While Phil is probably the second best player in this HU match, he is running bad to the tune of something like half a million in AIEV. Pretty important thing to mention.
Do you have a source or a graph that shows this? I only ask because the graph Phil posted early on when he was only down 350k showed his AIEV was -500k after 3938 hands of the match. He was actually running good in all in ev while still getting crushed. I haven’t seen any newer graphs or mention of the AIEV for the match but maybe I missed it.
Not sure what “aiev” is but saw golfond play K8ss from the big blind which seemed pretty donkish.
Playing only 2 cards in Omaha would be pretty donkish.
4 Likes
All In Expected Value. In other words, the amount of money we expect him to win based on the straight odds of winning after all cards are known because everyone is all in.
Isn’t AIEV measured against expected outcome? So -$500k is actually running very bad?
Here is the graph I was referring to:
https://twitter.com/PhilGalfond/status/1222579058128482304
It shows that he was running about 150k above ev in all in pots at the time. Of course he could haven been running bad in other ways such as being coolered too often.
1 Like
If AIEV = -$500k and you’re only down $350k then you ran $150k above expectation.
holy fucking shit that graph is incredible
Is that AIEV calculation just showing the hot and cold preflop EV for each hand, and not those that actually go all in? If so isn’t it just saying that he has been getting worse cards than his opponent?
The way I’ve always understood it is it’s the sum of non showdown winnings + the sum of EV for each hand that went all in, regardless of who won the hand.
Where is @ChrisV - I think he would know for sure.
Hello!
AIEV is only calculated on hands where someone is allin. It’s calculated from the equity on the street on which someone ends up allin. It’s accurate at what it tries to measure but it’s not the whole story in terms of measuring luck. The graph line adjusts to what the outcome “should” have been, so that graph shows that Galfond should have lost an additional 150K on fair allin runouts.
It’s likely that Galfond is running bad in other ways based on the fact that he’s losing in both showdown and non-showdown pots, that either means runbad (getting dealt worse hands/coolered/getting bad pre-allin runouts) or it means that Galfond is straight up terrible at poker. The latter is unlikely to be true. Whether this runbad is enough to account for the losses or whether Galfond is just losing is hard to say, I have zero experience with HUPLO, but that graph should definitely alarm him.
I don’t understand what Galfond says in the tweet, he seems to think that graph shows him getting unlucky in AIEV, but it shows the opposite. Someone else pointed out the same in the comments to the tweet.
Exactly my understanding as well. Thanks.
Has anyone ever started one of these challenges and it go well?
I was going to issue my challenge next week, but i better rethink. 
I doubt he’s making such a silly error. He’s not addressing the fact he’s run above expectation at all, he’s just whinging about the expectation itself. He reckons getting it in behind so often you have an AIEV of -500k after just 4k hands is pretty unlikely even if you’re a long run 15bb/100 dog.
1 Like
There’s definitely a logic error on his part there somewhere.
I dunno. A 15bb/100 dog would be expected to lose an average of about 120k over that many hands. If Galfond had run as expected he’d be down a total of around 850k 500k (thanks Chris).
Oh, right, I get it. He’s saying that -500 (the AIEV-adjusted actual result) is far beyond the loss expected even if he sucks at poker enough to be a 15bb/100 loser. Although you’re wrong in the second post. You don’t add the lines like that, it’s just -500 total.
Yeah I just misremembered the Galfond tweet that started this whole discussion. He’s probably right that a lot of time he’s at the bottom of his range and vini is at the top of his or whatever,- but I’m not sure how healthy that mindset is here.