ITH Poker Forum

[toronexti]

If you want to do an MTT I would suggest something like making a lot of different outcomes and then weighting them.

So let's say you look at 180's and you're completely average.

90% of the time you win 0 buyins
5% of the time you win 1.98 BI
.5% of the time you win 2.805 BI
.5% of the time you win 4.29 BI
.5% of the time you win 5.775 BI
.5% of the time you win 8.25 BI
.5% of the time you win 10.725 BI
.5% of the time you win 13.2 BI
.5% of the time you win 19.635 BI
.5% of the time you win 33 BI
.5% of the time you win 49.5 BI

(This is actually missing .005% cause I rounded somewhere).

Now this is actually a losing player (about -16.5% ROI) due to rake.

If you have someone who is cashing say 15% of the time I would do it something like this:

85% no cash
7.5% you win 1.98 BI
0.527% cashing in all other positions.

This would be a really simple way of adjusting for the extra cashes (and probably yield a player who's like -7.6% or so ROI).

Another way would be to scale it meaning that you're cashing more in the higher positions than the bottom.

So with 15% cash rate something like this:
85% no cash
7.0% you win 1.98 BI
.7% of the time you win 2.805 BI
.75% of the time you win 4.29 BI
.8% of the time you win 5.775 BI
.85% of the time you win 8.25 BI
.9% of the time you win 10.725 BI
.95% of the time you win 13.2 BI
1% of the time you win 19.635 BI
1.1% of the time you win 33 BI
1.175% of the time you win 49.5 BI

This is~ 60% ROI. But you get the idea. Then after you run like 1 billion trials you can see the type of swings.

So I guess just from this exercise (without the graphs) you can see that you have to do MUCH better than "average" to get a decent ROI. After 100 $4 MTTs you can expect scenario 1 to be down $72.60. In Scenario 2 to be down $33.44. In Scenario 3 to be up $264.

[janeg]

deleted

[toronexti]

Willem, isn't the data skewed by having Hero lose the pot instead of what he bet during the hand?

There is no betting going on in Willem's sims.

What happens is that the simulator picks a pot size from the range (2bb to 8bb), the simulator then flips a coin. If the coin lands Heads hero wins the pot along with an additions .02bb (so if pot is 2bb hero actually wins 2.02bb). If the coin lands Tails; villian win the pot minus .02bb (so villian wins 1.98bb if hero loses).

[janeg]

Willem, isn't the data skewed by having Hero lose the pot instead of what he bet during the hand?

There is no betting going on in Willem's sims.

What happens is that the simulator picks a pot size from the range (2bb to 8bb), the simulator then flips a coin. If the coin lands Heads hero wins the pot along with an additions .02bb (so if pot is 2bb hero actually wins 2.02bb). If the coin lands Tails; villian win the pot minus .02bb (so villian wins 1.98bb if hero loses).

Thanks toro ... yeah, realized that after I posted but no delete post button in this forum

[Willem]

So with 15% cash rate something like this:
85% no cash
7.0% you win 1.98 BI
.7% of the time you win 2.805 BI
.75% of the time you win 4.29 BI
.8% of the time you win 5.775 BI
.85% of the time you win 8.25 BI
.9% of the time you win 10.725 BI
.95% of the time you win 13.2 BI
1% of the time you win 19.635 BI
1.1% of the time you win 33 BI
1.175% of the time you win 49.5 BI

This is~ 60% ROI. But you get the idea. Then after you run like 1 billion trials you can see the type of swings.

So I guess just from this exercise (without the graphs) you can see that you have to do MUCH better than "average" to get a decent ROI. After 100 $4 MTTs you can expect scenario 1 to be down $72.60. In Scenario 2 to be down $33.44. In Scenario 3 to be up $264.

When you don't cash, you actually lose 1 buy-in. We need to have a similar payoff structure like this, only with the net result being completely zero. Some thing like this:

90% of the time you lose 1 BI
10% of the time you win 10 BI

Then we can multiply these payoffs with a factor that gives a different ROI. For instance, you'll get a 120% ROI when you multiply everything with 1.2.

Also, I don't think it's realistic that you have a bigger chance to win 49.5 BI compared to winning 2.805 BI.

[toronexti]

So with 15% cash rate something like this:
85% no cash
7.0% you win 1.98 BI
.7% of the time you win 2.805 BI
.75% of the time you win 4.29 BI
.8% of the time you win 5.775 BI
.85% of the time you win 8.25 BI
.9% of the time you win 10.725 BI
.95% of the time you win 13.2 BI
1% of the time you win 19.635 BI
1.1% of the time you win 33 BI
1.175% of the time you win 49.5 BI

This is~ 60% ROI. But you get the idea. Then after you run like 1 billion trials you can see the type of swings.

So I guess just from this exercise (without the graphs) you can see that you have to do MUCH better than "average" to get a decent ROI. After 100 $4 MTTs you can expect scenario 1 to be down $72.60. In Scenario 2 to be down $33.44. In Scenario 3 to be up $264.

When you don't cash, you actually lose 1 buy-in. We need to have a similar payoff structure like this, only with the net result being completely zero. Some thing like this:

90% of the time you lose 1 BI
10% of the time you win 10 BI

Then we can multiply these payoffs with a factor that gives a different ROI. For instance, you'll get a 120% ROI when you multiply everything with 1.2.

Also, I don't think it's realistic that you have a bigger chance to win 49.5 BI compared to winning 2.805 BI.

Well you'd be wrong, look up the OPRs of the top players and you'll see that they come in top 3 a higher % than they should.

[Willem]

Well you'd be wrong, look up the OPRs of the top players and you'll see that they come in top 3 a higher % than they should.

Hmmm, I didn't expect that. But I never play MTT's so I'll take your word for it. Besides, it doesn't matter much anyway,

Can you make the numbers such that the ROI is exactly zero? With that, I can program it somewhere in the next few days and produce some nice looking graphs.

[toronexti]

Well you'd be wrong, look up the OPRs of the top players and you'll see that they come in top 3 a higher % than they should.

Hmmm, I didn't expect that. But I never play MTT's so I'll take your word for it. Besides, it doesn't matter much anyway,

Can you make the numbers such that the ROI is exactly zero? With that, I can program it somewhere in the next few days and produce some nice looking graphs.

Does it matter if u load the % so that the missing ROI is all made up from 1 position?

So something like:
85.66% no cash
6.84% 1.98 BI
.5% 2.805 BI
3.5% 4.29 BI
.5% 5.775 BI
.5% 8.25 BI
.5% 10.725 BI
.5% 13.2 BI
.5% 19.635 BI
.5% 33 BI
.5% 49.5 BI

This is an ROI of about .0032% (if I did the math right) I think that's close enough to zero.