Updated Simulation of Biggest Buyin Drop Expected: 43 (15/14/13 ratio)

[Nicholasp27]

my previous code had gotten maximum negative value that the br could dip, which is basically risk of ruin because i made a fundamental mistake

i've revised it and it now gets the biggest drop over 10k tournaments...it simulates tournaments at a 15/14/13 ratio of 1/2/3 and then tracks the highest the br ever gets and then the furthest it ever gets from the current highest br, thus getting the largest dip, which would allow you to withdraw anything more than that amount of buyins whenver you want



after 10,000 simulations of 10,000 tourney sets, the largest drop was 43 buy-ins...thus if you have a 15/14/13 ACTUAL ratio, then you could deposit 44 buy-ins into your account and withdraw the extra whenever your account goes above that amount


example:
you are a 10+1 player, so you deposit 44*11=$484 into your account. You can now withdraw the extra anytime your account balance goes above $484, bringing the balance back to $484. So if you play for a living, you now know what bankroll you must keep in your account for poker and what winnings you can cash out for food/rent/fun

[eastbay]

Would would be more interesting is a table of drops and % likelihood in a stretch of 10k.

buy-in %
10: 92
20: 68
etc.

eastbay

[Nicholasp27]

so just track the count of each specific drop amount and return the % for each?


btw, 24.46 was the average drop for 1000 sims of 10k sets

[Nicholasp27]

yeah, a risk of ruin for when you withdraw the profits instead of leaving them in like the normal calculation

i'll run this later and do a table of %chance of losing X buyins with a/b/c ratio...i'll have it run for a variety of possible ratios

[Mr_J]

"i'll have it run for a variety of possible ratios"

How about just labelling ROIs, pretend 1st-3rd is evenly distributed or something. Could you also calculate it as if most or all profits are reinvested up to say a 100 buyin BR? Wouldn't mind knowing my chances of having a go at the $109s and "never looking back".

[Nicholasp27]

rois can be made up of different ratios and most people aren't even on 1/2/3...some are more aggressive on bubble and get more 1s than 2/3...some are just more aggressive once itm so get lots of 1/3 and not so many 2s...some suck hu so get lots of 2/3 but not so many 1s...etc


it's more comprehensive to give a large variety so anyone can look at their personal sngtracker(etc) results and see exactly what their ror is if they only keep X bankroll in account


i'm not sure i understand what u are asking with regards to the reinvesting...the standard risk of ruin on eastbay's excel file gives u the ror for 100 buyins assuming you keep the profits in your br and don't withdraw them

if u want something different, give me an example so i can see what u are talking about

[bones]

Nice work, but I'm still the sweet 16 master.

[Nicholasp27]

sounds like a challenge

[Mr_J]

"rois can be made up of different ratios and most people aren't even on 1/2/3"

I understand this, but hardly anyone will have a 15/14/13 ratio, thats 42% ITM for 31% ROI!!! How much does variance increase for 40% ITM vs 38% ITM but both at the same ROI??

"it's more comprehensive to give a large variety so anyone can look at their personal sngtracker(etc) results and see exactly what their ror is if they only keep X bankroll in account"

Yeh that'd work /images/graemlins/wink.gif

"i'm not sure i understand what u are asking with regards to the reinvesting..."

Eg what would my ROR over my first 1000 sngs be if I started with a BR of 25 buyins, never withdraw profits and hit 10% ROI at 36% ITM.

[Nicholasp27]

oh ok

that's what my original function did..it got the lowest point the br dropped from the original br in 10,000 sngs...with 15/14/13 it was 23 or something

u can use aleo's to get 13.33% ror with 36.25%itm/10.333%roi, but that's ror EVER, not for 1000 tournies

[Mr_J]

"with 15/14/13 it was 23 or something"

Remember that ratio gives 31% ROI, and no-one should be achieving that longterm (if you are it's time to move up).

"u can use aleo's to get 13.33% ror with 36.25%itm/10.333%roi, but that's ror EVER, not for 1000 tournies"

Eh same thing. If you can hit 10% over the first few hundred sngs you're pretty safe.

[Nicholasp27]

i don't have time to do the whole chart right now...but if u wanna tell me your 1/2/3 ratio, i'll tell you what your average expected biggest drop is in 10k sngs

[Back In Black]

How about for 13.5, 12.5, 11.5? That is 37% ITM and 15% ROI for me in the 50's right now.

[Nicholasp27]

53 buyins was biggest drop in 1000 sims of 10k sets

29.74 was the average, so 30 buy-ins is pretty safe, but 55 is better

[Irieguy]

[ QUOTE ]
How about for 13.5, 12.5, 11.5? That is 37% ITM and 15% ROI for me in the 50's right now.

[/ QUOTE ]

Yes, that's a more realistic distribution for a very good player. I'd like to see the ROR/max downswing numbers for this, too.

Interestingly, though, I will say this: I've played thousands of SNGs at just about every level and have never had a losing "block" of more than 500. My long term distributions and ROI are in line with the numbers above. Yet, I'm certain that I've had a bigger downswing than should be possible and I know at least 2 other pros who could say the same thing.

There must be some pretty significant intra-session and inter-session variables contributing to losing stream probabilities. Everybody that I know who plays 1000 or so SNGs a month and beats the piss out of them drops 30+ buy-ins quite regularly.

Similarly, you can "prove" mathematically that a 2BB/100 limit holdem winner should never ever lose more than 400 BBs... but I know of more than 10 players who win at this rate or better and have dropped 500+ BBs on a cooler.

Meh.

Irieguy

[Nicholasp27]

well this is based on your numbers being a certain ratio...but most people, when on a very bad streak, will not play 100% and thus will be lower than what they think their ratios are...remember, this is based on ACTUAL ratio, which you never know for sure and can't really approximate very accurately for thousands of sngs...and even then u change your game, etc and your ratios change..

[Irieguy]

[ QUOTE ]
well this is based on your numbers being a certain ratio...but most people, when on a very bad streak, will not play 100% and thus will be lower than what they think their ratios are...remember, this is based on ACTUAL ratio, which you never know for sure and can't really approximate very accurately for thousands of sngs...and even then u change your game, etc and your ratios change..

[/ QUOTE ]

Yes, I see.

Anyways, when you ran the numbers for the more modest ratios the downswing max fell more in line with what I've seen anyways.

Thank you very much. This kind of stuff is helpful for sanity, I think.

Irieguy

[Slim Pickens]

[ QUOTE ]
well this is based on your numbers being a certain ratio...but most people, when on a very bad streak, will not play 100% and thus will be lower than what they think their ratios are...remember, this is based on ACTUAL ratio, which you never know for sure and can't really approximate very accurately for thousands of sngs...and even then u change your game, etc and your ratios change..

[/ QUOTE ]

This is why I was hoping you'd put some uncertainty into the distribution during a single run of 10,000. I made this post (http://forumserver.twoplustwo.com/showflat.php?Cat=&Number=2257706&page=&view=&sb=5& o=&fpart=1&vc=1) a while ago after thinking about doing something similar to what you've done. Do you think you could put some randomization into the finish distribution, say, chaging it every 10 tournaments selecting from a normal distribution? Make the mean 38%ITM 15% ROI with some set 1/2/3 relative fraction and then build the distribution around that? Nice work so far BTW.


SlimP