Quote:
Originally Posted by 7OAD
I am curious if the guy who did the math can tell me if this is better or worse in terms of now having more hands in the sample while running worse.
If he also wants to answer this question, that'd be great:
What is the number for 99% or 99.9% in terms of under expected EV? I am genuinely curious as to how much worse someone in the 1% or .1% would run over 30k hands.
As I mentioned in a later post I shouldn't have really done the calculation as there isn't the right data present to answer your question: "How Bad Am I Running? (PT4 Graph)".
First of all I think you should probably try to avoid thinking about this Ev adjusted gap as it is in the past.
There are reasons to monitor data but to get any good from them you have to understand the data and it's source.
Looking at this graph is very unlikely to help your poker development at all, if you are interested in mathematics or statistics well trying to understand this can be an interesting challenge. For poker the only use is perhaps to get a slightly better guess at your ability, ie. if you are at 0% roi but have an unlucky run from allins maybe you are slightly positive in the long run compared to break-even.
To work out how well you are doing in AllinEvs you should really filter for these hands only and find out the std deviation of these only. Using this std dev and mean is a fairly simple way of checking what variation or spread a future run of X hands may bring. (Again it is pretty useless info though)
What I showed was how likely a future run of all hands would spread from your 17bb/100. We don't know your true bb/100 well enough yet and this spread involves all types of variation, bad run outs, poor pocket cards, unlucky domination etc.
The difference shown of the two lines on your graph only involves the Ev adjusted part and so this difference is only produced by a small part of the data producing it.
When I did I just guessed at the std deviation from my SnG results and assumed yours would be a bit higher as MTTs have double or more starting stacks.
Here are a couple of equivalent graphs of me playing 9865 games of non-rebuy turbo 180s SnGs. The first is all the hands unfiltered, the second is just the allin on Turn or earlier hands.
This has a bb/100=5.45, an EV bb/100=5.66, the sample std dev according to Hem is sd(bb/100)=52.16 or sd(per hand)=5.22
This is filtered for allins pre river and has a bb/100=164.1, an EV bb/100=167.0, the sample std dev according to Hem is sd(bb/100)=159.34 or sd(per hand)=15.93
As you can see the second graph is more informative for Allins and if I used the data I could work out how probable my future X hands of allin would spread. But why bother? It is of no real use to me. I am currently running about 1000bb under the adjusted but so what - it doesn't matter in the long run.
By eyeballing this 2nd graph you can see that from about the 3500 allin hands to the 5500 hand I ran like god and quickly built up a 1000 bb amount of luck but in the next 1000 hands I lost it all.
If I play 100k more games I'll end with a graph with a gap of perhaps a few thousand bbs but percentage wise by then it won't matter as I am so much closer to the long run. You can see that at times my allin ev is above the standard and some times it falls below, yours will too if you play enough games.
You can also see that the first graph on paper is more variable, more bouncy. I doubt if my true bb/100 changes much during this and so it indicates that throughout the whole game the other forms of luck like spells of being dominated etc exceed the effect of the allin ev. This is even for turbo Sngs and allins are so very common in them.
Many players do get obsessed on these EV differences but the best players just take what comes and continue to try to improve over the other competition. If you are using you db software use it to find leaks, this grpah is mostly just a form results oriented thinking.