Online Poker

If you have no interest in things so geeky as DIY database queries and obscure poker statistics, look away now.  Next time I'll probably post a picture of one of my socks or something to bring a little balance back into the world.

So then, how many times have I flopped a set, and how much did I win?  From just over 50,000 hands in my database now, we'd expect to see about 3000 pocket pairs dealt (1 in every 17 hands).  The odds of flopping a set are are 7.5-1, so somewhere in the region of 350 flopped sets would be about right.  I had no idea what to expect as a win rate for these hands alone.

If you could care less about SQL, just skip past the code section, otherwise here's what I did.  I failed to find any more elegant way of checking for a flopped set than simply comparing the card rank of one of the hole cards with each card on the flop in turn.  This query will actually count flopped quads as well as flopped sets, if I have any.

select count(*) as freq,
       sum(gp.total_won) - sum(gp.total_bet) as won
from game_players gp, game g
where gp.player_id = 11
and gp.game_id = g.game_id
and gp.pair_hand = 1
and substring(gp.hole_card_1 from 1 for 1) in (
         substring(g.flop_1 from 1 for 1),
         substring(g.flop_2 from 1 for 1),
         substring(g.flop_3 from 1 for 1))

Let the results speak.  I've seen 317 flopped sets (close enough to what I was expecting) and overall made $2884.50 profit from those hands.  That's just about $9.10 - 9 big bets at 50NL - each time.  Look at it another way: just over every five flopped sets, I'm winning a full $50 stack from another player.  That sounds OK.

Look at it yet another way: compare to pocket aces.  Poker Tracker tells me that I've been dealt AA 253 times and that I won an average of 4.86 big bets each time.  Three hundred of anything is no huge sample size, but it's a start in the right direction.  We could say that flopped sets are, roughly, twice as profitable as pocket aces.  That's definitely nothing to complain about.

The difference, of course, is that pocket aces always start off as the best hand - and by a long way too.  What's been killing me lately is the abundance of set-over-set action, with me on the wrong end of it.  I'm sure I'm remembering things worse than they actually were, so what I want to find out is both how and when I've been getting beat.

Adding the condition

and gp.total_won < gp.total_bet

to the end of the query will find only those hands where I flopped a set and did not win.  This includes split-pot situations, as the rake taken from the pot will make the amount returned less than the amount I put in the pot.  I used this condition to find only losses (no splits), although it assumes there are two players at showdown.  Not strictly true, but close enough, and do it any more accurately would create a monster query.

and gp.total_won < gp.total_bet - (g.rake / 2)

The results: 32 losses, 8 splits.  87% of my flopped sets won at showdown (compare: I won with 89% of my pocket aces).  I can live with that - there's still two cards to come, after all.  The combined loss on those 32 hands was $984.15, or $30.75 per hand.  A big chunk for sure, but almost certainly I was ahead when my money went in.  If only there was an easy way to check that...

The best I've come up with is to find all my set-over-set confrontations.  To do this I have to find the times both I and my opponent have a pair that matches one of the flop cards.  The query looks a little something like this.

select me.hole_cards, opp.hole_cards,
       me.total_won - me.total_bet as won
from game_players me, game_players opp, game g
where me.player_id = 11
and me.game_id = g.game_id
and me.pair_hand = 1
and substring(me.hole_card_1 from 1 for 1) in (
         substring(g.flop_1 from 1 for 1),
         substring(g.flop_2 from 1 for 1),
         substring(g.flop_3 from 1 for 1))
and opp.player_id != me.player_id
and opp.game_id = g.game_id
and opp.pair_hand = 1
and substring(opp.hole_card_1 from 1 for 1) in (
         substring(g.flop_1 from 1 for 1),
         substring(g.flop_2 from 1 for 1),
         substring(g.flop_3 from 1 for 1))

It's ugly for sure, and still a little flawed too.  I can't even start to get my head around how this might cope with three players all flopping sets at the same time.  This query just dumps out a list of every time I was in a pot with a flopped set against another flopped set, rather than counting wins and losses - or times I flopped the bigger set or smaller set.  It's only a small sample size - 15 confrontations - but it proves a point.

Won 3, lost 12.  Overall, I'm down $386 from those hands.  A loss of $32.17 each time I got shafted is understandable, but I'm down an average of $25.70 - nearly 26 BB/hand - across all set over set scenarios, simply because I never seem to be on the right end of them.  I would expect that these are usually breakeven situations over the long term: the times you're ahead you expect to get paid off, and the times you're behind you're almost certainly going to get stacked.

Boy I am owed.  And if I talked to you about how poker's been going for me lately and spouted some far fetched story about how bad I'm running... just look.  Told you so!

The royal flush was an anomaly.

When you run bad you start doubting things.  You start to ask yourself things like whether flopping a set really is that good at all.  Because yesterday, I flopped four sets, and every one of them lost, and every one of them cost me my entire stack.

OK, as you insist, I'll indulge you with a bad beat story. 

I have pocket fives on a king-five-deuce flop, all different suits.  It all looks so gorgeous.  The pre-flop raiser fires on the flop, I call and the other player - who it turns out couldn't get away from pocket queens - also calls.  Magic.  The turn brings a second king and he leads out again, but for a much smaller bet this time.  Does he hate the king, or does he have AK and feel like he just got invincible?  Well, I can't put him on K5 or K2 and I don't expect he'd have played KK so strong on the flop.  Right now is where I'm going to make him pay, so I move all in.  Pocket queens feels like he's pot committed and comes along for the ride...

"I have AA", he types in the chat box.  I don't know why, because almost half a second later he's called anyway.  Considering I was hoping that he would have ace-king, I'm especially pleased.  Just four cards left in the deck - the two remaining aces and two kings - can improve his hand to beat me now, whereas AK would have had 3 aces, 1 king and 3 deuces for the win.

But when you run bad, you don't just get beat, you get beat in the cruelest possible ways.  A third king on the river gives us both kings full, his aces both playing but my hand being reduced to what may as well have been a five and a joker.

On days like this you start to wonder whether it's just a matter of karma.  If you somehow did wrong by someone, but it surely couldn't be that when I hadn't left the house all day, and the only phone call I answered was from the bank, telling me that a cheque I dropped through the letterbox at the weekend wasn't signed.  I now know I must try harder to check these things to make the nice lady's job so much easier.  She ended the call by asking, with a sigh, whether I wanted her to send it back to me.  Clearly, some of their customers are indifferent about money.

And so, finally, to the business of reassuring myself that flopped sets are actually good hands.  Of course I know this really, but it's a great excuse to conjure up some stats that Poker Tracker can't provide by itself.  So I've started to construct a few queries that can be run against a PostgreSQL Poker Tracker database to analyse set performance.

You lucky lucky people have a stat-packed post, full of SQL code and probably just proving something we alreayd know, to look forward to - hopefully tomorrow.

Royal flush, baby!

It's been a while.  Definitely over a year as I've not had a picture to post since I started the blog.  I so definitely would, as you can now see...!

Possibly the only thing more awesome than the royal itself is to have hit it backdoor style.

I hope you enjoy the picture as much as I did.  Click on it if you would like to join me in basking in its full size glory.

At 4.10am the pop up banner announced a ten minute break.  I never had a ten minute break on Poker Stars before.  Well, I guess that's something to be pleased about if I don't make it now.

Morning has broken, it's getting light outside.  I need to be up early in the morning to take Claire to school and her car to the garage.  I wouldn't mind being fit to drive, but that looks like a long shot.  373 players remain.  231 get paid, of which 220 walk away with an $11k package.  Which really is an $11,000 cash prize, as Stars cannot register players into the WSOP.

The chip leader is at my table, and he could easily fold to victory, yet he's still playing, calling big bets, and knocking players out.  I'd quite like to move tables.  I'd also quite like to see a big hand and have it hold up.  I'm below average now, but it feels like I'm one coinflip away from standing a damn good chance.

That'd be a $5500 coinflip then.  Where did my comfort zone go?

EDIT: Nearly 6am.  The four figure coinflop was 66 vs AJ.  I raised, and the flop looked good.  An ace on the river and it was all but over.  256th.  8 hours, no cigar.

Welcome to stupidly large World Series of Poker satellite night.

On Poker Stars, 150 seats are guaranteed to be given away and on Full Tilt another 100.  In fact, between the two sites there's over $4 million dollars in play - enough for 359 ten thousand dollar seats, and a bit of pocket change to the runners up.

I never even thought of playing one of these until today, yet here I am, battling with nearly 7000 others on Poker Stars in Probably The World's Largest Satellite Poker Tournament Ever.  OK, actually it was ever so slightly bigger last year, but not so much you'd notice.

Let's rewind a bit.

Turbo satellites are silly.  With the blinds at $1500/$3000, plus a $150 ante, I have little more than one small blind remaining.  Click on the thumbnail to see the full table image.  For Harrington fans, does the fabric of the universe falls apart when you have an M that has to be expressed as a fraction?  It all looks grim, and yet I'm loving it.  Believe it or not, I'm in great shape here.

This was an $80+$8 qualifier to the main satellite.  The lobby called it a "last chance" tournament, but I will argue that it was actually my first and only chance.  One in five got a seat, and it looked like a good way to use up my W$ balance, which has been doing pretty much nothing for as long as I can remember.  Sure I could have sold them for 80% of value, but I figured eventually there'd be a more interesting way to spend them.  This was going to be it.

With 350 players remaining and 317 getting paid, I'd been fortunate enough to get two successive table breaks that landed me in a good seat, just as I was about to be blinded out.  Game of skill my arse.  Cards were irrelevant by that point.  Almost nobody could survive one round of blinds, so all that mattered was hanging on longer than everybody else.  With 38 tables left, 33 players left to be eliminated and the luxury of five free hands before I was forced in on the big blind, I needed 6 or 7 players to go bust for every hand played at my table.  No problem.  I was all set to fold pocket aces.

Two hands later, you could probably hear my woohoos.  I'd got a result in the turbo poker lottery, which from start to finish took just 75 minutes to eliminate 80% of the field.

The main satellite will be somewhat slower, however.  30 minute levels and 6702 players to money.  Could be a late one.