Mean reversion or have I forgotten how to play?

[spots123]

Ok, perhaps just a cry for a pep talk here. But after playing my first 100 SNGs at pokerroom $10+1 I had done pretty well, ITM and ROI both almost 50%. Now, I haven't really changed the way I play but have dumped 21 games in a row. Best finish in any of them, 5th. Did I forget how to play or is it mean reversion? Thanks in advance.
Spots

[J-Lo]

21 in a row seems like alot,but most would agree this could be normal variance.

[Karak567]

I dumped 12 in a row today after a 40 % ROI over the last week.

Normal variance... I hope.

[ilya]

A lot of it may just be variance, although your sample size is small. However, 21 OOTM is a big streak. Are you sure you're not tilting?

[11t]

Look over hand histories and be honest.

[Karak567]

I had AA and KK busted in 5 of the tournies today... so I think that had a lot to do with it for me.

[AtticusFinch]

Let me be geeky for a moment and throw in an explanation of mean regression, because you clearly have the wrong idea.

Mean regression does not mean that you can expect to have a bad run after a good one, so it all "evens out". Assuming independent random trials (for simplicity, I know this is not wholly accurate, so save the flames /images/graemlins/smile.gif ), a run of 10, 100, or 1000 wins in a row has no effect on the likelihood that you'll win next time.

Regression to the mean is really an expression of how your results smooth out over time after a large number of trials. Let's say your ROI is 25%, but you happen to win your first 20 tourneys in a row. Your mean result will be way off from your "true" ROI for a long time. But if you then play 10000 tourneys that average out to roughly your true ROI rate, those first 20 will have a negligible effect on your overall average, and your mean result will come closer and closer to your true ROI.

It's just the tendency of a large number of trials to overwhelm short term spikes. Nothing more.

[skipperbob]

U better hope that Irieguy is in a good mood when he reads this /images/graemlins/grin.gif

[spentrent]

21 OOTM in a row is A LOT. Where's that thread that predicts how often a 45% ITM player will lose 12 in a row? I dunno, but it's not very often. Perhaps you went on a great card run, got over-aggressive. Did you say this is your first 100 games ever?

[spots123]

No, not first 100 games ever, just first 100 since I started keeping track of it with the spreadsheet posted on the site a couple of times. Could be I got a bit more aggressive, but my current bad streak is spread over a two week period which I would hope would mitigate possible tilting.

[spots123]

Atticus: Not to get nerdy on your nerdy post, but isn't what you described, by adding trials to my sample, I am really showing that my ability isn't really to be in the 40% ROI range as shown by smaller sample, but the total sample ROI is reverting towards my true mean ROI of 2% as demonstrated over my 150 or 1000 trials?

[stinkyman]

When I'm running bad I find that I am playing to aggressive with speculative hands. Early level 1-3...AK to KJ.

Throw some stiff jabs and then get popped with a mean uppercut. Gotta learn to go to the corner and wait for the bell. More than a three round fight. Most players pp 10+1 (where I play) don't have any legs for the later stages. Older George Foreman would be superb in sng's.

[raptor517]

21 in a row is VERY unlikely. i would love to see the percentage of times this would happen for a 40% itm player. worst run i have had was 14 ootm. its usually just variance, but the negative streak can cause strain on your psyche, thus causing you to lean towards incorrect play. i never have a problem loosening up too much early, i pretty much fold everything, but when the blinds hit 50-100 i turn into a maniac, and am trying to double up. you can go quite a while without winning a coinflip, and it sucks sometimes. definitely take a look at your early game play and see if you are giving much up there. post a few hand histories and maybe have some 2+2ers look them over, and watch them yourself, keep playing and you will be back on your feet in no time.

[AtticusFinch]

[ QUOTE ]
Atticus: Not to get nerdy on your nerdy post, but isn't what you described, by adding trials to my sample, I am really showing that my ability isn't really to be in the 40% ROI range as shown by smaller sample, but the total sample ROI is reverting towards my true mean ROI of 2% as demonstrated over my 150 or 1000 trials?

[/ QUOTE ]

That's certaintly possible, and if that's what you meant, then cool. I've just seen many people misinterpret the concept, so I thought I'd chime in.

Still, even with a 2% ROI, that many losses in a row would be very rare, and could mask your true ROI for quite a while. Heck, even with a negative ROI that many losses in a row would be uncommon.

So it's possible you were just getting lucky before, but I'd say the odds you're on a bad run here are 99%, whatever your true ROI is.

[SumZero]

The odds that a BE 33% player loses his next 21 in a row is about 1 in 5000. Which, if my Q&D math is right, means you'd expect it to happen roughly every 15000 SNG (as each losing streak happens on a lose after a win and that happens roughlky every third hand L,L,W,L,L,W,etc.).

The odds that a 45% in the money player loses his next 21 in a row is about 1 in 285000.

The odds that a 20% ITM player loses his next 21 in a row is about 1 in 108.

Conditional probability question: Who thinks OP is a winning player given he's just lost 21 in a row?

[lorinda]

The difficulty here is overcoming Lorinda's Paradox.

If _I_ (or in your case you) have a run of 21 OOTM, that's probably such a small percentage chance that I should worry.
However the odds of this post appearing on a forum with 20000 users, is really pretty likely.

So on the one hand it's a pretty unlikely event, and on the other hand it's almost certain.

This makes a reply difficult to say the least.

Lori