Worst case scenarios (Downswing simulations)

[dmmikkel]

The point of this post is to look at how downswings are related to what level you play at (or in other words your ROI)

These numbers are simulated from 1 million SNGs per ROI.
These are 10+1

ROI (finish distribution)

24% (14-12-13)
Longest OOTM streak: 27
Longest ITM streak: 15
Largest downswing: $522

20% (13-12-13)
Longest OOTM streak: 24
Longest ITM streak: 14
Largest downswing: $734

11% (12-11-12)
Longest OOTM streak: 29
Longest ITM streak: 12
Largest downswing: $1122

4.5% (11-11-11)
Longest OOTM streak: 33
Longest ITM streak: 12
Largest downswing: $2252

My own conclusion
- Playing lower buyins will definately lower your swings
- Impressive enough there's room for more variance than I thought over a million SNGs
- You can have pretty naaaaasty downswings

[runner4life7]

If if ever take 100 buy in downswing im killing myself.

[dmmikkel]

[ QUOTE ]
If if ever take 100 buy in downswing im killing myself.

[/ QUOTE ]

Then I would like to bet you would never ever hit a 200 buyin drop =)

[Nottom]

A million SNGs is a bit unrealistic for anybody but raptor don't you think?

[Xenod]

How do you conclude that the size of the buy-in effects the variance when this is all 10+1's? I know it's simmed anyway, and it seems like common sense, but it's not really proven.

[bkbluedevil]

Wow if you extend this to the 200 level, a winning player could have a 486,432 dollar down swing. I'd be thinking PP were rigged and switching sites long before that point though.

[Nottom]

[ QUOTE ]
How do you conclude that the size of the buy-in effects the variance when this is all 10+1's? I know it's simmed anyway, and it seems like common sense, but it's not really proven.

[/ QUOTE ]

Well the swings are obviously twice as big if these were 20s /images/graemlins/wink.gif

[flyingmoose]

Yeah, but after a million SnGs a 215 player with, say, a 10% ROI has made 215*0.1*1,000,000 = A [censored] TON OF MONEY, and he probably doesn't care much about the 500k downswing.

[dmmikkel]

[ QUOTE ]
Wow if you extend this to the 200 level, a winning player could have a 486,432 dollar down swing. I'd be thinking PP were rigged and switching sites long before that point though.

[/ QUOTE ]

No it would be more like 50k

[dmmikkel]

Because ROI goes down the higher you play. These numbers goes for all buyins, the only reason for me to say this is 10+1 is because largest downswing is in $$$

[DonButtons]

I hit -80 buyins at one point. Not cool.

[45suited]

Very interesting information.

I think that the estimates regarding the # of buy-ins needed to move up in levels tends to be a little low because of a factor that would not come up in a simulation, but would come up in real life for most players: after a prolonged downswing, even if due to bad luck, many (if not most) players will see, at least temporarily, a decrease in their level of play. Just guessing here, but I think that after a certain amount of bad luck, this would accelerate a downswing and make the numbers in your sim even more drastic. It would take a special player to not have his play affected at all in the midst of a 50 buy-in downswing IMO.

[Degen]

[ QUOTE ]
Wow if you extend this to the 200 level, a winning player could have a 486,432 dollar down swing. I'd be thinking PP were rigged and switching sites long before that point though.

[/ QUOTE ]

you miscalculated something there bro

[Degen]

great post..awhile back i went down 50-60 BI's total between 33's and 55's and made a post about whether this were statistically possible as a winner player and most seemed to think that it was not possible. tho i haven't played a million sng's so i guess i coulda been on tilt /images/graemlins/grin.gif

[ewing55]

What about the distribution of downswings? (ie. With a 20% ITM the biggest downswing was $734, What were the next biggest downswings and how many were there?)

I guess I mean you have a "one in a million" $734 downswing (sorta) but how about other downswings? What is the biggest "likely" downswing? (I know it's a vague question, but if you have the figures it's easier for you to look at the results and post a summary.

Great Work, but scary as hell!

Thanks!!!!

---------------Jeff

[FieryJustice]

If I had a 50k downswing, I would not be happy. I know how a 10k one feels and it made me quite sad.

[Slim Pickens]

I would also like to to see the distribution of downswings if you can get that from your program.

[eastbay]

Conclusion: don't play a million SnGs.

More to the point, this kind of simulation is probably more interesting if you break it up into 1000 SnG stretches, and then measure the probability that you hit various levels of downswing in those stretches.

eastbay

[kamrann]

Yep. Obviously stats for a million SNGs are kind of irrelevant anyway. But more to the point, since you give a longer longest OOTM streak for 24% ROI than for 20% ROI, this indicates that you just ran a single simulation for each finish distribution. Not surprising since 1 million SNGs could take a while to simulate. As eastbay suggests, you would get more accurate and meaningful results if you ran, for example, 1000 simulations for each distribution, over 1000 SNGs each, and averaged the results.

[jcm4ccc]

[ QUOTE ]


More to the point, this kind of simulation is probably more interesting if you break it up into 1000 SnG stretches, and then measure the probability that you hit various levels of downswing in those stretches.

eastbay

[/ QUOTE ] yes

[dmmikkel]

[ QUOTE ]
But more to the point, since you give a longer longest OOTM streak for 24% ROI than for 20% ROI, this indicates that you just ran a single simulation for each finish distribution. Not surprising since 1 million SNGs could take a while to simulate

[/ QUOTE ]

These should be pretty accurate numbers because, well nobody will play this many SNGs, but what you say is absolutely true.

The most interesting thing about this is that SOMEONE is going to have these kind of downswings. But even better, someone will have the opposite swing. Some -5% ROI player will earn $50k at the 215s.

[dmmikkel]

Ok. here are some more calculations. These are not as accurate as the other ones, but very interesting results

Simulations are 1,000 SNGs where worst run was noted. I ran this 70-80 and noted the worst run in each.

It was done using a finish distribution of 13-11-12

In 1,000 SNGs your worst run will be over
- 20 buyins 87% of the times
- 30 buyins 23% of the times
- 40 buyins 4% of the times

Small samplesize, but you get an idea

[valenzuela]

ouch, anyway I made my own simulations...with infinite SNGS.
24% ROI:
Biggest OOTM streak: 93483
Biggest ITM streak: 5402
Biggest downswing: $856900

20%ROI
Biggest OOTM streak: 457907
Biggest ITM streak: 2908
anyway u get the idea...while I apreciate the job done by the OP most of us wont even play 100k. Still those swings arent so unrealistic and they just might as well happen, Im not sure Im ready for a 102 buy-in downswing.

[dmmikkel]

Well your numbers can't be right.

With infinite number of SNGs the biggest downswing would always be infinite too.

It's possible to place OOTM in every SNG you play during a lifetime

[valenzuela]

[ QUOTE ]


It's possible to place OOTM in every SNG you play during a lifetime

[/ QUOTE ]

Isnt 457907 a lifetime?? /images/graemlins/tongue.gif. My PC collapsed with all the numbers and thats the data I colected, my data aint infinite but its a very high number that would colapse two plus two if I wrote it.( even with scientific notation or powers)

[DMACM]

I thought this was a very interesting and useful post. Is there any way the OP could run the numbers for a higher ROI?

[sahala]

Great post. Makes me feel a little better about my recent downswing.

Can I ask how you ran the simulations and how it was modeled?

[Newt_Buggs]

[ QUOTE ]
I thought this was a very interesting and useful post. Is there any way the OP could run the numbers for a higher ROI?

[/ QUOTE ]
why?

[curtains]

How do you guys do these simulations!! Tell me how, I want to do cut them into 32 sit and gos and then figure out what % of days I should be up $$$ /images/graemlins/smile.gif

And I can think of many other uses too...

[AleoMagus]

just whipped up a little excel program out of another one of my excel programs for you.

www.aleomagus.freeservers.com/Spreadsheet (http://www.aleomagus.freeservers.com/Spreadsheet)

the 'sample confidences' file

this will let you define any sample and it will tell you with what % you can expect to be running better or worse than any given profit value.

Math beats sims. Am I the only guy who actually just does the math?

Regards
Brad S

edit: If you don't know your SD, just estimate by multiplying your buy-in by 1.7. That or download the confidence calculator (from the same site) and it will tell you based on your finish distribution.

[raptor517]

oh god.. so it can get worse? /images/graemlins/frown.gif holla