belts, there are 4 stripe blue belts who have 2-3 extra years experience on me and are much better. So now that I am starting to take it more seriously, there is still a large roadblock to competing at the blue belt level.

this week was the Pan American Championships, one of the bigger jiu jitsu gi tournaments in the world, and it was basically down the street from me. I had 3 teammates (one blue, 2 purple) and my professor (black) competing this year so I went out for the last 3 days and had a blast. My professor finished 3rd in the lightweight (168 and under) 29-34 year old black belt division which was pretty awesome to watch and the 2 purple belts did pretty well.

Even watching a teammate lose in the first round was a blast. So next year I am going to compete. The pan ams is the perfect tournament to do it. I am 29 right now (turning 30 next month) so ill be in a division with only 29-34 year olds, which means I likely wont be going up against some beast future world champion or ex college wrestler that just switched to BJJ and ill have a full year of blue belt to catch up in skill to the more advanced blues (and the top guys will be purple by then). Also its a big tournament so even though its a pretty small subdivision of the sport (29-34 year old, 169-181 pound blue belts) it still got 29 competitors this year.

Ill probably do a few smaller tournaments beforehand just to get the nerves of competing out of the way and to practice really going at full speed against someone of a similar skill level, but this will be the first tournament I really care about. 12 months to prepare, starting off the training right:

some pictures and videos of the guys from my academy:
round 1 match of my teammate jimmy (bearded guy), he ended up making it to the 3rd round/quarterfinals of the 29-34 168 pound purple division: https://www.facebook.com/photo.php?v=10106134351896514

round 2 match of my teammate chris, tough match against a guy whose strategy throughout was to get a sweep then basically stall out and it got him to at least a top 3 finish: https://www.facebook.com/photo.php?v=10106136535775004

a few pictures:
my professor is 2nd from the right:

-competitive league with poker friends: lots of sharks in here and 2-3 complete fish. overall tough league, finished 1st/12. buyin 400, cashed for 2k
-7 yahoo pro leagues: these are 10 man public leagues with buyins of either 20, 100, or 250. i ran 1x20, 2x100, 4x250. Won the 20, won a 100, and finished 3rd in 2x250. buyins 1220, cashed for 1100
-league with friends from home: they are pretty bad, 14 man league, won total points for season and finished 3rd on year. buyin 50, cashed for 100
-league with college friends: finished 4th, buyin 100
-league with friends from home: finished 4th, buyin 150+$2 transaction fee * 32 transactions = 214, cashed for 95
-auction league, co owned with roothlus. didnt make playoffs, buyin i believe was 500
-national fantasy football championship league, buyin 350, won individual league for 1500, finished 26/1800 in overall standings for additional payout of 500.

total buyins: 2834
total cashes: 5295

Feel like year went pretty well and i could feel my edge in every league. also learned a lot and will be a bit stronger next year.

Daily Fantasy
As the year progressed, I started getting more and more into daily fantasy sports. At the beginning of the year there was a nice guaranteed tournament over on draftkings (www.draftkings.com, has some poker crossover, their VIP manager is Jon Aguiar aka fatalerror and they have a partnership with the WPT) and a deposit bonus, so i threw 1200 on, fired 6 lineups in the $200, think i had one small cash, fired most of it away and had like 38 bucks left in the account. Let it sit there for most of the season, but a bunch of my friends had been getting into daily stuff so I decided to start firing.

didnt feel like redepositing but i wanted to make as many lineups as i could each week to try and learn as much as possible for next season. so first week i made 19 lineups and entered them in the $2 tournament, got back $49. the following week i fired in 1 $11 lineup and 19 $2 lineups, did well, got back $113. then fired in 2 $11 lineups and 45 $2 lineups, did very well in a few with 4 top 30 finishes out of 8500 and cashed for $240. first place was 3500 and i was pretty close to that so it was a little frustrating to have so many entries do so well and cash for so little but a good sign that im getting a much better grasp of things. this past week there was a bigger $150 guaranteed tournament, and since it was week 16 i fired in a lineup and put in around 40 smaller lineups, did bad in the large one and only got back $40 which is ok cause week 17 is so unpredictable with resting starters to prevent injury. most of the games do matter this weekend but a lot of teams that are eliminated just throw in the towel and some of the later game players know their fate by then and the games become meaningless.

but now having entered a few hundred lineups the past 4 weeks i feel good about my daily fantasy game and think ill put in a lot of work over the next 8 months and plan on going hard in it next year cause it feels like there is more money to be made here than in year long stuff, albeit with smaller edges and more variance. will probably deposit 10-20k next year and work out a year long plan for grinding it out. hopefully it goes well! if you dont play daily, i suggest you look into it, its really going to explode in the next year and its big for NBA/MLB/NFL.

The past few years I have gotten increasingly into fantasy football. This year has been by far the biggest jump for me in terms of committment, money wagered and time spent studying and researching. This is probably the last year of my preparation before I really delve into lots of big buyin leagues.

This offseason I spent probably around 100-150 hours studying, researching and running simulations in order to improve everything I could about my fantasy knowledge. I've done 10 of my 13 drafts so far this season and may add a couple last minute ones to do 15 leagues this year. They varied quite a bit.

some of the different types:
-10 team 1 qb/2rb/3wr/te/DST/K no PPR
-12 team 1 qb/2 rb/2wr/1 flex/te/DST .25 PPR
-12 team 1 qb/2 rb/2wr/2 flex/te/DST/K .25 PPR
-12 team auction keeper, $155 bankroll, 1 qb/2rb/2wr/1 flex/TE/DST/K .5 PPR
-12 team 1 qb/ 1 OP (offensive player, can include 2nd qb)/2 rb/3wr/flex/TE/DST/K 1 PPR
-14 team 1 qb/2 rb/2wr/flex/TE/DST/K/LB/DL/DB .5 ppr + other random scoring changes
-12 team national fantasy football championship: http://nffc.stats.com/football/home/nffc/index.asp

ive done a ton of work on all these drafts and everything worked out very well. Most of my work was on optimizing draft order by position round by round for each draft slot. I feel I have a much better grasp on this than almost anyone else and also a better idea of VBD (value based drafting) and dynamic value based drafting as well as player tiers and how to maximize value in them.

Since the season starts thursday and there are probably some last minute drafts going on and since everyone loves fantasy football thought id spread some wealth on what ive learned. ask me anything, but be specific with questions. i.e. dont ask what sleepers should i aim for, tell me the scoring system of your league, how many teams, what draft position you have, who you aim for, what goals you should have with each position etc.

Lloyd Irvin, a famous scumbag (those stories are for another day) grappling instructor put on an invitation brown belt no time limit submission only tournament he has dubbed the BJJ kumite. His student, Keenan Cornelius, is the next big name in BJJ. He won all of the biggest tournaments (worlds, world no gi, pan ams, european championships, brazilian national championships) at middleweight and absolute as a purple belt last year and was recently promoted to brown. Lloyd wanted to prove that Keenan was the best brown belt in the world, so he invited some of the other best guys in the world to compete against him.

Check out the 3 episodes:

Really enjoyed watching these. Lit a fire under my ass to start training a bit harder.

My mom's kindle was tilting the shit out of her. She would order stuff from amazon and it wouldnt show up in her library, its a simple device with no buttons and hard to get used to and she preferred reading on her ipad so she gave it to me. Since then i've read 7 books and can barely put the thing down.

books ive read:
hunger games trilogy - was an enjoyable and quick read.

Speaker for the Dead: http://www.amazon.com/Speaker-Dead-Ender-Book-2/dp/0812550757
The sequel to Ender's Game. Well kind of the sequel. The author later decided to add lots of prequels and interquels so there are a few books that take place in between, but this is the next book he wrote in the series and I have decided to skip over most of the others. Loved it even more than Ender's game.

Xenocide: http://www.amazon.com/Xenocide-Ender-Book-Orson-Scott/dp/0812509250
the third book in the Ender's game series. its somewhat different than the first two, its turned very philosophical and talky but I love it. I have never been as infuriated with characters in a work of fiction as I am with this book and I think thats a good sign. I imagine most people who read Ender's Game either stopped after the first or second book and never reached the third. Def worth continuing forward.

My february reading list is the last few books in the Ender's game series. Children of the Mind and whatever comes after that in the main ender series. Then I will probably move on to some nonfiction works.

you guys recommend anything else that falls into line with the books ive been reading? or any recommendations whatsoever,im on pace to read about 85 books this year so probably have time for just about everything