After 8k 30+3 sngs recorded, I've discovered something...

As the title suggests I have a crap load of 30+3 sngs recorded, and maybe another 2000-ish not recorded, and another 1000 50+5 and 100+9 sngs not recorded. I was looking over my stats and some batches of 1k sngs showed some really weird things.

I have 2 sets of 1k SNGs at 29% roi and 27% roi, and I have one set of 1k SNGs at negative 2% roi. My total (recorded) roi is at 17% roi for the 30+3s. What I rationalized is that it definately couldn't be only luck that could affect a set of 1k SNGs so dramatically, I mean there's a 31% roi difference on the end of each spectrum of my 1k batches.

All my sngs were done 8-14 tabling, 8 tabling maybe 5 months ago, started 12 tabling a few months, then 14 tabling, and now i 10 table (not because I can't handle 12, but because ive become lazy and decided to just use 1 party account). Anyways, what I realized is that I have gone some weeks where I go through 700 SNGs. I tend to get lazy periods where I play around with MTTs and such and I also get some weeks where I get super motivated to play massive SNGs. Well the point is multitabling enables a person to amass a large sample size within a short period of time.

By getting a decent sample size, you could probably say to yourself, "hey look I've got 1k games, I now have a pretty good idea of what my roi is." What's missing here is the human element of performance in these numbers. I think we can all agree that we can maintain our best performance in whatever we're doing for only a certain amount of time, after that our performance deteriorates, until we rest again.

For poker it's not so simple, it's a constant grind week over week, month after month... etc. The time we 'rest' isn't really 'rejuvenating' our juices, as poker is a slow slow game. Our best performance at any game must last many hours in any poker session to show any effects at all. To truly 'rest' we must take a prolonged break, maybe 2 weeks a month?

I sort of got side tracked, but stay with me here- because we as poker players are probably not at our 'best' most of the times, it means that the element of 'real time' must be considered in calculating our stats. So, if I played 1000 SNGs in a short 2 week span, that 1k set is good for a indication of perhaps my best performance. If I played 1000 SNGs over 4 months it would be even better, because now this set of 1k SNGs is also indicative of a longer period of time, where I might have had streaks of tiltiness, mixed in with times where I have performed great.

The 2nd set of 1k SNGs is far superior to the 1st set because it has a sort of 'average' of my performance capabilities (a-game,b-game,c-game,...whatever game you can deteriorate to).

So what does this all mean? A lot of the stats that we are getting from multitablers are probably very inaccurate, because they do not inidicate 'true' roi. 'True' roi I think must be a combination of a large sample size and it also must span over a good period of time, perhaps at least several months. If you can play well this month, does it mean you can play as well next month? What about the month after that, can you say for certain that you won't tilt?
A lot the roi we see posted are probably isolated periods where the player is playing near his peak performance. We may not realize this because it is at least a semi large sample size, which was achieve in a short period of time by multi tabling.

Sorry this got sort of long, I just had to write down these thoughts somewhere, so I don't forget and fool myself with inflated periods of roi and get too overconfident.

[Irieguy]

Interesting post. Thank you.

Irieguy

PS- I also like your use of the word "batches"

[microbet]

On the other hand, many people change their style of play a bit over time and a sample over a longer period of time may not reflect on the ROI of one particular style. Also, the competition can change over time.

[ QUOTE ]
On the other hand, many people change their style of play a bit over time and a sample over a longer period of time may not reflect on the ROI of one particular style. Also, the competition can change over time.

[/ QUOTE ]

Agreed. Your best bet of knowing what your real ROI is just to calculate your standard deviation.

Still a great post though.

[citanul]

[ QUOTE ]
Interesting post. Thank you.

Irieguy

PS- I also like your use of the word "batches"

[/ QUOTE ]

grrr... i'm gonna have to disagree with you here, this post was, for me, quite mind numbingly poor. very little in here hasn't been said hundreds of times before, and more concisely, by yourself, for instance. (hey, i went all capt kirk again in that last sentence.) not only does he harp on the obvious, he clearly overestimates the value of a 1000 sng sample, and underestimates the true value of luck (ie, hasn't played around very much with an ROI simulator).

continuing, he "takes in to account" a few factors, but totally doesn't really. such as, if you play poker in say, 1000 games in a week, it's not actually that much more likely that you avoid tilt during that time period than if you play them over a 1000 day span, as well, the poker stuff still hits you. in fact, i'd say that the more spread out your games are the less likely you're going to experience massively intergame tilt. that seems fairly intuitive. if i lose today, get upset, and don't play my next game for a month, chances are i'm going to be over it when i get to the next game. i find that most players i speak to who have tilt problems have those problems hit them hardest when they try to massively multitable their way through it.

the observation, however, that he makes, that it is likely that good stats we seen posted are often from short time spans of running good and playing good is a reasonable one. again, one harped on by hundreds of other posters, including yourself.

blah, i really don't want to go on much more about this.

c

[ilya]

[ QUOTE ]
one harped on by hundreds of other posters

[/ QUOTE ]

not hundreds, cit...thousands! thousands.

[FlyWf]

Well, the problem with reviewing 1000 games over 3 years is that you're probably a very different player in game 884 than you were in game 72(unless you were already fantastic or are extremely dense). In 'batches' from short time periods it's likely you were playing the same general style.

[mlagoo]

please post more. would love to hear your opinions on hand lines, general philosophy, etc., having had so much experience. nice post.

[citanul]

[ QUOTE ]
Well, the problem with reviewing 1000 games over 3 years is that you're probably a very different player in game 884 than you were in game 72(unless you were already fantastic or are extremely dense). In 'batches' from short time periods it's likely you were playing the same general style.

[/ QUOTE ]

my general point is that much in the same way as there are tons of factors which make large differences in one's game in the super spread out way, there are tons of factors that make your roi vary in the super short run as well. if i had the time to do entirely pointless exercises, i'd say "for any stupid thing that can be counted as a long run thingy, i'll give you a short run thingy" or whatnot.

for example: while it's very likely that over hte course of 3 years playing 1 a day your play changes stylisticly, it's also very true that playing 150 games a day for a week probably means that you're not playing close to the same way at either the end of each day as you were in the beginning of that day, or at the end of the week the way you were playing at the beginning of the week.

c

[FlyWf]

Do you think tilt and fatigue are comparable to the playstyle changes caused by experience/reading/thinking about poker?
I doubt it(and if they are, that's a "not able to be a longterm winner" level problem). My Xth SNG in a day might be full of ridiculous mistakes, but N years ago I would've been stack donating with J9o at l1.

There are obviously tons of "But well..." [censored] counters to each side(which is why stats posts are masturbatory wastes of time), but if you made me pick shortterm or longterm batches shortterm will tell you a lot more than long about a player's true ROI.

[SumZero]

What happens if you break your 8000 SNG into 8 batches by taking each 8th SNG? What are the ROI and what not of those batches. That might help clue you in to how much the changes were due to random luck in how you happened to bunch them or how much the changes were due to actual changes in your playing skill or the environements in which you play.

[Insty]

[ QUOTE ]
What happens if you break your 8000 SNG into 8 batches by taking each 8th SNG? What are the ROI and what not of those batches. That might help clue you in to how much the changes were due to random luck in how you happened to bunch them or how much the changes were due to actual changes in your playing skill or the environements in which you play.

[/ QUOTE ]
Interesting, do this.

[citanul]

[ QUOTE ]
true ROI.

[/ QUOTE ]

i think there's some massive misunderstanding of what this term means, even by a player like you, who seems totally wanting to figure things out, and approaching the largest sample sizes reported.

[ QUOTE ]
As the title suggests I have a crap load of 30+3 sngs recorded, and maybe another 2000-ish not recorded, and another 1000 50+5 and 100+9 sngs not recorded. I was looking over my stats and some batches of 1k sngs showed some really weird things.

I have 2 sets of 1k SNGs at 29% roi and 27% roi, and I have one set of 1k SNGs at negative 2% roi. My total (recorded) roi is at 17% roi for the 30+3s. What I rationalized is that it definately couldn't be only luck that could affect a set of 1k SNGs so dramatically, I mean there's a 31% roi difference on the end of each spectrum of my 1k batches.

All my sngs were done 8-14 tabling, 8 tabling maybe 5 months ago, started 12 tabling a few months, then 14 tabling, and now i 10 table (not because I can't handle 12, but because ive become lazy and decided to just use 1 party account). Anyways, what I realized is that I have gone some weeks where I go through 700 SNGs. I tend to get lazy periods where I play around with MTTs and such and I also get some weeks where I get super motivated to play massive SNGs. Well the point is multitabling enables a person to amass a large sample size within a short period of time.

By getting a decent sample size, you could probably say to yourself, "hey look I've got 1k games, I now have a pretty good idea of what my roi is." What's missing here is the human element of performance in these numbers. I think we can all agree that we can maintain our best performance in whatever we're doing for only a certain amount of time, after that our performance deteriorates, until we rest again.

For poker it's not so simple, it's a constant grind week over week, month after month... etc. The time we 'rest' isn't really 'rejuvenating' our juices, as poker is a slow slow game. Our best performance at any game must last many hours in any poker session to show any effects at all. To truly 'rest' we must take a prolonged break, maybe 2 weeks a month?

I sort of got side tracked, but stay with me here- because we as poker players are probably not at our 'best' most of the times, it means that the element of 'real time' must be considered in calculating our stats. So, if I played 1000 SNGs in a short 2 week span, that 1k set is good for a indication of perhaps my best performance. If I played 1000 SNGs over 4 months it would be even better, because now this set of 1k SNGs is also indicative of a longer period of time, where I might have had streaks of tiltiness, mixed in with times where I have performed great.

The 2nd set of 1k SNGs is far superior to the 1st set because it has a sort of 'average' of my performance capabilities (a-game,b-game,c-game,...whatever game you can deteriorate to).

So what does this all mean? A lot of the stats that we are getting from multitablers are probably very inaccurate, because they do not inidicate 'true' roi. 'True' roi I think must be a combination of a large sample size and it also must span over a good period of time, perhaps at least several months. If you can play well this month, does it mean you can play as well next month? What about the month after that, can you say for certain that you won't tilt?
A lot the roi we see posted are probably isolated periods where the player is playing near his peak performance. We may not realize this because it is at least a semi large sample size, which was achieve in a short period of time by multi tabling.

Sorry this got sort of long, I just had to write down these thoughts somewhere, so I don't forget and fool myself with inflated periods of roi and get too overconfident.

[/ QUOTE ]

Are you CrzyChinaman on Party?

[FlyWf]

I'd define "true ROI" as "expected return on the next investment." What is your definition?

[citanul]

[ QUOTE ]
I'd define "true ROI" as "expected return on the next investment." What is your definition?

[/ QUOTE ]

it's exactly that. but you have to "expect" your tilting runs, your game improvement, and all that fancy crap.