The importance of seat selection

[elindauer]

Another thread got me thinking about seat selection, and it's impact on your bottom line. Most people have heard the following:

[ QUOTE ]
You make most of your money from the players 1 or 2 seats to your right. You lose most of your money to the players 1 or 2 seats to your left

[/ QUOTE ]

I propose we try to test this theory by showing where our EV comes from in various positions.

Pokertracker already tells us, for example, that we make .08 BB / 100 in the hijack, or .06 BB / 100 button + 5. What it doesn't tell us, is how much of that money comes from what other seats? How much do we make in the hijack from the player UTG? How much do we lose to the button?


I think the hijack is an interesting seat to examine because the two players left and right of us are not posting blinds. Anybody have any suggestions for how to make this calculation?


My hope is that this project will offer some more insight into the age-old questions of where to sit and how important seat selection is to your bottom line.


thanks,
eric

[JAA]

Unfortunately I don't have the know-how to conduct this study. However, I will predict that if it is done, the results will show that the typical "most of the money you make comes from those 1-2 seats to your right and most of the money you lose goes to the players 1-2 to your left" is way way overstated. I would predict that you make noticeably more from the dude one to your right versus the dude one to your left, but the correlation between "seat in relation to you" and "profit made" will not be as strong as some would think.

I hope someone smarter than me does this.

- Jags

[Dazarath]

I think this would be an interesting project. I guess the way to go about doing this would be to run SQL queries on the database. I only know a little SQL, and I could take a look at it later, but I can't guarantee anything. If someone does come up with the queries, would people be willing to run them and share the data so it can all be combined into one large table?

[elindauer]

[ QUOTE ]
If someone does come up with the queries, would people be willing to run them and share the data so it can all be combined into one large table?

[/ QUOTE ]

Naturally, I'd be willing to run these queries and donate results. Anybody else?

-Eric

[Steve Giufre]

[ QUOTE ]
[ QUOTE ]
If someone does come up with the queries, would people be willing to run them and share the data so it can all be combined into one large table?

[/ QUOTE ]

Naturally, I'd be willing to run these queries and donate results. Anybody else?

-Eric

[/ QUOTE ]

Sure. Im pretty bad with poker tracker so somebody will have to walk me through it, but I have a lot hands. I usually sit next to the guy who I think is going to aggravate me the least. Im not sure how much money is costs me, but I wouldnt mind finding out.

[Danenania]

This sounds like a job for Peter_rus. Anyone seen him around lately?

[Dazarath]

I took a quick look at the PT database earlier. Here are a couple issues that I think could cause trouble:
1) Dealing with the rake. It's pretty easy if 3 people all go to showdown and the rake is $3, but what happens when people fold the flop? I'm not sure how one would want to define what percent of the rake comes from what player. It's possible to just ignore the rake, but then the BB/100 will be overinflated.
2) Chopped pots. It just looks like a pain to write a query to deal with these. Maybe I'm just lazy.
3) Short-handed pots. I'm thinking that the queries should just be written to only look at 10-handed games. Otherwise the stats would be kind of skewed up if you're taking 7-handed stats combined with 10-handed stats.

I was thinking of writing a simpler version of the query that just adds up the total amount won and then return what percent of it came from positions 1-9 to the right (or left) of you. I think this way, the rake could be ignored, and the stats would only be taken from 10-player hands. This isn't as in-depth as what was originally suggested, but it's a starting point.

[The Truth]

[ QUOTE ]
I think this would be an interesting project. I guess the way to go about doing this would be to run SQL queries on the database. I only know a little SQL, and I could take a look at it later, but I can't guarantee anything. If someone does come up with the queries, would people be willing to run them and share the data so it can all be combined into one large table?

[/ QUOTE ]

I've read 2 SQL database books. I did this to be able to write PT queries. Queries are actually very simple. I don't have microsoft access on my computer any longer, as I only had a pirated copy and had to reformat. If someone will buy me microsoft access, or send me a link to a pirated one with a key, I'll write em up within a week, as well as some other cool ones.


blake

[Dazarath]

If you have postgres installed, you can use pgadmin to run the queries.

[Peter_rus]

I made a quick query. Not with almost strict rules. But it shows tendency at first glance. Levels 15/30 and 30/60 included.

All positions counted beginning from the first to your left:
in terms of % of total winnings
1.+2.6%
2.-6.3%
3.+13.7%
4.+2.7%
5.+6.3%
6.+16.9%
7.+12.8%
8.+21.6%
9.+29.6%

Yes, hijack is really interesting to examing.