11-15-2012 , 07:56 AM
Warning: Wall of Text Alert. TL;dr = I'd buy this service if I were a serious online poker player. In a heartbeat.

I conducted an unscientific experiment. I ran 7 simulations. The first 6 involved 1,000 units. These could be hands, sessions, weeks, a chunk of hands, whatever. I am not smart enough to figure out whether I am in the right ball park on this but I thought I was going in the right direction. The 7th simulation involved 10,000 units. I used heads-up situations to keep it simple.

I assumed:
--that the potential customer of insured play has a 5% edge over his typical opponent
--that 25% of the units are insured

I used random.org to obtain 1,000 (10,000 in the 7th simulation) numbers from 1 to 1,050 and placed those in column A. These represent hero results.

I did the same for column B, only the numbers ranged from 1 to 1,000. These represent villain results.

Column C is the profit subtotal of A-B. The sum of column C represents profit over the 1,000 or 10,000 hands/other appropriate unit of measure.

For Column D I inserted a random number from 1 to 4, where 1 represents an instance of insurance.

For Column E I said if D=1 then 23.75 (insured), otherwise whatever was in column C stands (not insured). I chose 23.75 because the average result in column A is 525.5 and the average result in column B is 500.5, and the difference there is 25, so that is hero's EV each hand/other unit of measure. 25 is 5% of 500, representing the 5% edge. If the edge is reduced to 4.75% because of the insurance fees, hero's EV each hand/other unit of measure is reduced from 25 to 23.75. So the insurance situations are supposed to impose an EV result, so that's what I did even though in practicality it will fluctuate more wildly. But as your poker account shoots up your insurance account plummets and vice versa, so that creates an equilibrium that just isn't there without this product, so I think it was fair of me to analyze it this way, but I'm a history major so don't think I'm trying to be some probability or math pooh bah cuz Im not.

Finally in column G I calculated it if you could get this insurance product for free (if they ran the business without charging any premiums).

Here is the summary, and it's in spoiler because I put it in all caps because I didn't plan on posting it until after I was done, so if that bothers you don't hit spoiler.

Spoiler:
SUPERSIM -SHEET 7
HERO IN SUPERSIM RAN 531.6333 (EV 525.5)
VILLAIN IN SUPERSIM RAN 502.0148 (EV 500.5)
SIM RAN 2,529 TIMES (EV 2,500)
--PROFIT WOULD HAVE BEEN \$296,185 WITHOUT INSURANCE (EV \$250,000)

SIM1
SO COLUMN A IS 1-1050 (HERO) AVERAGE IS 525.5 (IN SIM RAN 532.39)
AND COLUMN B IS 1-1000 (VILLAIN) AVERAGE IS 500.5 (SIM 493.103)
SIM RAN 264 INSTANCES OF INSURANCE WHERE EV WAS 250
--PROFIT WOULD HAVE BEEN \$39,287 WITHOUT INSURANCE
--PROFIT WAS INSTEAD \$28,540 WITH INSURANCE
--EXPECTED PROFIT WAS \$25,000.
--PROFIT WITH INSURANCE IF NO 0.25% PREMIUM HAD BEEN CHARGED: \$28,870 (1.16% MORE)

SIM2
HERO IN SIM RAN 532.508
VILLAIN RAN 507.244
SIM RAN 260 INSURANCE INSTANCES
--PROFIT WOULD HAVE BEEN \$25,264 WITHOUT INSURANCE
--PROFIT WAS INSTEAD \$22,969 WITH INSURANCE
--EXPECTED PROFIT WAS \$25,000
--PROFIT WITH INSURANCE IF NO 0.25% PREM HAD BEEN CHARGED: \$23,294 (1.41% MORE)

SIM3
HERO IN SIM RAN 519.566
VILLAIN RAN 515.214
SIM RAN ONLY 230 INSURANCE INSTANCES
--PROFIT WOULD HAVE BEEN \$4,352 WITHOUT INSURANCE
--PROFIT WAS INSTEAD \$1,194.50 WITH INSURANCE
--EXPECTED PROFIT WAS \$25,000
--PROFIT WITH INSURANCE IF NO 0.25% PREM: \$1,482 (24.07% MORE)

SIM4
HERO IN SIM RAN 527.037
VILLAIN RAN 482.659
SIM RAN 272 INSURANCE INSTANCES!
--PROFIT WOULD HAVE BEEN \$44,378 WITHOUT INSURANCE
--PROFIT WAS INSTEAD \$36,102 WITH INSURANCE
--EXPECTED PROFIT WAS \$25,000
--PROFIT WITH INSURANCE IF NO 0.25% PREM: \$36,442 (0.94% MORE)

SIM5
HERO IN SIM RAN 534.348
VILLAIN IN SIM RAN 503.554
SIM RAN 260 INSURANCES
--PROFIT WOULD HAVE BEEN \$30,794 WITHOUT INSURANCE
--PROFIT WAS INSTEAD \$19,239 WITH INSURANCE
--EXPECTED PROFIT WAS \$25,000
--PROFIT WITH INSURANCE IF NO 0.25% PREM: \$19,564 (1.69% MORE)

SIM6
HERO IN SIM RAN 531.741
VILLAIN IN SIM RAN 499.691
SIM RAN 255 INSURANCES
--PROFIT WOULD HAVE BEEN \$32,050 WITHOUT INSURANCE
--PROFIT WAS INSTEAD \$21,649.25 WITH INSURANCE
--EXPECTED PROFIT WAS \$25,000
--PROFIT WITH INSURANCE IF NO 0.25% PREM: \$21,968

11-15-2012 , 08:38 AM
i don't see what the problem is in NL players using this. Emre said the players that use IP are generally of a higher caliber than the "industry average". So these are weekend warrior regs, perhaps shortstackers, reg-fish, etc.

Let them use it. If they're smart enough they'll read up, ask around and generally inform themselves. If not, survival of the fittest; evolution baby.

Quote:
I'm interested to hear people's theories on why the first 6 simulations always showed hero making more money without the insurance than with it.
Gee, I wonder why.
11-15-2012 , 09:31 AM
Unless I am missing something there does not seem much difference between this service and staking or selling shares. You are limiting your upside and limiting your downside. The "fee" IP collects above the EV is their revenue. The player pays this fee to feel mentally more comfortable and in return probably will allow him to play better, more detached. That said, it is not for everybody, but I know winning players who get staked for this precise reason.
11-15-2012 , 09:58 AM
IP, don`t waste your time with these FR nits. There is better market for you. HU Hyper Turbo Sit `n Go. I`d snap pay 1bb/100 to make sure i run at my ev (we rake 11bb/100 anyways).

No ICM in heads-up, that discounts problems you might get at other SnG
11-15-2012 , 10:55 AM
Quote:
Originally Posted by zilltine
IP, don`t waste your time with these FR nits. There is better market for you. HU Hyper Turbo Sit `n Go. I`d snap pay 1bb/100 to make sure i run at my ev (we rake 11bb/100 anyways).

No ICM in heads-up, that discounts problems you might get at other SnG
ICM based all-in equity insurance for SNGs is something we are looking into.

And yes it's very easy to do for heads up SNGs.
11-15-2012 , 12:15 PM
Quote:
Originally Posted by Emre Kenci
• KK is all-in on the flop with 80% equity in the \$100 pot
• AQ calls and KK gets automatically insured for \$20
• Ace drops on the river and KK loses the hand
• KK gets \$100 in his account on InsuredPlay
Hey,
Isn't the above quote incorrect? I notice it on the front page of your website too.

The KK wouldn't get 100 in his account, he would only get 80, no?
11-15-2012 , 12:19 PM
Hey,

AQ calls and KK gets automatically insured for \$20

KK pays the \$20 premium first and then gets \$100 in his account.

So KK's net winning is \$80.
11-15-2012 , 01:59 PM
Ok, thanks - maybe should make that clear [clearer?] on the site.

Thread is an interesting read. Ultimately people will decide for themselves whether insurance is worth it for them.

Fwiw, I've started to use it as an experiment ... so far so good.
11-15-2012 , 05:17 PM
Emre, I know you seem very proud of the fact that you have a whopping 5 SuperNova Elite customers, but let me try and break the numbers down for you real quick.

There are roughly 250 SNE players in the world, you have 5 of them, that means you have a monopolizing outrageous 2% share of the SNE market out there, not to mention there are about zero competitors in your industry. But of course your price is not too high, you have stated this, and the reasoning is because you and your partners are very smart people and know everything.
11-15-2012 , 06:33 PM
Quote:
Originally Posted by jaxtraw
Unless I am missing something there does not seem much difference between this service and staking or selling shares. You are limiting your upside and limiting your downside. The "fee" IP collects above the EV is their revenue. The player pays this fee to feel mentally more comfortable and in return probably will allow him to play better, more detached. That said, it is not for everybody, but I know winning players who get staked for this precise reason.
The difference is that cash game EV insurance is a totally new product that is more comfortable for the insurer than traditional staking because the EV difference (pre-river AI swing) is independent of the player, hence there's no need to check how well the insured person plays and no need to filter out bad players. The insurance is bad for bad players, unlike a traditional stake, but they can be unaware of it and take the bait. The profitability of tournament insurance does depend on the player's style, but the company hopes that no person will play suboptimally only to destroy them, for normal playing styles the company gains profit on average.
Quote:
Originally Posted by starrazz
<tl;dr> ... Consider this my "veteran" post.
That was a nice try by a historian to do sims, but they tell hardly anything imho. The first 6 sims are sooo swingy that Hero has a whopping 45-49.5% chance of hitting such a cooler that he makes more money with insurance than without it, whereas clearly insurance costs him 312.5 on average. Since the standard deviation is ~4564, four-digit and even five-digit swings are not uncommon. The chances of Hero profiting from insurance in exactly one of six sims are 9.8-13.6% (by the binomial distribution law, 6*0.495*0.505^5~9.8%, 6*0.45*0.55^5~13.6%).The supersim is swingy too, the chances of the insurance being profitable are 40.5-42.1%. Boring math details:
Spoiler:
The difference between what Hero makes by uninsured and insured (1/4 of the time) play in a single 'hand' is a random variable that takes value 0 with probability 3/4 (when insurance isn't triggered) and values from -525.75 to 523.25 (the fee; btw it should be -2.5 as it's IP fees are percentages of the whole pot) with probability 1/4200 each. It has mean -0.3125, and it follows from the formula for the variance of a uniform discrete distribution that it has variance ((1001*999)/12)/4+(1.25-0.3125)^2/4+0.3125^2*3/4=999999/48=20833,703125. By the Korolev-Shevtsova estimate, the sim result - a sum of 1000 independent variables identically distributed as above - is quite close (the difference in distrubution functions is less than 0.022) to a normally distributed variable with EV -312.5 and standard deviation ~sqrt(20833703)~4564.4. Insurance brings him more money if the difference is more than 312.5/4564.35~0.068 standard deviations above EV, which happens in ~47.3% of all cases; subtracting the <2.2% error of the normal approximation, we arrive at a >45% chance of insurance being profitable. In the supersim, the EV is -3125, the std dev is ~14434, the error of the normal approximation is <0.67%, the chances of running 0.2165 std dev above EV are ~41.42% in the norm. approx, so chances are of winning from the insurance are >40.5%.

If a person plays the same stake all the time, all-in insurance is of course -EV. The positive effect from it comes when the person using it moves up stakes faster. Boring higher math rants again (why and when the Sharpe ratio defined in my above post is important):
Spoiler:
Assume that the gambler has a logarithmic utility function (a common assumption), e.g. he's glad to wager his entire liferoll in a lottery where he gets back one tenth of it if he loses and ten or more times his roll (but not less) if he wins. To maximise utility EV, a gambler should use the Kelly betting strategy which tells to buy in for the share of the roll that equals the ratio of the winrate (in buy-ins, in comparison with a risk-free asset like a real job) to the variance. E.g. for PLO with winrate 0.034 (3.4 bb/100) and variance 1.7 (std dev 1.3 BI/100, 1.3*1.3=1.69) it tells to buy in for 0.034/1.7=2% of the roll, i.e. to have a roll of 50 BI.

This view is simplified because winrates and variances are different at various limits, and also grinders are more risk averse and have bigger rolls, partly because they don't know their winrate for sure. To find out which kind of bets (e.g. with or without insurance) will lead to a faster roll growth under the Kelly strategy, the grinder should compare their Sharpe ratios. A sample bankroll management guideline based on adjusted Kelly betting is in Quadrophobia's blog post.

Btw I think Quadrophobia can promote the company very well .
11-15-2012 , 06:43 PM
Can you add player transfers as a way to deposit/cashout ?
11-15-2012 , 07:02 PM
Quote:
Originally Posted by MeleaB

Even the 0.25% fee is very, very, expensive.
LOL, are u a rec player or a cheap bastard?

Last edited by paletokio; 11-15-2012 at 07:04 PM. Reason: Great service btw!
11-15-2012 , 07:05 PM
Quote:
Originally Posted by K2AA72
...If a person plays the same stake all the time, all-in insurance is of course -EV. The positive effect from it comes when the person using it moves up stakes faster...
I'm unsure of the thinking behind the claim that using this insurance service allows you to play under-rolled. Suppose you wish to shot-take for a period of time and so choose to use IP to help you in case you end up running \$x below EV. The fact that you are equally likely to run \$x above EV, means that your IP account needs to be funded with money that would otherwise be part of your poker bankroll, so it doesn't really allow you to play under-rolled in the way it is claimed.

E.g. a 100NL player with 40 buy-ins (\$4k) decides he wants to use IP to shot take at 200NL for the weekend as he doesn't want to risk running below EV and his bankroll isn't big enough to play 200NL on his own. He's advised that he could easily experience a \$2k+ downswing in EV over that short time period so IP sounds ideal for him. The only problem is he can't sit down at 200NL with his \$4k roll, as he has to fund his IP account with half of it!
11-15-2012 , 08:17 PM
Quote:
Can you add player transfers as a way to deposit/cashout ?
This would be sweet if posible.
11-15-2012 , 08:38 PM
I assume that shot-taking is a more continuous process than MeleaB described, i.e. if he 9-tables (I don't know his multitabling coeff, it's likely high if he complains about money flow at all), if the roll is up to \$4500, he can open 8 tables of 100NL and 1 table of 200NL, playing effectively 111NL, then, if he's up by 5 BI, close the worst of the 100NL tables and open a second 200NL one, moving up to 122NL, preserving the bankroll-to-buy-in ratio at about 40 or w/e. As only 20 BI are needed to put in an hour 9-tabling session comfortably, I think 15 BI can be held in the IP account (chances are 90% that it will suffice for 30-40 insured hands).

But of course player transfers as a banking option would solve the problem. However, that may require hiring escrowers - reputable HS players who can play such amounts through, as tossing money back and forth without playthrough will be suspicious.
11-15-2012 , 10:39 PM
I wouldnt complain any transfer limits
11-15-2012 , 11:29 PM
Quote:
Originally Posted by MeleaB
I'm unsure of the thinking behind the claim that using this insurance service allows you to play under-rolled. Suppose you wish to shot-take for a period of time and so choose to use IP to help you in case you end up running \$x below EV. The fact that you are equally likely to run \$x above EV, means that your IP account needs to be funded with money that would otherwise be part of your poker bankroll, so it doesn't really allow you to play under-rolled in the way it is claimed.

E.g. a 100NL player with 40 buy-ins (\$4k) decides he wants to use IP to shot take at 200NL for the weekend as he doesn't want to risk running below EV and his bankroll isn't big enough to play 200NL on his own. He's advised that he could easily experience a \$2k+ downswing in EV over that short time period so IP sounds ideal for him. The only problem is he can't sit down at 200NL with his \$4k roll, as he has to fund his IP account with half of it!
This is an excellent point. If there were some way to have the poker site facilitate an "instant transfer" to and from insurance, that would solve your problem, I think. The only problem is what if put your entire roll on the table and run it up big because you keep sucking out on your 2 outers? The software won't let you go south. So part of the table closing process needs to intercept the funds between their status as being "in play" and being "available." It would have to be integrated into the software. Heck, if I were pokerstars, I would write that in myself and cut out the middleman, as insurance is not a concept that can be patented. Ooops, sorry company spokesman. But don't worry, these sites probably aren't listening to me!
11-15-2012 , 11:50 PM
Quote:
Originally Posted by Mexican_Natis
Emre, I know you seem very proud of the fact that you have a whopping 5 SuperNova Elite customers, but let me try and break the numbers down for you real quick.

There are roughly 250 SNE players in the world, you have 5 of them, that means you have a monopolizing outrageous 2% share of the SNE market out there, not to mention there are about zero competitors in your industry. But of course your price is not too high, you have stated this, and the reasoning is because you and your partners are very smart people and know everything.
any SNE or anyone putting in SNE required volume on other sites and using this service has to be a complete nutcase.

well, maybe if since it's already mid november and they are late re-acquiring SNE and had to move up to get more VPPs i guess i could understand using IP but if any one of them uses IP for the whole year it would be laughable

so yeah, i'd be proud to have 5 SNEs if they stayed the whole year
11-16-2012 , 12:02 AM
Quote:
Originally Posted by Lessu
any SNE or anyone putting in SNE required volume on other sites and using this service has to be a complete nutcase.

well, maybe if since it's already mid november and they are late re-acquiring SNE and had to move up to get more VPPs i guess i could understand using IP but if any one of them uses IP for the whole year it would be laughable

so yeah, i'd be proud to have 5 SNEs if they stayed the whole year
What's wrong with players taking advantage of this opportunity to monetize variance and pay a comparatively tiny fee to reduce it substantially?
11-16-2012 , 12:44 AM
Quote:
Originally Posted by starrazz
What's wrong with players taking advantage of this opportunity to monetize variance and pay a comparatively tiny fee to reduce it substantially?
Absolutely nothing, that would be great.

The problem here is that it's not a tiny fee. Using the rough calculations from an earlier post in this thread, here are the cumulative profits of a small stakes break-even Supernova Elite cash player over five years.

Assuming a best/worst case scenario of +/- \$50k in EV over the five years (a very small chance at these stakes) and comparing each of those models with a +\$0 EV model, and an InsuredPlay model, it's clear to see that the costs aren't not tiny.

In defence of the InsuredPlay model, you would want to argue that you would gain value (increased winrate, play at higher stakes, etc.) but good luck justifying an increase in value that surpasses your \$24k yearly cost.

11-16-2012 , 12:48 AM
Quote:
Originally Posted by MeleaB
I'm unsure of the thinking behind the claim that using this insurance service allows you to play under-rolled. Suppose you wish to shot-take for a period of time and so choose to use IP to help you in case you end up running \$x below EV. The fact that you are equally likely to run \$x above EV, means that your IP account needs to be funded with money that would otherwise be part of your poker bankroll, so it doesn't really allow you to play under-rolled in the way it is claimed.

E.g. a 100NL player with 40 buy-ins (\$4k) decides he wants to use IP to shot take at 200NL for the weekend as he doesn't want to risk running below EV and his bankroll isn't big enough to play 200NL on his own. He's advised that he could easily experience a \$2k+ downswing in EV over that short time period so IP sounds ideal for him. The only problem is he can't sit down at 200NL with his \$4k roll, as he has to fund his IP account with half of it!
Im pretty sure it wasn't me who posted what you quoted with this response... I'm on the same page as you with regards to insurance...
11-16-2012 , 12:57 AM
Quote:
Originally Posted by K2AA72
Im pretty sure it wasn't me who posted what you quoted with this response... I'm on the same page as you with regards to insurance...
I apologise. It was coon74. (I think I must have selected quotes by you and him, and then trimmed the quoted posts down but accidently used your header instead of his.) It was also more of a general comment anyway, rather than a response specifically to his post.
11-16-2012 , 01:04 AM
Quote:
Originally Posted by starrazz
What's wrong with players taking advantage of this opportunity to monetize variance and pay a comparatively tiny fee to reduce it substantially?
because of the volume SNE requires.

maybe someone could count numbers how much an SNE would pay for this service at for example 1/2 or 2/4 6-max nlhe for the whole year and how it would affect them when they are for example breakeven, 2bb/100 or 5bb/100 winners (the lower your winrate the more swongs you have but i'd be curious to see how much it matters with SNE required volume)
11-16-2012 , 01:04 AM
oh someone posted a garph while I was typing
11-16-2012 , 01:30 AM
thanks for the graph, beats my wall o' text earlier.

Devils advocate here...what about the fact your graph shows each year's end point and not the journey...we've all seen those HEM graphs with EV and how mountainous and valleyfilled they are, surely this reduces that so that graph is only one line...notice how the EV line is always less swongy than the actual line? If I know my variance is reduced it allows me to do things riskier. So of course it costs money, question is it worth it. \$24,000 a year probably not. But I am not clear on how you get that big a figure from 0.25%. I assume all profits in your graph is a SNE who is good enough in the game just to beat the rake and that's it...its all SNE rewards right? And why not throw in a player who is closer to a standard deviation up or standard deviation down, that would be interesting to see instead of just that narrow range. But thanks for making the graph.

Im still for it because in exchange for giving up a tiny edge I can make correct AI decisions without having to worry over things I can't control such as which cards come on the turn and river--luck. My results skew closer to my true skill. My variance is reduced. You agree these are facts, right?

m