Thursday, January 26, 2006

Grinding away.... and my SNG Pipeline

I've finally cleared my Party VIP and Cryptos bonuses for January. I am specifically not playing more at Party after clearing the current target because I don't want to be issued an even bigger target next month. The only other bonus I still particularly want to get cracking on is the Martinspoker one mentioned by Scurvy, which I have not yet started. I'm planning on going the €100NL route there, and with 3000 points to accumulate, I'm hoping that 60% of the pots at those tables are sufficiently raked (min pot size requirement appears to be €7.50) so that it won't take more than 5000 hands to clear.

I barely have 3k VPPs at Stars for January as this promo has by far been the least lucrative of the bunch (and additionally the flu bug knocked me out of commission during the whole first week of the month). I'm debating on whether or not it is worth the effort to complete the next 1k VPPs to reach Gold status (and Platinum is out of the question given the deadline). Hmmmm, assuming that I planned to accumulate at least 4k VPPs in February, then not reaching Gold level in January will "cost" 2000FPPs worth ~$32. My VPP accumulation rate is roughly 1000VPPs/3 hours for an approximate $10/hour in toppings. Ahhh, the petty thoughts of a low limit grinder.

Yes, it is clear that I was very much a low limit grinder this month, representing over 80% of my poker time. This is not how I intended to spend this month. At this stage, I let learning take a back seat to bonus whoring? Apparently the tail wags the dog in my world... (i.e. this month's promotions highly favoring playing full-table low limit ring games dictated my playing)

On the other hand, it looks like this month will be my 6th best month in earnings, but more significantly it was certainly the least stressful month of poker in at least a year. I don't think that the relative ease of the month should be overlooked. I feel that I played virtually tilt-free poker during this time, and that is something to be pleased about.

SNG's via a dual-pipelined 4-stage processor?

There seems to be some resemblance between the managing of SNG's and the managing of CPU instructions in a multipipeline, multistage microprocessor. (My educational background is in Computer Engineering) These are my current thoughts:


With players like Elky or Pokernerd tearing up the online world with their SNG's exploits, I've been considering how I would go about playing a lot of SNG's. I am certainly no expert at SNG's, and my experimentations have been down at the $20+2 Party SNG's. (Party SNG's play a fixed number of hands, 10, per blind level)

First off, for some perspective, what is the theoretical maximum earn rate of 8-tabling $20+2 SNG's? Experts say that a good ROI for multitabling $20+2 SNG's is 20% or $4.40 per SNG. Assuming it takes an average of 40 minutes to complete a single SNG, then (in steady state) 12 SNG's could be run through per hour for an hourly rate of $52.80.

It is a very simple to say that, for a winning SNG player, the more SNG's played, the more money you make. However, there are "scheduling" issues that make it difficult to achieve this simple ($ROI-per-SNG * #-tables / #-hours-per-SNG) rate while maintaining a high level of play in all of the SNG's played.

There is more table management activity done in multitabling SNG's than multitabling ring games. With ring games, so long as all of your current tables are good (and no substantially better tables are discovered), there is no other table management activity. SNG's, on the other hand, transition through early-mid-late phases with (on average) increasing amount of attention required as progressing through these phases. You need to get the maximum number of SNG's running without having too many of them simultaneously in the high-attention-required late phases. (Perhaps in higher stakes SNG's there is a more even distribution of attention-requirements between the various phases, but I will ignore this; in a $20 Party SNG there is absolutely no need to get involved with marginal hands in the first couple of levels)

The obvious thing to even out the collective amount of attention needed for all tables is to stagger the starting times of your SNG's so that you have them somewhat equally distributed amongst the various phases.

Initially, start 2 SNG's. When both SNG's have moved halfway through level 2, launch 2 more SNG's, and repeat this process until 8 SNG's are running. Everything is nice and evenly spaced out.

Consider that when 8-tabling you have 8*60=480 table minutes per hour (with each table minute being worth $0.11 given the $52.80 hourly rate). Let's assume that the first 1.5 SNG levels take 12 minutes. Given the above approach, you have already lost 144 table minutes (or $15.84) while "filling the pipeline"! The same problem is going to occur at the end of your session when you "drain the pipeline". There is no way around this, but it does suggest that you would tend to only choose SNG play when your available session time is substantial, say more than 2 hours.

Now, staggering the starts of the SNG's is all well and good if hands finished at the same rate at each table and you complete each SNG in the same amount of time, but this is not the reality. Some tables will run faster than others, and you will naturally complete any given SNG at a somewhat random time. This has 2 implications:
i) despite the best scheduling efforts you will still run into situations where you have perhaps 2-3 tables down to 2-3 players, 2-3 tables down to 4-5 players with large blinds, and remaining tables in the early levels. Even when completely ignoring the tables in the early levels (as you should), the quality of the decisions you make at the late stage level tables will suffer.
ii) you may have to insert some dead time before kicking off a new SNG to retain the staggering of start times, and this again is lost table minutes. You also need to spend the effort to periodically check if it is time to launch the new SNG(s).

Depending on the session duration, the lost minutes from filling and draining the pipeline may represent the majority of the lost table time, however, at least these are predictable. You can integrate other activities into this: reading/writing email, reading news/blogs/etc, doing any nearby things that can be worked on in 30 second spurts, and even temporarily jumping in some ring game(s) for which you have sufficient data mining info available.

Everything I have mentioned is very simple and obvious. The main point is just that table maintenance becomes a considerably larger factor when multitabling SNG's compared to ring games. The table maintenance is essentially another layer of decisions and actions in online poker playing.


For sessions of any significant duration, I am already close to maxed-out when 8-tabling ring games, so I may be incapable of effectively 8-tabling SNG's. (e.g. my win rate for 8-tabling may be the same or worse than 6-tabling)

So far, I have been running well in the $20s over a very small sample size, 74 SNG's. My ITM is 44.5%, and ROI 43.73%.


In terms of screen setup, I have 2 20" LCD monitors. My tables are organized on the 2 monitors like this:


To help me more quickly recall the state of the SNG's, the tables are organized such that the right most tables are always the SNGs at the highest level. As SNG's finish, the tables in each row will be moved from left to right.


However, I attempt to keep SNG's of approximately the same level in the same column, so sometimes there will be periods of time when there will be less than 8 active tables organized like:

8-6........ -2

or (more unfortunately from the point of view of lost table minutes)



F.J. Delgado said...

sup my friend... fellow pro just dropping a line, hope the SNG's treat you well.

i have you linked up over at my crib, check it out sometime whenever you get a chance.

that Wharton kid is one of my boys, he's a funny dude and a pretty sick player, in my not so humble opinion.


Payday Health Star said...

Check Out For The Best Real Estate Deals In Kerala