The reason I point this out is that this information was available a long time ago, and it's interesting how, even till today, so few people are aware of it and how it can impact tournaments. That's why this discussion thread exists and it's also why late registration exists.

Also, it's probably fair to say that those people who are running tournaments who allow late registration, especially if it's very late, have a lot to learn.

Does this mean that all the insults you have sent my way in the past, and this includes long discussions on your podcast, were just jokes?

This paragraph shows that your understanding of mathematical modeling is minimal at best. However, this is nothing to be ashamed of, very few people have a background in statistical theory and, in general, when statistical theory and mathematical modeling come up, most people are just confused.

Let me try to explain it to you and everyone else who reads this, and I'll start with a mathematical model that virtually everyone is familiar with.

Many years ago a man named Copernicus said that the planets orbit the sun in circles. Notice that planets going around the sun is a simple mathematical model that answered a lot of questions relative to how the planets moved in the night sky. And this is exactly what you want from a mathematical model -- simplicity with many questions being answered.

Today, we know that the planets actually go around the sun in complex elliptical orbits. But this doesn't detract from the brilliance of the Copernicus model.

Notice that ICM does something similar. The idea behind it is relatively simple, yet it answers many questions related to poker tournaments. This includes estimating the value of a chip stack as well as giving some guidelines as to how the holders of the different size stacks should play their hands -- see my book Gambling Theory and Other Topics for more information on this second point.

Also, notice that ICM is not perfect. But it has stood the test of time.

Here is part of your fundamental misunderstanding of mathematical modeling. ICM is not part of the equation. It is an equation, and this equation gives us much guidance relative to tournaments. Also, while it does not account for differences in skill between different players, it does give guidance towards what skills are needed relative to the size of the chip stacks and to a certain degree how these skills should be applied.

Okay. But again, to demonstrate the power of ICM, you don't have to be near the bubble. This is a serious misunderstanding on your part.

MM

Agree with this point. If 100 players reg at start time, and there are 99 remaining when you register, you steal (a tiny bit) of EV from each of the 99 players in the tournament.

Hi ravager:

This is correct. At the beginning of the tournament, the ICM effect is minimal. But the later in the tournament, the more powerful ICM becomes. This is why ICM (in percentage payback tournaments) points towards some strange plays late in a tournament.

Also, another way to look at late registration is as follows. First, suppose you're in a 100 player tournament and everyone is a playing clone of each other. Then your chance of finishing in any position is 1 percent.

Now suppose when you enter this tournament there are 79 players left, it started with 99 players, you are the 100th player (meaning there are now 80 players left), and again everyone is a playing clone of each other. Notice that your chance, according to ICM, of finishing first is still 1 percent since you have 1 percent of the chips. But your chance of finishing in each of positions 2 - 80 is now higher than 1 percent, and that's because you can't finish in positions 81-100 since 20 players are already knocked out.

Best wishes,
Mason

It certainly true that very late registration increases the EV for the average player. This is even true for very skilled players….recall the story of Chris Ferguson routinely letting people move ahead of him in the registation line at several WSOP tournaments after registration had closed just so he could enter a few minutes later.

But in my experience, the worst aspect of extreme late registration is the way that it affects the play of the actual tournament.

If a lot of people are registering late, the table are going to be filled with short stacks at a time when blinds have increased way past the starting levels. These new players are going to just have to play push/fold poker, and as a result all the other players who have been playing the whole time and may have built up deep stacks have to just defend against push/fold poker, and this destroys any potential for post-flop play in the middle stages of the tournament.

I used to go the WSOP every year and play 4-6 events in the $1500 range. For the first several years, I though the structure of these events was pretty good through the tournament (or at least throughout the first day and a half). But the last few NLHE events I played (in 2018 and 2019), whenever you made it back from the dinner break (with like 50-70% of the field already gone), you’d suddenly have four new pros join your table with short stacks, and any strategy beyond preflop charts was out the window until everyone either doubled up or busted.

This pretty much soured me toward playing any NLHE events at the WSOP in the future unless they fix the late reg problem. (Though I will still play limit & mixed game events, which have the same ICM issues with late reg, but where the strategy doesn’t significantly change with more shorter stacks at the table.)

The problem with Mason Malmuth in threads like these is that he usually turns this into a posturing exercise, where he wants to demonstrate how much more mathematical knowledge he has than anyone else.

Therefore, you can't even have a discussion with him (even on the general same side of the argument) unless you 100% agree with everything he types.

Let's say I enter a 300-man poker tournament, and the other 299 opponents are approximately 50% huge fish and 50% good players, without any skill levels in between. One could argue that entering on time is an advantage over registering somewhat late, because otherwise you'll be missing out on a lot of the chips of the huge fish, who are basically giving them away.

In the above example, this is where ICM can't be used to explain everything regarding late registration. As I said before, ICM does not account for skill levels, and assumes all players to be average compared to the field.

There is a school of thought that entering at any point when a player has busted is advantageous over entering before anyone busts, due to ICM.

However, as I stated above, that doesn't tell the whole story, since there are varying skill levels, and it's true that bad players will bust (or lose most of their chips) early more often compared to the better players.

This is why my "bubble" example is so powerful. It demonstrates that there is some point in late registration where, no matter how good the remaining field (and no matter how bad the initial field), it would be hugely +EV to register at that point. Therefore, if the bubble is super-advantageous to register, and if one-off-the-bubble is similarly super-advantageous, then obviously this demonstrates that very late reg can be a huge edge.

This isn't the only way to explain the concept, but it's one which is easy for the average poker player to understand, and one which shuts down all debate on the matter.

I realize that Mason can try to present a more convoluted way that he proved mathematically in 1986, but that's not necessary here, nor will it convince the biggest doubters.

I definitely understand both sides of it. ACR I late register for quite a bit and have noticed the advantages and my success level going up quite a bit compared to when I use to never late register.

On the other side, the biggest tourney I ever personally played was a $2,500NL tourney at the M resort in Vegas back when it had poker like 7 years ago. At the beginning of the tourney there were around 250 people registered and I would say around 100 to150 were people who won their seats at all the giveaways by the M resort owned casinos around the U.S. My opening table had 3 people I can tell where regs and rest were seat winners. Spent the entire day just exploiting, building my 30k stack up to about 80k by days end. It felt so easy. Halfway through day one though I saw a long line of people late registering and by end of late registration there were a total of about 600 players. Day 2 I was put on a table where it was obvious every single player at my table was a reg. Day 2 I couldn't couldn't find any good edges and ended up busting like 20 players before the money.

Again, ICM is a mathematical model that gives certain information including, under the assumptions of the model, the later you register in a tournament the better off you are. If you want to adjust your opinion based on other information available, that’s fine. Statisticians call this Bayesian Statistics and there are some examples of it in my Gambling Theory book. Also, I wouldn’t adjust it much. When statisticians might error, they try to error on the conservative side.

MM

No, they would pay (250/200)*100=$125. The amount can only go up.

A simple late fee per hour addresses the situation somewhat, but not as accurately as the sliding amount based on the ratio of entrants to "remnants". There would be more registrants closer to the end of the hour, and the fee could be woefully low or unfairly high depending on the number of players remaining. The sliding scale is trivially easy to do online and the software used in live tournaments could be modified fairly easily (The live tournament software I've seen keeps record of players remaining so the tournament directors are already inputting that information).

That entry amount assumes 200 players remain. The original number of players is irrelevant. You divide the number of total entrants by the number remaining.

Here is the bottom line(s):

1.Great players usually increase their EV by entering early. This is especially true in big bet games where live ones often go broke early. In limit games it would be less true because only a small percentage of stacks are at risk. It could also be untrue if fatigue is a factor.

2. Above average players usually lose EV if they enter early rather than late. But entering early does increase their chances of winning the tournament. The exceptions would be if the fatigue factor exists or if the reason they are merely good rather than great is because they are weak at very big stack poker.

3. Almost all winning players, even the great ones, will decrease their hourly rate in almost all tournaments if they enter early.

4 Whether your hourly rate should be the deciding factor usually depends on what you would be doing with the extra time available if you don't enter early. If you could be playing in a profitable side game or entering a different tournament late, you probably shouldn't enter this one early. However if you are a great playe,who doesn't easily tire, who will be twiddling his thumbs if he doesn't enter now, he shouldn't wait. Other players should wait if EV is their goal unless they are placing extra importance on their chances of coming in first or if they are just trying to have fun.

Good summary. One thing I would add is late regging, even though it might be higher theoretical hourly under the right circumstances, will increase swings substantially. To what extent this is an issue depends on a case by case basis. One thing is for sure though, playing tournaments for a living with 10-20% ROI is not a fun time.

Why would it increase swings? Your results will less likely be zero or top three than if you entered early. (Your standard deviation per DAY would go up if late entering meant you played more tournaments, but not per tournament.)

Who came up with ICM first, the complicated math behind who benefits most from entering late, and ICM is all fun to talk about, but would a soft field not increase one's EV more than registering late or early? If we're talking about whether the best players have an increased edge based on when they enter, would that not be heavily determined by the field?

In a high roller where you are the worst player, whether you enter early or late, your EV would still be negative, right? Which goes back to my point of how do we get new players in the game? Confusing them with 8 hour late reg, more chips/less rake if you enter early, and showing them the math behind their EV depending on when they register, doesn't seem to solve the problem.

New players aren't entering the game. They're playing fortnite and other games on twitch. People would rather watch a hot Asian girl play chess, than even try to play online poker in today's environment. If "late reg can guarantee an increase in equity of x%, xx%", then how much can a softer field with novice players learning the game increase one's equity?

If late reg deters new players from entering the game, would this not have a negative effect on the perceived increase in EV of when you register? Tougher fields means lower EV for each player, does it not? I appreciate all the words from you David Sklansky, but one thing people would always say about you in the card rooms I've played in over the years, is you are a master at game selection. If they see you in a game, it's a good game. If that's true, is this not because tougher games have a lower EV for the individual players?

Main point is that late reg seems to be deterring people from wanting to play poker tournaments. Not just any people, but the people we need and want to register. The fish. The new players. The ones that keep the ecosystem going.

Chances of winning the tourney in this scenario would be <1% as you'd be more likely to enter the bubble period and subsequent pay jumps with a short stack, meaning you have to play more conservative than normal. Meanwhile the big stacks can take advantage of this so their x% of chips in play actually translates to a >x% chance of winning the tourney.

I think the upshot is that in an equal field, late regging gives more of a chance of cashing but less of a chance of a big score, resulting in a neutral ev situation overall.

What I'd like to know is how late regging affects ev in a satellite, assuming equal ability in the field. My hunch is that it might be +ev, as the big stacks will have some chips in their stack that are effectively redundant, meaning we buy into a 100 player field with >1% of the effective chips in play. Would be interested to know if anyone has any data on this.

Its not a neutral ev situation. Not if you know what you are doing, The changes in strategy are only used if they add more than they subtract.

I'd contend that correct ICM strategy is needed to bring it up to neutral. Anything else would be -ev.

Eleven players are left. Nine are deep. Two have tiny stacks well below the initial buy in. Ten players make the money. Tenth place pays two buy ins. If you buy in as the twelfth player your EV is hugely positive even if you did nothing to adjust to the prize situation.

Someone gets it.. You need to give fun players/recs the flexibility to play the way/when they want! Sadly this might benefit some regs who latereg, but so be it, because, in the end, it helps the sites to generate bigger guarantees and more fun players, because poker tournaments take a long time and not everyone wants to sit on a computer playing one tournament for six to ten hours and win $50.. People want to play few hours and win a lot of money. That's where the world is going. Entertainment is fast and quick.

Casino streamers(twitch) have 10x bigger viewership than the poker players. It's quick, potential big win, and dumb. Attracts a lot of fun watchers. Poker needs this sort of population too.

I am afraid, if we would take out the latereg it would take out a lot of the fun player pool too or some people would just lose interest in poker, like I said, their time is limited. They are just doing it as a hobby.

As a professional poker player, my attitude is: fun players rule the poker world. We need to make the poker world better for them. Not for regs.

If we took out late reg it might initially take out the fun player pool, but like all things, people will adapt. They'll see less regs having 8 hours to fire unlimited bullets, and eventually see people having more fun playing the game. If anyone has seen the BenCB and Charlie Carrel youtube video where they analyze their hands, Charlie Carrel keeps saying, "Mate, in a freeze out, you can't do things like flat 64s on the button."

Eventually people will realize that the game is completely different during Late reg and Unlimited re Entry times. You're playing a game where people are doing things they won't do in a Freeze out. Bluff frequencies of players firing unlimited bullets completely changes the game. Tournaments have been hijacked solely for the purpose to give Regs a higher edge, and casinos more money for less space/personnel.

Also with all the ICM talk and things like that, do these guys even play the tournaments? It doesn't look like it. I'd assume it's because they probably don't have an edge. The fields are much tougher with less recreational players. If the thread is about how much an increase in equity is by late registering, remember we're talking about tournaments where some people have a minimal edge like 1%. So late regging they increase their edge to 3%. That's a 200% increase. OMG. But that's also lying with statistics. Nobody cares about an actual increase in equity from 1% to 3% unless you're putting in a ton of volume.

Instead, why don't we get new players into the game, stop listening to people who don't even play the tournaments, mathematically figure out how they hypothetically could, but never will because the fields are too tough. A softer field would solve all problems. But why would anyone tell anyone to play poker in today's environment? A tournament on ACR can take 12 hours. They've basically time conditioned every single person who has a job, doesn't have a full day to play, and isn't a pro, to not be able to play. Or, they've deceived them and got them to enter without them realizing they stand no chance. And in sales they say this: One happy customer tells 5 people. One angry customer, TELLS EVERYONE. It's only so long before everyone is saying F*** tournament poker.

I watched David Baker stream while playing on ACR a couple months ago, and he won the $55 PKO tournament. Multiple times on the stream he kept saying, "This takes so long, I don't think I'll play this again". If the winner of a mother ******* tournament is telling people even he probably wouldn't play it again because it takes so long, what are we really doing? Word of mouth is essential in any business.

And one time I also watched PokerGuru Rob Kuhn streaming , and he said that he talked to the owner of ACR and he got him to make the structures better, or deeper. He has also stated that he has more of an edge in deeper structures than almost everyone, from experience and studying. I respect that, but why are the owners catering to the people and making tournaments structured to give the people with the highest edge, even more of an edge?

People say they want a slower structure, but again, Henry Ford quote, they also said they wanted faster horses. People may not know what they want, but one thing I'm sure they do want, is time. Their Time. Time to do anything else. Time to spend their winnings. Time to not be forced sitting in front of a poker table/monitor.

Late reg for almost any major LIVE tourneys are fine but should really be cut off no less that about 20bb's

Taking chips off your new stack or paying a fee is the stupidest damn thing ever. Players will hate your tourney an not come back.

I do not see a single advantage live and I dont care what your stats say. Some people actually want to play longer if they love the event and want to be able to re-enter themselves fairly late.


As for ONLINE they for sure leave it open too long. Stats are much more reliable to show a possible advantage here as they are easier to database and there are tourneys that dont stop until 30mins before the money.

Yes, as an amateur, I can only schedule time for MTTs a certain time of the year especially for ones that are multiple days. And the ones with the best structure are multiple days. Being able to re-enter as a late reg means I get to play a lot more poker than I would otherwise. That way I get more value for any time I might have scheduled off work.

Coincidentally or not all of my biggest live tournament scores came as a late reg re-entry so I'm definitely in favor of continuing it as an option. I agree that for live tournaments, the ICM advantage that it gives simply is not that big a deal and has a very small impact compared to a million other edges that exist in live poker.

As far as short stacks entering the field, facing a table with a wide variety of chips stacks is simply a part of tournament poker. There will be short stacks all throughout the tournament anyway, so some injected in late reg doesn't matter.

It would be interesting if a winning player played a couple thousand Tourneys on Pokerstars late registering at the last second to see what their ROI looks like.

Great post. Sites should consider fun players the most. Regs and their edge can wait.

Doesn't the question of "Why doesn't everybody do it?" beg the question "What if everybody does?"

In this hypotehtical tournament with late reg up to one before the bubble, I know I am gaining edge so I register at the last moment. So do 100 other people. Now we're 101 from the bubble. I bust out, it gets back to one before the bubble. I know I am gaining edge so I register again. So do 100 other people. And so on. Everyone is gaining edge, but the tournament never ends. We would be in Dante's 6th circle of poker hell. I'm never getting paid, so it seems my edge is now exactly -100%. Where did it go?