Check out our brand new Local Poker Communities! Get updates and interact with poker players in your area.
Visit the United States Poker Community | Visit the California Poker Community | Read more about the Launch of P5s Local

Sample Sizes & determining "true" ROI

Reply to Thread
  1. BlueRenegade BlueRenegade is offline

    Vancouver, WA
    United States

    Posts: 58
    Joined: Jul 09

    May 22nd, 2010 (5/22/2010 11:50pm)

    Hey guys,

    Anybody know the numbers (or got a link to an online calculator) on how big a sample size you need to have X% confidence that your actual ROI is close to your "true" or expected future ROI?

    For instance, I've played about 600 DoNs and have an ROI of around 6%. I'd like to figure out how big a sample size I need to be confident that that ROI will hold. I know it can be done with standard deviation & whatnot, but it's been a damn long time since I've had a statistics class.

    Mainly interested in the #'s for DoNs/single-table SNG's, but would be cool to know or be able to figure out the #'s for 180-mans/MTT's as well.

    Can any kindly stats geeks point me in the right direction?

    Thanks,

    Blue
  2. EyeKnows EyeKnows is online now

    Posts: 6,647
    Joined: Sep 06

    May 23rd, 2010 (5/23/2010 12:13am) in reply to BlueRenegade

    just gonna guess you will never play enough before you could get to the 90% confidence level, like 15-20K sample size, totally guessing and would also like to see.
  3. ntanygd760 ntanygd760 is offline

    Fort Drum, NY
    United States

    Posts: 5,495
    Joined: Jul 08

    May 23rd, 2010 (5/23/2010 12:13am) in reply to BlueRenegade

    play 100 billion of them then u might have a close to true roi
     
  4. BlueRenegade BlueRenegade is offline

    Vancouver, WA
    United States

    Posts: 58
    Joined: Jul 09

    May 23rd, 2010 (5/23/2010 1:10am) in reply to ntanygd760

     
    Originally Posted by ntanygd760 View Post

    play 100 billion of them then u might have a close to true roi

    So you don't know either! Great, along with EyeKnows, that's three of us that could use a good point in the right direction.

    Blue
    Thread Starter
  5. ntanygd760 ntanygd760 is offline

    Fort Drum, NY
    United States

    Posts: 5,495
    Joined: Jul 08

    May 23rd, 2010 (5/23/2010 1:52am) in reply to BlueRenegade

    no one knows, you can estimate but it will never be perfect. i mean somewhere in the universe im going to ship the sunday million by getting AA every hand
     
  6. CadderlySoaring CadderlySoaring is offline

    TX
    United States

    Posts: 216
    Joined: Feb 10

    May 23rd, 2010 (5/23/2010 2:37am) in reply to ntanygd760

    What's a "true" ROI? Is that like finding a mythical part of female nether regions?
  7. savvy savvy is offline

    Deer Park, NY
    United States

    Posts: 3,815
    Joined: May 05

    May 23rd, 2010 (5/23/2010 2:39am) in reply to ntanygd760

     
    Originally Posted by ntanygd760 View Post

    no one knows, you can estimate but it will never be perfect. i mean somewhere in the universe im going to ship the sunday million by getting AA every hand

    my head just exploded
  8. BlueRenegade BlueRenegade is offline

    Vancouver, WA
    United States

    Posts: 58
    Joined: Jul 09

    May 23rd, 2010 (5/23/2010 3:09am) in reply to ntanygd760

     
    Originally Posted by ntanygd760 View Post

    no one knows, you can estimate but it will never be perfect. i mean somewhere in the universe im going to ship the sunday million by getting AA every hand

    Yes, an estimate. Like I said, to X% confidence...that's the best statistics can really EVER do, and that's all I was asking for.

    And I think you'll need multiverses to ship the Million that way.

    Peace,
    Blue
    Thread Starter
  9. ntanygd760 ntanygd760 is offline

    Fort Drum, NY
    United States

    Posts: 5,495
    Joined: Jul 08

    May 23rd, 2010 (5/23/2010 4:17am) in reply to BlueRenegade

     
    Originally Posted by BlueRenegade View Post

    Yes, an estimate. Like I said, to X% confidence...that's the best statistics can really EVER do, and that's all I was asking for.

    And I think you'll need multiverses to ship the Million that way.

    Peace,
    Blue

    not true (not like any of us know) but in theory if the universe is endless then there will be acopy out somewhere, actually endless copys but thats just theoretical stuff meant to fuck with your head
     
  10. ntanygd760 ntanygd760 is offline

    Fort Drum, NY
    United States

    Posts: 5,495
    Joined: Jul 08

    May 23rd, 2010 (5/23/2010 4:30am) in reply to ntanygd760

    to actually answer your question id say 10k double or nothings would prob get you to +/- 3%

    im guessing though, not in the mood to actually do math
     
  11. BlueRenegade BlueRenegade is offline

    Vancouver, WA
    United States

    Posts: 58
    Joined: Jul 09

    May 23rd, 2010 (5/23/2010 11:32am) in reply to ntanygd760

     
    Originally Posted by ntanygd760 View Post

    to actually answer your question id say 10k double or nothings would prob get you to +/- 3%

    im guessing though, not in the mood to actually do math

    Thank you. That sounds, intuitively, about right...guess I'm gonna have to find some kinda stats program that works with standard deviation or else get a book and break out an abacus and a slide rule, lol.

    Peace,
    Blue
    Thread Starter
  12. StaySuckaFree StaySuckaFree is offline

    Posts: 671
    Joined: Feb 08

    May 23rd, 2010 (5/23/2010 11:43am) in reply to BlueRenegade

     
    Originally Posted by BlueRenegade View Post

    standard deviation or else get a book and break out an abacus and a slide rule, lol.

    You just found your "Point in the right direction"

    IMO, for DoN's I would say you could start to gather with about 5% confidence your True ROI after about 1500-2k.

    Yes I know that's an extremely small Sample.
  13. lordxixor101 lordxixor101 is offline

    Westland, MI
    United States

    Posts: 6,364
    Joined: Mar 06

    May 23rd, 2010 (5/23/2010 3:13pm) in reply to BlueRenegade

    Everybody seems to think the number is huge. In a SNG type setting, I really have to think 1-2k will get you a pretty good idea of what your ROI is at a setting (assuming it's a fairly accurate sample, meaning you weren't drunk for half of them, etc).

    People think for a poker tournament, you need a huge sample (and for MTT's, you'll need more, just because you have so many people in them, that the variance is huge). But, remember, the polls they show on the news for Presidential elections is usually based on a couple thousand people. A couple of thousand will get you a pretty good idea of where you stand (now,this is just informational, though I believe this, I think your nuts if you play 1,000 DON, show you win 10% and then quit your job)
Reply to Thread