Cross post from MTT

[Marm]

Posted this under MTT, but havent gotten any replies yet, figured Prob is also a good place to try.

This is a concept I have been playing witht for a little while. I like it, some people don't, maybe I'm just crazy. I figured If I want to get flamed for a bad idea, 2+2 is the place to go.

Take the inverse of the product of your ROI% and ITM%. (ROI x ITM)^-1. Whats that give you?

I'm thinking its the minimum numbers of buyins you should have before playing in an event, as long as both numbers are positive. At the very least it gives an index of your performance. The closer to 1, the better you are. Of course this requires a decent sample size to be accurate.

Thoughts? Not too harsh now please..

[Siegmund]

Well, lessee.
Your ROI is basically a weighted average of your placings (in a sit-n-go it would be 5*P(win)+3*P(place)+2*P(show)-1, in units of buyins; change the -1 to -1.1 or -1.15 to account for entry cost.)

Instead of saying you will win various prizes various percentages of the time, you are essentially treating a tourney as you play 1 unit to play every time and win (ROI+1)/ITM units with probability ITM.

A series of n "oversimplified tourneys" (not an oversimplification at all if you are playing something like a TEC tourney on UB, where all the placers get the same prize!) is just a series of coinflips, expectation ROI and variance (ROI+1)^2*(1-ITM)/ITM each.

Risk of ruin and expected time until showing a profit are both proportional to Var/EV^2 = (1-ITM)^2*(1+ROI)^2 / [ITM*ROI]^2 for your scenario.

This will (quite badly) underestimate your variance in an actual series of tournaments with graded payouts.

[Marm]

Thats about what I figured I'd get.