February 17, 2016

Starting Out On My Next New Downswing :)

The title is a bit misleading and sounds a bit sour, hey, isn't that quite like the usual BaseMetal blog post? Well yes but this one is more of a celebration.

About 3 or 4 years ago I started out trying to get reasonably competent in 180 seat MttSng's and things went fine for the first 4 or 5 thousand games. I have archived off the data for these but I think I ran at perhaps 20% roi or so even though I lost plenty of allinEv. Not a great result but not too bad as I was still learning a good style, I still didn't play that many 180's and mostly concentrated on the smaller field games.

At some point the poker gods decided to show me how variance worked in these 180s – and it was a hard lesson.

I quite quickly dropped into a 600BI downswing, was this sloppy play or just bad luck? Who knows? I did start to pull back and recovered to about about 300BI's or so but then a second variance wave hit me and pushed me back down to 600 below again.

I was a bit peeved with 180's and pretty much stopped playing them. I think at this point overall, even after the good start, I wasn't very much above break even in them. Over the last year I played a few more here and there and I did put in a bit of study at them as I thought it would be good to get a proper feel for the push fold game at pre ICM points as this knowledge would probably help in some other game analysis.

Late last year I decided I would have another proper go and started to put in a little volume. At first I ran a touch badly but then hit a heater and what do you know I have finally managed to dig myself out of this gaping 600BI hole. Yippee!

For the last couple of months at these 180's I have actually ran pretty good and have even managed to cash in on this good fortune, over the 1.5k games I've hit a 50% roi!

So I've now played about a total of 15k games and I think I have seen about a 5% roi in them but I still don't really know what sort of roi I could expect from them in the future.

At some point I was down about 4.5 million chips to allinEv but about half of this was from late in final tables so it is difficult to assess how much this FT stuff is worth. I think the ~2 million chips of pre FT is worth quite a lot though and I suspect about 800 to 1000 BI's so it is still possible that I may be one of the 25% plus bruisers. Well I can hope as not only did the allinEv stuff hit hard but I think I also ran extremely badly overall for the other hands. It is easy to see why you need more like 100k games to really get a decent estimate from just looking at the results in 180's.

So I now feel ready to take on my next sparkling new, shiny, downswing, I doubt if the next will be as big but I survived it and coming out the other side I do feel it has strengthened me.
(I should add that this 600 swing is only when looking at my 180's, the other smaller field games I played did help to protect me – it would have felt so much worse if I had only been playing these)



Posted By BaseMetal at 03:10 PM

59 Comments

July 21, 2015

Powered by BaseMetal

A few weeks ago I finally got around to calculating the probability of allinAdjusted to NetWon and now I have I spent, or wasted you may say, a little time to automatically graph these results similar to the way Hem does. When I did I noticed an interesting difference between mine and Hem.
Here is the Hem one for the last 4 weeks of 180 SnG hands, filtered by at having least 5 players.

User Uploaded Image

 Here is my equivalent also filtered for at least 5 players.

User Uploaded Image

 As you can see mine is spikier and more saw-toothed. This is because I process the hands in play order for each tournament rather than just order by the time of the hands. However in Hem as I play several tournies concurrently each hand is brought into the graph just by the hand time so several of these concurrent hands overlap from different tournies playing at different stages and blind levels.

If you look closely I cheekily put “Powered by BaseMetal” on my graph but really most of the heavy work is done by a variety of open source libraries, c++ Boost Spirit for my H-H parser, then Swig this into Python and I also use numpy/scipy, pandas, transitions and a few more python libs. I use the Python(xy) distribution and I find this works pretty well and is quite well supported for scientific stuff.

I have also analysed quite a few spots and one thing I have noticed that surprised me is how well the bad Ax hands do in certain circumstances compared to hands like K9o. For example with a 5bb stack and antes shoved from utg 8 or 9 handed A2o does better than K9o. I used to avoid bad Ax hands as they do get dominated so often but actually they have a lot of equity versus some calls and also act as a blocker so that your shove gets through sometimes. In fairness 97s does even better than either though.

I have also decided now to calculate how to improve my luck at these allin's ;), this last month again is appalling, here is the output from these graphed hands:
sumNetWonChips: -22037
sumEvGain: 367920.930925
sumVariance: 23150391473.2
sumTotalSD: 152152.527002
Prob of this score or worse: 0.00518948605303 or 1 in 192.69730948

So as it's 1 in 192 it is about as bad as I should expect for any month in a 16 year period, ouch!
It's time I got a break on these allins, c'mon poker gods you know you can do it!

Incidentally, most players when filtering for allin's pre-river and 5 or more players in 180 SnG's should have a similar sized variance per hand. For me here this is ~27,000,000 per hand (edit: I haven't really calculated the best average estimate for 180s 5+ handed is yet, somewhere between 20m and 30m), and roughly 3100 qualifying hands per 1000 games . If you want to roughly calculate how often you are likely to score your difference then multiply this variance by the number of hands, yes it's a huge number, but then take the square root of this and this will be your standard deviation for this block of hands. It won't be super accurate but it could give you a very rough ballpark figure. I haven't really checked how much spread occurs for these filters but maybe in the future I'll see if a reasonable estimate can be made. Incidentally I think the  trackers could very easily do these calculations same as I did and give precise values - I think it is good to know how often bad or good things happen in order to plan for the future.

Posted By BaseMetal at 07:15 PM

26 Comments

July 08, 2015

A Long-expected Moan

(or, Oh No He's Back Again )

Lately I haven't been too productive as far as blog posts go and hey I suspect that many of you dear readers probably think that one Base post every 6 months is one too many.

So have I been up to anything interesting in the Poker field – no not really so there isn't much to say.

Well in the past couple of months I've been looking for a better way than HEM to analyse my hand histories and I have finally managed a decent program/script to do so.

One of the side pots of this work is that I can now also analyse the allin adjusted hands better than HEM does, I don't think HEM is a great product for tournament players, I think it only pays lip service to tournaments with poor filters / Notecaddy – then again I also think PT4 is nearly as bad. Oh dear I seem to be criticising my poker tools I guess I mustn't be a good tradesman.

Anyway in some earlier posts I did say I would try to find out how often my none adjusted net chips should drift off from the mean of the adjusted line. This is actually pretty easy to do as all you need to know is the variance, or standard deviation, of the hands and viewing enough hands will show a Normal bell-curve like relationship to the 'expected'. Now HEM does give the variance/standard deviation for cash hands in bb/100 but for some reason can't be bothered to calculate this for tournaments players. Maybe they think that tournament players can't even speel varience so it would only confuse them. Well now I've found a program with a spell checker and I can also calculate the variance exactly of each individual hand - so there HEM.

Using the program I developed I have found I should have been pushing wider in the later positions and perhaps even a touch tighter than Nash in the early positions. I have made a few adjustments to my play but it is still too early to tell if I will get any pay off from this. In fact at the moment it doesn't look good but it is easy to see why. It has been about 3 weeks since the change and I have fallen further into the deeps of a downswing – here is the latest graph of allin adjusted for these games (all 180's) (Note that this HEM graph is still a little off in that I can only filter for hands that are allin pre-river for the Hero and so I get a couple too many.)

User Uploaded Image


So as you can see it looks like I have taken a good few beatings from the lucky-stick.
Ah, well Base can you now tell me how often such a pounding happens in general? Yes I can and it is far too often.

Oh, you wanted numbers…note that these are the actual correct figures and not the HEM ones.

Net Chips Won: -159910
Expected Chips: 170817.044012
Total Variance: 25032021841.2
SD for these hands: 158215.112556
Prob of getting Net Chips Won or less: 0.0182925944037 or 1 in 54.6669312145

So if I play roughly the same amount of games for another 54.7 * 3 weeks, or about 3 years, then this last period is likely to be equal to the worse of all these future 3 week blocks. Phew, I can't wait to see what comes next.

I have archived off my histories before 2014 but after this I have played about 5k 180 games since then and here is the graph for these.

User Uploaded Image

 
Again not too good and how often is it likely to be this bad, well here are the results for these (Again note the HEM values are a little off).
Net Chips Won: 4051914
Expected Chips: 6272849.37582
Total Variance: 1.23047250501 e+12
TotalSD: 1109266.6519
Prob of getting 4051914 or less: 0.022633469597 or 1 in 44.1823555029

As you can see the variance when measured in chips squared gets to be a pretty big number but you take the square root to get the sd and this gives a more reasonable human figure. As these results will be Gaussian roughly 68% of the time the none adjusted should fall between +/- 1 sd of the mean, ie, the 'Expected Chips' value. Alas, for me I am about 2 sd's away from the mean and on the wrong side – but don't worry about me, at some time in the future I intend to get lucky!

In some ways this probability doesn't seem as bad as I thought it would be. For every group of 50 fishy BaseMetals out there swimming in the Pokerstars sea there is likely to be one that is even more allin unlucky than I am. Tbh, before I calculated this I did think it would have been more like a 1 in 100 or worse case so I should stop the complaining but then I what could I possibly blog about.

Posted By BaseMetal at 01:15 AM

19 Comments

November 27, 2014

A Small Update

Not too much to report and I've got little to complain about this time although I could have done better – I guess I'll always be complaining in this blog.

For the last few weeks the adjusted allin's have pretty much balanced the actual. (I have even actually done slightly better by a couple of hundred k chips.)


User Uploaded Image

  I still haven't played anywhere like enough games but feel I have improved a little in some areas although I am only hitting about 12% roi, I think I still haven't ran too well though, I'll just keep trying to make some improvements.


About 2 or 3 months ago I decided to try using a lot more raise/fold/call lines when 15 to 10bbs deep and I'm not really sure if there is any real improvement over the simpler just shove line. I think over the next month or two I will cut back a little on this and shove more unless there are a few aggro's present behind. If it does seem pretty neutral either way I might just vary the style randomly to make things a little less readable for other regs.

Posted By BaseMetal at 04:22 PM

20 Comments

October 15, 2014

Keep on Grinding

Lately the AllinEv has been pretty bad with only a couple of small patches where I have run ok. I think I have to get used to these and just simply weather the storm but it has been quite hard from the mental game standpoint.
In the previous blog I had just managed a small section with a +ve roi and thought I was through the worse alas the poker gods decided to test me with some more really bad AllinEv - I wish they would pick on some of my opponents for a change.

At the worst point it got to a 3.8 million chip difference and this seems really bad to me for a set of just 11.5k SnGs, I mean that is effectively 2.4k worth of buyins at their full price, phew!

Again it has now recovered a small amount and it's just 3.3 million off but this bad stretch of poor luck is effecting my mojo, I have been doing a bit of work on my game and I reckon the allin hands are the main reason I am only sustaining a paltry 2% roi overall, every section when I do get a balanced stretch of allin I manage to hold 10% or above.

The real problem I have is that I do play a lot less games at these bad times and this is a pretty big mental leak to my game, I do know this problem is there and so I do intend to overcome it eventually. I don't think I really splash tilt much at all although taking lots of bad beats probably does knock my confidence and drop my roi a point or two. It seems so much easier to play my 'A' game for longer during the reasonable patches. I'll try to concentrate and keep improving my underlying game and also my mental game. These bad sections are like rites of passage and any self respecting poker player has to get through them. I'm still climbing out of a 500+BI downswing but it looks like I will get my head above the water line soon.

Here is the latest graph of allinEv, it covers about 11k games mostly 180's and 45's but with a few 18's, I've marked the approx roi for the sections.

User Uploaded Image

Obviously the green and the orange should really go parallel and from the relative 'un-bounciness' look of the green shows the variance is low compared to the 3.2 mill chip difference so I do at least think it is an unusual stretch and shouldn't expect too many several million diffs often in the future.

I'll try to stay positive and as Buzz Lightyear would  say “To infinity … and beyond!”

Posted By BaseMetal at 02:04 PM

20 Comments

June 09, 2014

A Bit of a Gift?

.. never look a Frequent Player Point in the mouth.

I haven't played that much lately and this is probably somewhat of a tilt - I do tend to reduce volume when things don't go my way and I guess I should pull myself together and suck it all up. A good mental attitude would help and actually in this regard you will be pleased to know I have just bought The Mental Game Of Poker 2.

I decided to use up some of my frequent player points and not being supernova a straight transfer to cash isn't that good. Sometimes I play a few 6 seater 235 point satellites as these are quite good value for me. I can easily do better than 1 in 3 cashes and so I think this works out as about 235 * 2.9 = $11.00 so 1 point = 1.61 cents.
Even at Platinum star you only get a $200 reward bonus for 15400 Fpps, so 1 fpp = 1.29 cents.

So it's much better to cash these points in via tournament tickets or as I have now found, rather than going to all this bother of playing :), I can change them into £ gift vouchers and this also seems quite decent at the moment. Maybe it is because of some changes in the exchange rate but I got £250 for 25900 points and this comes out as about 25900 = (25000 * 1.68) = 1.62 cents per point.

Well, armed with the £250 gift voucher I bought a Nexus 7, some sundries, and also 3 poker books.
1) The Mental Game Of Poker 2
2) Applications of No Limit Poker
3) Killer Poker by the Numbers.

All at first sight seem reasonable although (2) is published by 2+2 publishing and I have found in the past that their books tend to fall to pieces after relatively little thumbing. I wish they would find better quality binders.

I am a little way through the Mental book and at the moment this looks quite worthwhile with plenty of decent ideas and suggestions - but this is provisional and I may come back in with a full review after I have fully digested it.

I may add some thoughts about the other books later although there is a good chance I might never get enough time to go through the Mathew Janda, Apps. of NL Poker, properly. This doesn't look like the type of book you just read through.

Posted By BaseMetal at 07:36 PM

13 Comments

May 21, 2014

Easy Game

These 180's are easy, I've clocked up 48% roi for the last 500 - I must be getting the knack.

Nope it's just the variance talking, to get a decent shot at an even approximate roi you need about 20,000 games. I have been transitioning to the 180s but so far in the last few months I've only yet played 5000 games. Maybe if I keep on solely playing these 180's by the end of the year I might get to 20k games and be able to nail my roi to as narrow as 10% +/-6 with a 90% confidence - jeez these MTTs are tough.

User Uploaded Image

In the last 500 or so I've run very slightly above expected on allinEv, yippee. As you can see the last part of the graph has ran about parallel as you would normally expect. A few other things have probably also gone right and hence the big roi score in these last few games.

I am now about +3% roi overall, I still think I am likely to be at least 10% when I get near the long run but that is still a good way off - I hope to keep getting better at them and, who knows, perhaps I might get as high as 20%. Some people do seem to own 25% at low buyins, maybe with a bit of study I can join these crushers.

I have done a slightly better attempt at trying to calculate how much roi the bad stretch of allinEv cost and it comes out at about 700BI's. I did this by checking on the allinEv diff for HU only and this suggests 1 chip is equivalent to 0.000061of a BI at this point, this is due to at the HU stage we can win an extra 16.5 BI's by getting all the 270,000 chips, so 1 chip = 16.5/270000

If we are 3 handed a chip is worth about 0.00011 as having all chips gets us 1st or an extra 31BI's from this point, 4th...etc.

If I use this quite basic method then I seem to have lost slightly over 700 BI's to the allinEV (ok it's not accurate but it's decent approximation). As most of this allinEV loss happened as I plummeted down into a 450BI downswing I feel it was almost exclusively this and not bad play that dug this hole.

In my early STT days I used to worry about 100BI swings but now I realize that with 180's you have to prepare for some 500+ ones. Each time you go through a bad swing you harden up, get callouses I think is the term JaimeStaples used. Each one is like a rite of passage as you pass through your poker.journey.


Posted By BaseMetal at 08:26 PM

11 Comments

May 09, 2014

If it wasn't for the bad luck...

...I'd have no luck at all.

Well you can probably guess where it's been going lately and it's not the direction I'd hoped for.

I still seem to be getting smashed by cold decks and playing 180's with poor allinAdjEv is so, so uncomfortable.
I started out reasonably ok in these 180s but for the past few months I've hit a terrible spell and it's now finally pushed me into -1.9% roi in them, for this db of roughly 5000 games. I still feel I am playing these games pretty fine and checking over a good few of these tournaments it does seem like somebody at Pokerstars has it in for me.
Here is the graph of allin before river for all the 180's I've played in this db, it's about 4600 total games now.

User Uploaded Image

As you can see I am now 2.3 million chips behind – I didn't really think I would see this large a gap open up and it does feel like it must be quite an unusual occurrence.
Is it unusual or just something 180 SnGMtt players have to get used to?
Does this kind of gap show up every 4 or 5k games? If so I think these 180s are too depressing for me to play.
I think at some point I may take some time out and analyze how frequently such a bad run should occur. I would like to know to figure out how comfortable playing these 180's is for the future. They do have the advantage of finishing reasonably quickly and seem like after learning a good style they are moderately easy to multi-table.
In the last 250 games I've roughly broken even and this doesn't seem too bad when so many hands went bad – approx. 500k chips below in expectation on this set.
Although I'm -2%  roi in these I can't really see me being a losing player and tbh I think of myself as at least +10%, only time and more volume will tell.
I am pleased to say that I haven't spiraled downward but have kept my style and play fairly constant but I do suffer from my own form of tilt and that is playing less when things start going wrong, I do find it very hard to keep the motivation to play when bad luck strikes. On the plus side it does cause me to reassess my play and do a bit more study and so if anything I still feel I am improving at these as my roi drops.
I started off this blog when I had A Bad Start To a New Stake and so this seems like a complete revolution of the wheel, it's a bad start to a new field size. The wheels are still in spin and I just need to pump a bit of air into these tyres to get up to a bit of crushing speed.
Goodbye for now – I'll re-vent or rejoice in a few weeks and let you know how it's all going.

Posted By BaseMetal at 10:59 AM

12 Comments

April 13, 2014

How Much is a Tournament Chip Worth?

How Much is a Tournament Chip Worth?

Everybody knows this – it depends....

First a warning this is just a bit fun, don't take anything in this too literally but I'm going to try to estimate the value the chips I have lost in the pre-river allins for the 45 and 180 tournaments I've been playing recently.
The most usual way to assign a chip value is by the icm algorithm but Hem2 (a program I still really dislike) does not calculate icm type values for MttSnG's. I can't blame it really as there is no way for the program to know the chip distribution across all the tables so I'll just have to try to get a rough approximation.
Ok, when you first sit down and you aren't a bit of a drooler the starting 1500 chips in a SnG should be worth about a buy-in, so 1500 = 100% of a buyin, this is easy.
If you get to the HU stage in a 180 the last part you both play for is 10% of the total prize pool, so at this point as all the chips are still in play it would seem reasonable to say that 1500 chips are now worth 10% of a buyin.
With 3 left you are fighting for about 40% of the pool so 1500 is about 40%, 4 left 50%... etc
When there are 10 or more players left the chips will be near their start value. Most of the time 1500 chips are worth near 100% of a buy-in except as the game progresses through to the final table.

So just to guess at how much the average chip is worth I would say, as I don't get near to HU often enough, that 1500 chips are usually worth 1/2 a buy-in at the very least. This seems a bit of a stingy valuation tbh but I'll go with it.

Well if you have read this far you've made a bit of a mistake because I have set you up for a moan. Here it comes....
I am running worse than the dog's dinner after the cat has done a doo-doo in it. At the moment I am in a 350 BI downswing – whoops. I'll have to put a bit of study in and see if I keep digging I can somehow come out the other bloody side.
I suspect quite a lot of this swing is caused by run bad and if it is ok to assign roughly an average of 0.5 BI's per 1500 chips the AllinEvAdj alone more than accounts for this disaster.

Here is exhibit A m'Lord

User Uploaded Image

So if you notice In the last 7000 allin hands (approx 2400 games of 45 and 180 SnGs) I have dropped 1,763,823 chips into the Bank of Sklansky or about 730 chips per tournament. Ah, but who's counting - oh it seems to be me and if I go for the fairly conservative 1/2 BI value that's enough for 588 BI's in the last 2400 games or about -24% roi, phew!

Sorry about the vent people but it eases the pain and as Lance Armstrong once said “Pain is temporary. Quitting lasts forever.”

Lance, toss me a bag or two of those drugs over please.



Posted By BaseMetal at 10:33 PM

17 Comments

January 24, 2014

HU and DON Sng Downswing Graphs (Hyper level rake)

These graphs are produced to give a guide to what sized downswings occur for hyper HU or DONs using a rake value of 2.78%, this rake seems roughly inline with that changed by Pokerstars up to the mid level buyins. Your roi is the most important figure in the size of downswings and using smaller level for rake would not have a great effect.

1% ROI Link to HU and DON Downswing Probability Graphs (Hyper Level rake)

2% ROI Link to HU and DON Downswing Probability Graphs (Hyper Level rake)

3% ROI Link to HU and DON Downswing Probability Graphs (Hyper Level rake)

4% ROI Link to HU and DON Downswing Probability Graphs (Hyper Level rake)

6% ROI Link to HU and DON Downswing Probability Graphs (Hyper Level rake)

8% ROI Link to HU and DON Downswing Probability Graphs (Hyper Level rake)

10% ROI Link to HU and DON Downswing Probability Graphs (Hyper Level rake)

HU and DON for standard rake level (~5%)

Posted By BaseMetal at 04:01 PM

18 Comments


About Me

Sputnik

BaseMetal

Archive