Any thoughts on this?

Is it beneficial?

What win rates would translate at which levels in NL vs people?

Do the bots react to you, will they respond to different play styles?

Do the different bots maintain the same play styles on each profile or is it randomized each session?

I'm only a few hundred hands in but winning 1.06 bb per hand (which is obviously a heater) but I don't really feel half as challenged as if I were playing even NL2

latest session https://imgur.com/a/T0lIAG0

https://imgur.com/a/bssBNVs

Alfie bots are quite odd

The blinds nearly ever defend, even in spots where they really should with almost atc, like when mp min raises, btn calls, sb calls and bb folds. $1 to see a flop with a $8 pot...

Also, if Alfie 3bets you and hits top pair, they are going to get it all in by the river, making it very easy to bust with hit draws, overpairs and sets

https://www.americascardroom.eu/poke...be-undefeated/

"several players including Serbian poker pro Srdjan Pavlovic Nislija have tested their skills against the app. So far not a single one has been able to win against PokerAlfie over a game sample of 5,000 hands in the game’s most challenging “Optimal” setting."

I need to do a headzup4rollz match against Nislija apparently

"July 31, 2017“I played against PokerAlfie 2.0 over 5000 hands and PokerAlfie was better for about 0.05 big blinds per hand.” - Srdjan Pavlovic Nislija. Nislija's worldwide highest ranking at PocketFives is 394."

4. PokerAlfie does not analyze opponents decision-history and does not adapt his play to opponents play style

http://pokeralfie.com/poker_ai.htm

Maybe i'm just such a fish that alfie doesn't quite know how to exploit my horrible tendencies?

https://imgur.com/a/81leYKJ

Hero MP A3o raise $3
fold to sb
SB 3bet to $9
Hero defends because pokeralfie is a fish and we have position

Pot = $19

248hh flop check, check

2 turn, check check

5 river sb bets $6, we raise to $30, sb calls with A high

how exactly is this bot undefeated?

Looks like you both played it bad

oh yeah, I'm playing far from optimally, that was by no means intended to highlight good play on my end

Anyways not surprised... marketing you know?

yeah that's very apparent by now but given that it seems to be an academic side project it's surprising they'd be so absurd with the "nobody has ever beaten it" thing... I actually came into thinking it'd be an interesting challenge, something more like what crushed the top regs

i think main problem is it's too nitty, doesn't defend enough and then if it sees a flop will gii with just top pair or even A high, it's highly exploitable - i want to know more about the "pros" who lost to it to bumhunt them more than anything else

just want to finish grinding the 5k and post it for the lols that a 2+2 nobody beat it with ease

So how'd ya do buddy? You've surely played 5000 hands by now.

lol i totally forgot about this and just kind of got bored with it, it just kind of got monotonous once i realized marketing was all a sham and i'm just clicking buttons

guess i'll grind out the extra thousand hands or so i'd need to qualify as having beat it - thanks for reminding me

laptop with alfie was out of commission for a while as it needed a new motherboard and i forgot about it for a while

today grinded a few hundred hands, nearly 4k hands in and a 10bb/100 win so now just need to close it out

This is a great read.

haha yeah i know, just got suckered in by their "unbeatable" claim so wanted to beat it

today i grinded out those final few hundo while watching succession, could have folded to victory like our lord and savior phil galfond but played it out and won a few more buyins, needless to say, if i can beat alfie for 14/100 then it's objectively terrible at poker - definitely accidentally folded AA and other premiums a few times trying to get through the j2o hands via fast fold and sticky mousepad double clicking it

alfie is way too tight and the entire 5k sample only went 3 handed to a flop, would find alfie sometimes deviate heavily from the nit profile and 3bet you with something like K8 but then alfie bots would usually check fold a lot of flops, alfie didn't bluff enough and would often check down value hands so it felt like i could have played atc so long as i had position because i usually had some options to check down and improve or steal it

definitetly felt like nl2 much harder than alfie

final results

emailed them my victory, no response, looks like i was a tree falling in a forest

i found it in the windows store of all places

but it can also be found here http://pokeralfie.com/

Thank You for playing PokerAlfie,

Sorry, for not answering your E-Mail, it was by accident.

When I made PokerAlfie in 2017 the only reference for quality of playing was
1. Mr. C. L. (He did not want his name to be published)
- He was the 31st in a Live-Tournament and collected €624
- He won against PokerAlfie with 0.0047 Big-Blinds per hand (we can say almost Tie) after playing 5000 hands
2. Srdjan Pavlovic Nislija
- Nislija's worldwide highest ranking at PocketFives is 394.
- Nislija played against PokerAlfie over 5000 hands and PokerAlfie was better for about 0.05 Big-Blinds per hand

In 2017 my poker skill was very bad, there was no chance for me to win against PokerAlfie.
Of course, writing an algorithm to play Poker and Poker-Skill are two very different things.

I thought, these two guys are competitive players and PokerAlfie plays decent.

So, I challenged the good players if they can win against PokerAlfie

In the mean time I know, that a serious player can beat PokerAlfie.

I still find PokerAlfie a very useful tool for improving poker skills or just for playing Poker.
Many users are very glad for using PokerAlfie.
Of course, people that reached such a descent level as you, see no use of PokerAlfie.

Best Regards,
Alex – Developer of PokerAlfie

I mean “Of course, people that reached such a DECENT level as you, see no use of PokerAlfie.”

There is a new version of PokerAlfie with improved decision logic and some new features

Hi Alex,

I was away for a bit, thanks for responding.

A few things about your "pros"

The credentials you gave are not credentials. If you and I put both put our grandmothers into some poker tournaments they could leverage some tournament finishes. 31st place in a tournament is a wholly meaningless result. There's a massive amount of variance involved in poker, especially tournaments due to the structure where people make sub-optimal decisions can have good performances over a short sample size. There have been finalists and even winners of the World Series of Poker, surviving a tournament of thousands of people, including many of the best in the world who are long term losing players.

Likewise, pocket 5s rankings only indicates gross results not wins and losses. Essentially, the people who lose the most money in the game and are who the ecosystem is built around will always rank very highly on that site. It's also where you need to opt into the program in order to get ranked. So it's neither a complete list nor an accurate one. For example, I never opted in there because there's really no benefit at all unless you want to deceive people who don't understand how their rankings actually work.

Now he's blocked trackers from seeing his history, so we don't have much to work with.

But it's pretty clear he's a recreational player and possibly has lost money overall.

His play history is all over the place, he's playing $109 tournaments and $1 ones at the same time. This screams casual losing poker player and is a big red flag.



Again, these are only the results where he cashed. We have no idea what he played in where he lost.

He opted out of the more objective trackers, which is also what a lot of people did too. I too opted out simply because I didn't want opponents to analyze my stats.

But the info we do have there seems to confirm casual player.


See this he's a losing player who got managed to win one big tournament for 42k. Take that one result away and everything is a loss. If we had a time machine and had him not win that for 42k but rather still crush it and finish in 4th place fo 8k then we have a losing player. Take away that entirely and we have a bad losing player.


I have no doubt he understands poker pretty well. It's also possible he's somewhat talented at it. But if I were to guess, I'd say he's not a very good player at all and would welcome him at my table. But even is he was a winning player, it sure doesn't look like he was thriving and would have made more money working at Walmart during this time. That number on pocket fives doesn't include losses, so even if profitable, his net is going to be a significantly smaller number - which broken down over several years is not much income at all. I understand that not much income would be required in Serbia and if he was able to play for a living then good for him. I don't mean to attack him too much, he could be a very nice guy. But he's not representative of an elite player by any objective measurement whatsoever.

And lastly, he played tournaments. Which is an entirely different kind of poker than cash games, which is what alfie is and tournament players are notoriously worse at cash games as they play a format that's more based on the structure of the tournament (IE a lot of your decisions are influence not by optimal play etc but by how many players and chips are remaining) whereas cash games are just pure unfiltered poker.

Again, I'm not trying to be mean. Your anonymous friend and Srdjan are the losing players the economy needs in order to keep people like me fed and clothed. But your marketing doesn't claim that Alfie beat some guys who've played a lot of tournament poker. Instead you claim it beats world class pros, which is simply not true. I don't think you intentionally tried to deceive people, I also don't think that these other two people tried to deceive you. They probably genuinely believe themselves to be good at poker.

I am personally small potatoes in poker. If I doxxed myself and you googled me you'd find far less info than what is available on Srdjan. I haven't played professionally for well over a decade and I mostly played live, which is a significantly easier game. People don't see me here and think "Oh rick he's so good at poker" they instead know me for things like microwaving steak.

If you wanted to work with genuine pros to help test the engine and improve it, I'm sure you could start a thread here and find some very good players. In the meantime, I highly suggest you change the language with which you market this software into something more truthful. Just say that it beat some highly knowledgeable poker players. At the very least you should remove the "unbeatable over a 5k sample" because I've clearly demonstrated this is no longer the case.

I will wipe the data and check out the update though

GLGL and feel free to reach out at any time. I'm in software myself, which is a major reason why I took so much interest in this in the first place because it's where two of my passions intersect.

Hi Rick,

Thank you for your detailed message.

I’ve always claimed: PokerAlfie won against these two players - not against other (perhaps better) players. Anyway, N. has won more than a million $ in tournament prize money – I thought that is significant.

After these 2 Matches in 2017, I was challenging players to see if someone could beat PokerAlfie (since PokerAlfie was unbeatable back then) - But not anymore.

My apologies if PokerAlfie website mislead you (and perhaps some others) into thinking that I still claim it's unbeatable. PokerAlfie is 100% free and with no ads, so it is not a big deal - if somebody mistakenly tried it.

With your assumption, that N. is not a winning player, I really advice everyone to play against PokerAlfie first and only after winning, to try to play for real money.

The new version has some incremental improvements, it plays better, but not two class better. I hope to release a new version with much better decision logic later this year (but who knows, if my expectations materialize).

PokerAlfie webpage:
https://www.pokeralfie.com

Looking forward to checking out the new version, big fan of projects like this.

As far as areas of improvement for the old version.

The AI is way too nitty when other players enter the pot. Nearly every pot was heads up. Very few went 3 ways to flop and I genuinely don't think we ever had more than 3 people to a flop and one of them would be sure to fold out right away.

There are just so many real world situations where AhKh, JJ, & 99 all find themselves seeing a flop together. Then on the flop it is Jh9x3h and now all three players going to want to continue to see a turn. These situations literally never happened. So I often wondered if it was just a fluke sample or the logic of the software made really strong hands fold out if it were multi way.

Sometimes, after an action, PokerAlfie has a habit of setting the hand range a little smaller, which then leads to playing tight afterwards.

In my opinion you need to randomize this setting. Otherwise it’s too exploitable once the human notices and they will notice it soon after realizing they never go multi way and when they they immediately fold to the smallest bet unless holding the nuts.