New 6-Handed Limit Holdem Bot at Harrah's Las Vegas

[The Shrink]

http://www.uspoker.com/blog/limit-ho...t-vegas/12818/

John Mehaffey wrote this article about a new 6-max limit hold'em machine at Harrah's in Vegas. Bot play is horrific but rake is outrageous (up to 25%). Thoughts on this from the limit players? In a 2+2 thread the author estimated that it would take about 150 hours grinding this thing to make Seven Star.

[Dan Druff]

http://www.uspoker.com/blog/limit-ho...t-vegas/12818/

John Mehaffey wrote this article about a new 6-max limit hold'em machine at Harrah's in Vegas. Bot play is horrific but rake is outrageous (up to 25%). Thoughts on this from the limit players? In a 2+2 thread the author estimated that it would take about 150 hours grinding this thing to make Seven Star.

Pretty good report by Mehaffey.

I agree with his conclusions that the game is likely unbeatable due to the awwwwwwwwwwful rake.

Even a 10% rake game tends to be unbeatable, though according to Mehaffey, the bot play is so bad (and is somewhat exploitable once you get used to it) that you can win far more than you would against human opponents.

Still, I can't imagine overcoming a 25% rake. As he mentioned, this 25% rake basically gives an opponent in a 4-way pot a free chance to suck out on you.

This still might be worth looking into just in case the rake can be overcome (probably not though), or that it might be the cheapest way to Diamond or Seven Stars.

Recall that the Texas Holdem Heads Up machine (which had no rake, but got increasingly tougher to beat as time went on) was initially the fastest/easiest way to Seven Stars. Now that machine is gone from all CET properties, and has been for awhile.

Mehaffey isn't totally clear about the tier credits earned. He wrote that you earn "$1 in comps or 50c in freeplay for $1000 wagered". This is an overly simplified (and a little bit inaccurate) way of saying 1 Reward Credit (RC) per $10 wagered, which each RC being worth 1 cent. You can trade in RCs for freeplay at varying rates, depending upon your tier credit (0.5 cents per RC at Gold, all the way up to 0.8 cents per RC at Seven Stars).

But he doesn't explain the tier credits earned. I am assuming it is 1 tier credit per $10 wagered, which would be the same as video poker. I'm assuming this because the RC accumulation he described is the same as video poker.

This was also the same accumulation initially offered on Texas Holdem Heads Up, which is why it was possible to earn Seven Stars so quickly at the time. This was NOT a bug or glitch, but they simply over-tiered the game because they must not have understood limit holdem action very well. Eventually they degraded it to one tier & one RC per $60 wagered, which was still worth doing. Then they got rid of the machines entirely.

The difference here is that you will be playing $2/$4 at the maximum, whereas you could play $20/$40 on Texas Holdem Heads Up.

However, due to apparently earning 1 tier per $10 wagered, Mehaffey estimates that you can earn 1000 tier credits per hour, which is about equivalent to what most players would earn from playing $5 per credit video poker.

The people on 2+2 saying that this would take 150 hours for Seven Stars are incorrect -- at least for the remainder of 2015.

This is because you will earn bonus tier credits for a certain amount played per day, which basically tripled your tier credits if done optimally.

If you play 2500 tier credits , you earn 5000 bonus tiers.

If you play 5000 tier credits, you earn 10000 bonus tiers -- and will make diamond in a day.

Both are very doable even at the less variance $1/$2 level (50c credit level).

So to get to Seven Stars optimally, you need to play 10 separate sessions (that is, 10 separate days) of exactly 5000 tier credits (or a tiny bit more if you can't be exact), and you will make Seven Stars.

Keep in mind that you can also play "time games" with the Total Rewards days. Most casinos restart their Total Rewards day at 6am (though some at 4am or 5am, so always check!)

This means you could accomplish THREE days of play in fewer than 48 hours. For example, let's say you live in Los Angeles and want to play one of these machines in Vegas.

Monday: Drive up to Vegas at 6pm, arrive 10pm. Start playing at 11pm for 5 hours at $2/$4, earn 5000 base tier credits + 10000 bonus, finish at around 4am. Go to sleep.

Tuesday: Wake up whenever you feel like it. Play 5 hours at $2/$4 again, earning 5000 base tiers and 10000 bonus tiers again. Go to sleep fairly early so you can wake up in the morning.

Wednesday: Get up at 8am. Play 5 hours at $2/$4 one more time, earning another 5000 base tiers and 10000 bonus tiers. It is now 1pm. Check out and drive home. You will be back by 6pm, 48 hours after you started!

With the above plan, you spent 2 full days in Vegas (including travel time from LA), but earned 45,000 tier credits -- meaning you're already 30% to Seven Stars.

So you could make two trips like these, and one a day longer, and you will make Seven Stars.

It's easier than the 2+2 nerds make it appear.

However, this may not be the machine to do it. There are still a few CET casinos left with 99.54% or better video poker with full tier credits being awarded, so the expected loss will be likely lower than this new machine, and the time required to play will be the same.

[Dan Druff]

I should also mention that if you are going to play against Texas Fold Up, you should do it soon. If it is beatable, eventually it will be tweaked to where it is not.

This happened with the Texas Holdem Heads Up game. For awhile it had some exploitable tendencies, such as the fact that you could triple-check-raise it fairly easily in certain circumstances. I came to realize that it would usually bet when checked to if holding top pair on a board like Q44, even if you check-raised it on the previous street. So if you have A4 and it has AQ, you can check-raise the flop, turn, and river and get paid off each time. They eliminated problems like this with subsequent updates, which I believe pooled actual player data (legal) and closed its own loopholes via an algorithm.

[Dan Druff]

Also, I have been told that this Texas Holdem Fold Up bot only exists at Harrah's Las Vegas for now, and is on a trial run.

I don't believe it's anywhere else in the country.

[Dan Druff]

Brandon (Drexel) went and played it yesterday, and reported back to me.

He found a few things:

1) He wasn't able to earn anywhere near 1000 tier credits per hour at the highest limits. He likely didn't play as quickly as John Mehaffey did (especially because he didn't realize until part way into the session that he could "end hands" early once he folded), but he only earned a little over 100 tier credits in an hour of play. So this may not be a good way to reach 2500 or 5000 tier credits in a day. I suppose it's also possible that Harrah's decreased the tier credit earning power since John played.

2) Brandon ran badly and lost money (nothing huge, but a lot compared to how long he played and the $2/$4 stakes). However, he theorizes that it is unbeatable due to both the rake and the ante.

3) He agreed with Mehaffey's statement that the bots play like the fish on Party Poker circa 2001.

4) He described it as a "card catching contest', as the bots do not fold with any semblance of a hand or draw, so therefore they cannot be bluffed. Therefore, the best hand always wins -- which is different than normal poker we are used to playing.

Since then, I have done a bit more thinking about the rake and whether the game is beatable.

I decided to do this in terms of the number of big blinds won per 100 hands -- a common metric to determine how good cash players are.

First, I approximated that at the $2/$4 limits, your average total bet per hand will be about $10. The minimum bet is $2 (the ante), and the maximum bet is $48 (capping every street), but I figured the average will be around $10. Whether I'm correct or not on that is immaterial, because it can't be too far off, and its accuracy isn't that important for what I want to show.

Since there are 6 players in the game (you and 5 bots), and since bluffing is virtually impossible, the hands always go to showdown, which means that the winner is always simply the player who catches the best cards. If all players were identical, this would mean each player would average 16.67% winning hands.

However, due to different standards regarding playing hands preflop, as well as different standards for folding postflop, the tighter player will win a smaller percentage of hands than that, but will also lose less on the hands he doesn't win.

It is fairly certain that a competent human player will be tighter than the bots.

So let's say that the human player wins 12% of the time, and one of the bots wins the other 88% of the time.

Again, this is just a guess, and the accuracy isn't important.

In a hypothetical session of 5000 hands (which is what would be required to earn 5000 tier credits, assuming a $10 average bet per hand), this would mean the player would win 600 of them and lose 4400 of them.

Let's say that the pot size in those 600 hands averaged $50. That would mean $30,000 worth of won pots, before rake.

Now of course we have to look at the 4400 hands that the player lost. These are not raked, but the player loses a minimum of $2 each. Let's say that the player averages $6 loss per hand, making a total loss of $26,400.

So before the rake is considered, the player won $3600 ($30,000 - $26,400) in 5000 hands of $2/$4. That breaks out to a profit of $0.72 per hand -- which is 0.36 big blinds (which are $2 each) per hand. That means the player won 36 big blinds per 100 hands played, before rake.

BUT WAIT!

The estimated 20% rake on $30,000 comes out to $6000! So the player actually walks away having LOST $2400 after the rake is taken out!

This is absolutely insane, given that 36 big blinds per 100 is an EXCELLENT win rate in limit holdem. To give you a comparison, when I have an extended period of time where I run well in limit holdem on Bovada, I average about 7.5 big blinds per 100 hands! This is nearly 5 times my best medium-term win rate on Bovada, and yet still I would LOSE if I achieved that rate! Ouch!!

So in our hypothetical above, the player would have to win $33,000 in those 600 hands before rake -- meaning a $6600 profit after losses are subtracted, in order to break EVEN after the $6600 rake is collected. This would be $1.32 per hand overall, and a staggering 66 big blinds won per 100 -- just to walk away even.

No matter how bad the opponents, I don't believe it is anywhere near possible to win 66 blinds per 100 (or anywhere near it) in limit holdem in anything but the super-short term.

Therefore, I declare this machine to be wholly unbeatable.

Even if my calculations are off and you simply need to win 30 big blinds per 100, that again is not possible to do for any extended period of time.

So what is the game's hold?

Well, even taking my first example (a good player beating it for 36 blinds per 100 before rake), that still equates to $2400 lost per $50,000 wagered after the rake. That's a hold of 4.8%, which is far more than most video poker machines in Vegas.

And if the player instead only manages to beat it for 10 big blinds per 100 (which may be more realistic), he would lose $4480 using the figures above. That would be a hold of almost 9%, which puts it on par with slot machines.

In other words, this looks like a sucker's game from my early calculations.

The rake is too damn high.

Also, developing a strategy to do better is tough. If you play tighter, you are costing yourself money because of the $2 ante per hand (there's no such thing as "folding a free hand"). If you play looser, you are going to have issues with having your weaker hands rarely hold up against the ever-chasing bots. That is, don't expect middle pair or ace high to win at showdown too often (whereas these hands are frequent winners at normal 6-max games, where people chase much less). Looser play also means that you will get raked more, as you will be entering more hands.

Therefore, playing tight has a downside, and playing loose has a downside. This will make it tougher to develop a play style which could negate the huge rake advantage that the house has in this game.

Conclusion: This machine looks like a waste of time and money.

[BeerAndPoker]

So you're saying I'd be better off degening at blackjack?

[Forum Wars]

Brandon (Drexel) went and played it yesterday, and reported back to me.

He found a few things:

1) He wasn't able to earn anywhere near 1000 tier credits per hour at the highest limits. He likely didn't play as quickly as John Mehaffey did (especially because he didn't realize until part way into the session that he could "end hands" early once he folded), but he only earned a little over 100 tier credits in an hour of play. So this may not be a good way to reach 2500 or 5000 tier credits in a day. I suppose it's also possible that Harrah's decreased the tier credit earning power since John played.

2) Brandon ran badly and lost money (nothing huge, but a lot compared to how long he played and the $2/$4 stakes). However, he theorizes that it is unbeatable due to both the rake and the ante.

3) He agreed with Mehaffey's statement that the bots play like the fish on Party Poker circa 2001.

4) He described it as a "card catching contest', as the bots do not fold with any semblance of a hand or draw, so therefore they cannot be bluffed. Therefore, the best hand always wins -- which is different than normal poker we are used to playing.

Since then, I have done a bit more thinking about the rake and whether the game is beatable.

I decided to do this in terms of the number of big blinds won per 100 hands -- a common metric to determine how good cash players are.

First, I approximated that at the $2/$4 limits, your average total bet per hand will be about $10. The minimum bet is $2 (the ante), and the maximum bet is $48 (capping every street), but I figured the average will be around $10. Whether I'm correct or not on that is immaterial, because it can't be too far off, and its accuracy isn't that important for what I want to show.

Since there are 6 players in the game (you and 5 bots), and since bluffing is virtually impossible, the hands always go to showdown, which means that the winner is always simply the player who catches the best cards. If all players were identical, this would mean each player would average 16.67% winning hands.

However, due to different standards regarding playing hands preflop, as well as different standards for folding postflop, the tighter player will win a smaller percentage of hands than that, but will also lose less on the hands he doesn't win.

It is fairly certain that a competent human player will be tighter than the bots.

So let's say that the human player wins 12% of the time, and one of the bots wins the other 88% of the time.

Again, this is just a guess, and the accuracy isn't important.

In a hypothetical session of 5000 hands (which is what would be required to earn 5000 tier credits, assuming a $10 average bet per hand), this would mean the player would win 600 of them and lose 4400 of them.

Let's say that the pot size in those 600 hands averaged $50. That would mean $30,000 worth of won pots, before rake.

Now of course we have to look at the 4400 hands that the player lost. These are not raked, but the player loses a minimum of $2 each. Let's say that the player averages $6 loss per hand, making a total loss of $26,400.

So before the rake is considered, the player won $3600 ($30,000 - $26,400) in 5000 hands of $2/$4. That breaks out to a profit of $0.72 per hand -- which is 0.36 big blinds (which are $2 each) per hand. That means the player won 36 big blinds per 100 hands played, before rake.

BUT WAIT!

The estimated 20% rake on $30,000 comes out to $6000! So the player actually walks away having LOST $2400 after the rake is taken out!

This is absolutely insane, given that 36 big blinds per 100 is an EXCELLENT win rate in limit holdem. To give you a comparison, when I have an extended period of time where I run well in limit holdem on Bovada, I average about 7.5 big blinds per 100 hands! This is nearly 5 times my best medium-term win rate on Bovada, and yet still I would LOSE if I achieved that rate! Ouch!!

So in our hypothetical above, the player would have to win $33,000 in those 600 hands before rake -- meaning a $6600 profit after losses are subtracted, in order to break EVEN after the $6600 rake is collected. This would be $1.32 per hand overall, and a staggering 66 big blinds won per 100 -- just to walk away even.

No matter how bad the opponents, I don't believe it is anywhere near possible to win 66 blinds per 100 (or anywhere near it) in limit holdem in anything but the super-short term.

Therefore, I declare this machine to be wholly unbeatable.

Even if my calculations are off and you simply need to win 30 big blinds per 100, that again is not possible to do for any extended period of time.

So what is the game's hold?

Well, even taking my first example (a good player beating it for 36 blinds per 100 before rake), that still equates to $2400 lost per $50,000 wagered after the rake. That's a hold of 4.8%, which is far more than most video poker machines in Vegas.

And if the player instead only manages to beat it for 10 big blinds per 100 (which may be more realistic), he would lose $4480 using the figures above. That would be a hold of almost 9%, which puts it on par with slot machines.

In other words, this looks like a sucker's game from my early calculations.

The rake is too damn high.

Also, developing a strategy to do better is tough. If you play tighter, you are costing yourself money because of the $2 ante per hand (there's no such thing as "folding a free hand"). If you play looser, you are going to have issues with having your weaker hands rarely hold up against the ever-chasing bots. That is, don't expect middle pair or ace high to win at showdown too often (whereas these hands are frequent winners at normal 6-max games, where people chase much less). Looser play also means that you will get raked more, as you will be entering more hands.

Therefore, playing tight has a downside, and playing loose has a downside. This will make it tougher to develop a play style which could negate the huge rake advantage that the house has in this game.

Conclusion: This machine looks like a waste of time and money.

This is probably pretty close to reality...here is the public IGT .pdf brochure about the game:

http://www.igt.com/~/media/05ed32177...50e362e05.ashx

You'll notice that they give the payback for the machine at 96.5-99.91%



I know that John Mehaffey, in the original USPoker.com article said this:

Bonus bet

There is an optional bonus bet on the machine that may be placed before the hand is dealt. It is irrelevant to the poker action.

The payouts are listed on the screenshot to the right based on two credits. The bonus bet pay table can vary with a range of 96.50% to 99.91%, according to marketing material distributed by IGT. The casino determines which pay table is used.

...but I think there's a chance that this is the payback to the casino for the main game (not the side bet), as the brochure seems to be talking about the main game.

Druff would be quite close with his analysis if the 96.50-99.91% payback listed is correct, since these would probably be the payback rates for PERFECT play against the machine. It would be pretty difficult to know how to play perfectly, so most players would probably be in the 90-95% payback range.

[nerakil]

Also, I have been told that this Texas Holdem Fold Up bot only exists at Harrah's Las Vegas for now, and is on a trial run.

I don't believe it's anywhere else in the country.

Drexel also mentioned the "maximum field test period" on the 1/6/2015 show, which, per Nevada gaming regulation 14.080, is a period not exceeding 180 days.

Regardless, I believe both of you are incorrect. The installation of this machine at Harrah's LV was... the summer of 2014 (and it was first unveiled at G2E in October 2012).

I was vacationing in Las Vegas in 2014, and visiting the casinos up and down the strip. At Harrah's LV, I stopped in to play $10 pai gow tiles by the high limit slots section. They just finished installing a bunch of new IGT machines nearby (I tried them all), including Texas Hold'em Fold Up, Stack 'em Poker, Multi-Win Draw Poker, etc. I believe they also put a Shoot to Win craps machine in that section (did not play). Obviously, the trial period has expired, so gaming must have determined Texas Hold'em Fold Up fits their minimum standards.

Just a minor correction, Mr. Witteles.

[Anyone else find that listening to the podcast on 1¹/₂ speed (iPhone) is more ideal?]

[Dan Druff]

Also, I have been told that this Texas Holdem Fold Up bot only exists at Harrah's Las Vegas for now, and is on a trial run.

I don't believe it's anywhere else in the country.

Drexel also mentioned the "maximum field test period" on the 1/6/2015 show, which, per Nevada gaming regulation 14.080, is a period not exceeding 180 days.

Regardless, I believe both of you are incorrect. The installation of this machine at Harrah's LV was... the summer of 2014 (and it was first unveiled at G2E in October 2012).

I was vacationing in Las Vegas in 2014, and visiting the casinos up and down the strip. At Harrah's LV, I stopped in to play $10 pai gow tiles by the high limit slots section. They just finished installing a bunch of new IGT machines nearby (I tried them all), including Texas Hold'em Fold Up, Stack 'em Poker, Multi-Win Draw Poker, etc. I believe they also put a Shoot to Win craps machine in that section (did not play). Obviously, the trial period has expired, so gaming must have determined Texas Hold'em Fold Up fits their minimum standards.

Just a minor correction, Mr. Witteles.

[Anyone else find that listening to the podcast on 1¹/₂ speed (iPhone) is more ideal?]

Wow!

I had no idea this thing has been here for 1 1/2 years. I guess its location at Harrah's and only one machine made it take a long time for a limit hold player (Mehaffey) to discover it.