Calendar

November 2008
M T W T F S S
« Oct   Dec »
 12
3456789
10111213141516
17181920212223
24252627282930

Archives

Categories

Derivitives of luck

What type of graph is this?  I have no idea.

I asked Excel for a line graph, but it looks like it doesn’t want to play ball, even though the x-axis data definitely is taken from continuous dates and times, with values to the nearest second.

I guess it’ll do.  It’s not like I’m going to ask the bloody paperclip what went wrong.

This is the output from my messing about with my PokerTracker database in order to try to quantify how lucky or unlucky I’ve been lately.

Since moving up a limit in the turbo sit-and-gos my results have streaked all over the place.  I started off with a $500 upswing which came from results that were almost good enough to put me on the Battle of the Planets leaderboard – but not quite.

This excellent run was immediately followed by a $1000 downswing.  It all happened very quickly, without me (knowingly) changing the way I played or the way I picked games.

There’s actually very little game selection you can do when a sit-and-go fills up in less than a minute and seating is randomised, but as a rule of thumb if I see a dozen tables in the lobby all with the same 4 or 5 players sitting at them the chances are those players all know what they’re doing so I’ll hang back a few minutes and wait for their games to start before I try some new tables.

I thought I was used to the ups and downs of playing turbo tournaments, but these swings seemed unusually big - and not just because of the higher stakes involved.  That $1000 loss on the $27s is equivalent to about $600 on the $16s, and the worst spell I’ve ever had has been a $400 loss before starting to claw it back.

In fact, it’s only because the results were so drastic that I suspected I’d been unusually unfortunate and wasn’t just getting crushed by other players in this game.

I’m not going to get complacent.  When you seem to be constantly finishing just outside the money, or only make 3rd place the times you do get there, or clock up more last place results than you’ve had in a long time there’s a good chance there’s some other reason for the bad results than just luck.

However 14 straight sit-and-gos without finishing in the money is a clear personal worst.  An actual monkey should get better results than that, as long as he thumps on the mouse often enough not to get timed out.

My dilemma was that I wanted to stick with it and ride out the bad streak but unless I could somehow show myself that I was statistically running bad (and therefore stood a chance of holding my own with "average luck") I was going to have to move back down.

I generated the graph above by pulling out of PokerTracker all the hands where I was all-in pre-flop, either raising first and getting called or calling someone else’s push.  In turbo tournaments the blinds get big quickly, so you have to take flips like this a lot.

The query I wrote only looked at hands with two players at showdown (about 6% of my all-ins were 3-way or worse) and ignored split pots.  There seems to be a quirk in the PokerTracker database when one player is forced all in on a blind so that it doesn’t know whether he "won" or not (it treats the small stack surviving and the big stack losing less than a big blind as a split pot) so I excluded those hands too (another 5% of my hands).

From this I plugged each hand in turn into poker-eval, grabbed my equity value and used this to plot a cumulative graph of my performance ahead of or behind expectation.

This is a similar concept to Sklansky bucks, but I treated every all-in situation as having the same unit value rather than counting how much money was at stake each time.  Converting the amount of tournament chips won or lost on each hand to an amount of real money would just be way too complex – all I cared about was whether I was actually having a bad run of luck, or spewing off my chips to superior players.

The massive slump at the end of the graph is somewhat reassuring, although to be honest I’d rather have my grand back.

The ups and downs in November certainly correspond to my leaderboard-challenging run followed by the doomswitch-fuelled plunge.  I moved up on November 5th, which is the 6th vertical line from the right – the one that reaches the highest point on the graph.  The values on the y-axis aren’t much use, but the distance of the points from zero and from the previous highs and lows gives you some idea of the extent of each hot and cold streak.

In fact, I would theorise that if you wanted to assess your "luck factor" at any given point in time, it is the gradient of the line on the graph at that point which matters, not the value at that point.  If you’re running good, the line is going upwards – regardless of where it started.

So, calculus fans, we need to find d luck by d time.

But I asked Claire how to differentiate my graph and she just laughed at me.  Apparently I’d need a formula to do that.

I’m not actually sure whether attempting to capture an abstract concept like luck on a two-dimensional chart is something I should be focusing my efforts on, or just a brief mad scientist moment.  Surely if I could find a formula for luck, it would cease to be luck by definition and the world would implode.  Muahahaha?

Nevertheless, I did manage to tidy the graph up a bit by removing the date values from the x-axis and plotting results hand by hand instead.

It still looks like it was drawn on blotting paper, but at least it shows what I needed to see in order for me to carry on possibly throwing money away playing poker at this level.

Good luck me?

Multitasking man

I’d been meaning to fish out this clip for a while but forgot all about it until I started to hear the results from the WSOP final table.

Craig Marquis is awesome.  He can like do twelve things at once cos he like grew up with computers man.  If you’re struggling to multi-table at online poker, the reason is probably that you’re not young enough.

Steve Friess from The Strip podcast was not impressed. (3 minute clip, direct link)

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Craig finished ninth.  Say it ain’t so.

Immortalised

A few years ago Claire had a birthday cake made for me that immortalised my likeness in marzipan.

Today, a similar honour has been paid to me by Silverton Jim who has immortalised me as an actual walking, talking donut.

Read episode 7 of Stick to Vegas to see what this is all about.  Jim assures me he was not on crack when he wrote it, but I’m not completley convinced…

Going for going for gold

I’m such a sucker for a tiered loyalty program.

The number of turbo sit-and-gos I play on Poker Stars each month tends to mean that I can just about retain SilverStar, and if I’m running a little short of points as the end of the month approaches I’ll make an effort to play a bit more in order to keep that precious status.

I don’t know why I bother really.   I worked out the actual value of doing this is about $12 per month.

<show working>

The benefit of SilverStar (1500 base points per month) over BronzeStar (no qualification required) is that your FPPs (the points you can spend, not the VIP points that determine your status) accumulate 50% faster at the higher level.  Once you have that level you keep it right through to the end of the following month.

So the difference between earning your first 1500 points in month as a BronzeStar player vs SilverStar is 750 FPPs.  One FPP is worth about 1.6 cents, so those extra points are worth about $12.

</show working>

Like I said, I’m a sucker for it.  But you know that without people like me these schemes just wouldn’t exist.

I do wish I hadn’t bothered going out of my way to keep my status at the end of last month though.

PokerStars are currently running an promotion where you get a bonus if you increase your VIP level this month.

If you upgrade from bronze to silver, you get $50, so if I’d actually dropped back to BronzeStar last month this would be an easy $50 for not really any more play than I’d usually put in.  Much better than the $12 worth of player points from maintaining that level.

But now I need to bypass silver (1,500 points) and get to gold (4,000 points) for a $100 bonus.

It’s not even real money, it’s a bonus in your poker account that you then have to earn 7x the dollar amount in points to unlock.  But the thing is, I just can’t say no.

$100 for free, and all I have to do is play about two and a half times as much as normal, and then make sure I play again next month too?  Sure, sign me up.

Well, I started off thinking that’s how it was going to be, probably looking at 3-4 hours a day to stay on top of it, but Claire convinced me I should take the opportunity to move up levels (which will earn points twice as fast, so I don’t have to play for as long).

I was some way off the latest win goals I’d set myself before taking another shot at the next level, but in the absense of having the stats to back up a decision it does help to have someone to blame if it all goes wrong. :) So I’m going to go for it.

I never wrote about what happened last time I tried this though.

Everyone loves a graph, and this one has even been annotated to add trend lines as I saw fit.  You can even click on it for a high def version.  You lucky, lucky people.

This is my entire PokerStars single-table SNG history for 2008 so far (apart from steps, which can’t be measured the same way).

I’ve marked four zones on the graph.  Starting from the left, the first section is all $16 tournaments, and I was winning.  Then I moved up to the $27s and crashed and burned.  Next, I dropped back down to the $16s but played 6 at a time instead of the 4 I was used to, where I carried on losing - albeit slightly slower.  Finally I dropped back down to 4-tabling and things seem to have settled back down to how they used to be.

I’d only given myself $500 to play with at the higher level, which lasted just over 200 tournaments.  That isn’t really enough to know for sure whether I was getting killed by the game or I was just running bad.  I’m still optimistic that it was the latter, but the downward trend seems pretty consistent after the initial spike.

The thing that really stands out when I look at those results is the number of times I cashed for the least possible money.

Overall I finished in the money almost as often as I did with the $16s (38% vs 40%) but on the $27s I had more 3rd place finishes than the total of my 1sts and 2nds put together!  With 1st paying two and a half times as much as 3rd, that’s going to make quite a difference to my overall return, and I’ll need to watch closely this time around to try to see if there’s a reason I’m a habitual bronze medalist.

What surprised me most though was the difference between my results on the $16s when 6-tabling versus 4-tabling.

I really didn’t think I’d be giving up too much by playing two more tables – hoping, eventually, to be able to move up to 8 or more tables at a time and increase my volume before increasing my stake.  I accepted I might not win as quickly if my concentration was being spread thinner, but I certainly didn’t expect to suddenly be losing money at the same rate I used to be winning it.

The reason I dropped right back to four $16s at a time was to see if I’d still got what it takes to beat that game.  It’s not a huge sample size (which is why I was hesitant to move up just yet) but things seemed to change almost instantly.  The graph appears to be going in the right direction and the last magenta line is virtually the same gradient as the first one.

So I don’t think I’m broken, but apparently four tables is as many as I can handle.

Anyway, the heat is on and the time is right for me to play my game.  Going for gold, four-tabling the $27s.  Watch this space.

Paradise spammed

Claire got this email from Paradise Poker trying to tempt her back to play, which almost certainly won’t work unless they bin the horrible Boss Media network and reinstate their kitchy old software with food and drink at the table and flaming cards when you hit a high hand.

Personal greeting – fail.  The broken images are a nice touch too.  Good effort.

Still, at least they want her back I guess.  I haven’t had a sausage.

Free kicks

It’s been a while since Sporting Index last sent out a bet refund promotion.  Thankfully the dry spell ended this weekend.

Their new spread game is, as usual, some kind of random number generator apparently based around a computer sports simulation, but with no possible way it could actually be representative of what you see on screen.

This time it’s penalty kicks.  The goal area is divided into 35 areas and points are awarded for every goal scored depending on where it hits the net.

That’s if you’re betting on a "points" market.  You could just bet on the total number of goals, or the differential of goals (or points) between the two teams.  Or you could plump for something much more exotic.

For example, take the shirt number of the striker and multiply it by the number of points for the goal he scores (if he scores) and add them all together.  If that number is higher or lower than some other number, you might win money.

The question is: how do you make your decision as to which way or how much to bet when you have no way of knowing how the trajectory of the football is determined?

Each square on the goal grid might be equally likely to be hit, but there’s no reason to assume this is the case.  And even if it was, what is the chance that any particular shot is saved by the goalkeeper?  You’d have to factor that in too to see what the actual distribution of scores is likely to be.

This is all assuming that the RNG is fair.  I’m sure it is though.  Online gambling is never rigged.

The workings of this game are further obfuscated by the fact you can choose two teams from a list of five to play the shootout.  Brazil is most likely to score from the penalty spot and Mexico the least, they say – and who am I to argue?

As you select your teams, you can see the markets move based on how much better one team is than the other.  In fact the values in the markets are the only clue you have as to what’s about to happen.

You can also pick the goalkeeper, although this appears to make no difference to the numbers at all.  It’s just there for comedy value: choose from Hans Blocker, Dmitri Tipitova or Claude Le Ballawei.  I groaned out loud (GOL?), just like you’re probably doing now.

The offer I received was for a refund of net losses up to £50 on this game, providing I placed at least ten bets each with a £5 minimum risk.  Clearly, there’s no way you can lose if you do this right.

I chose the total points market to meet the requirements for the refund, purely because of the numbers involved.  For this combination of teams, the price was set at 51 points to buy - meaning that a 10p stake would always have a maximum loss of £5.10.  That’s very much a worst case scenario too, as you can only lose the full amount if there are no goals at all scored in ten attempts!

In fact the variance on this bet was pretty low.  After 9 spins I’d had a loss of £2 and a win of £2.60 and seven other results somewhere in between.  Overall I was down £4.40.

The way I usually like to play these things is to take the full amount of money that I can still bet risk-free after meeting the requirements (in this case £45.60) and lump it all on one bet at even money or fairly short odds.

The only bet in this game close to offering fixed odds is the win index, which awards 25 points if the chosen team wins the shootout, 10 points for a draw or 0 if they lose.  I’ve actually played this same bet on a real football match before.

I simply decided to pick the market which would allow me to buy one team at as close as possible to 10 points, effectively resulting in a fixed odds bet at 3-2 (25-10).  I wanted to make sure a draw resulted in a do-over, rather than a win or loss that would affect the amount I had left to play with significantly.

The bet I chose was Mexico vs USA, backing Mexico @ 9.4 points for £4.85 per point.  Worst case, I lose £45.59 (just 1p short of getting the full £50 refund, when you add in what I lost so far).  The draw is almost irrelevant, but the win is worth a healthy £75.66 for £71.26 profit in total on the promotion.

Mexico surged ahead early, only to choke later on in a frustrating 3-3 draw, which ultimately landed me just £2.91 for the effort.  After I stopped shouting at the stupid little pictures of pretend footballers, I increased the stakes ever so slightly and ran it again.

Unfortunately the USA won 5-4 that time.

Was worth a shot though.  Ship that refund…