Poker: Rake and Bumhunting

I agree with the general point. If a site is upset about bumhunting, they should acknowledge that it’s their fault it’s happening. For anyone not poker – bumhunting is when good players only play against bad ones, rather than each other. Because poker sites rake each hand that reaches the flop, if you’re playing against equally skilled opponents, in the long term you all lose. If you’re playing against worse opponents, they can lose faster to you than the rake removes money from the game, so you can win. The lower the rake, the closer in skill you can be to your opponents and still win.

The unique selling point of poker over other popular online gambling games is that players can win. For a poker site to claw out a space in the market, it’s a vital point to understand. If the site starts trying to get rid of winning players, they end up competing with casino games (often from their own company), and if you continue along this path, casino games are much better suited to these remaining players than your poker site is. High rake erodes poker’s USP.

But there’s quite a lot in Callum’s article I disagree with, and I think most of my disagreements come from experience of running a site, so having access to information most players likely haven’t considered. Of course, this is no fault of Callum’s.

“nobody actually pays attention to the rake, right? Not according to Rob Yong, at least.”

@Callum_TH

I don’t know what Rob Yong said, but the general point is true – players are not sensitive to rake changes. When I was Head of Poker at Unibet, I was very eager to have the lowest rake possible at NL4, and we launched with 1% rake to 50c cap. If NL4 players were price sensitive, we would expect Unibet to steal all NL4 traffic from our competitors, and I can assure you that that did not happen.

We knew in advance that players wouldn’t flock to us for our NL4 rake though. We did it because we knew most new players test the waters in the lowest stake games. We wanted those new players to stay alive as long as possible.

If they were new to the site but experienced at poker, we wanted them to get used to the site, learn the unique features we had, and be comfortable playing their normal game there.

If they were new to poker, we wanted to keep them alive as long as possible, to let them learn that poker is a fun game, and to decide that they should make another deposit.

And if they were players who were used to NL4, we wanted them to move up to NL10. Increasing their win rate by reducing rake boosts their chances of doing so.

Of course, NL4 is not NL400. You would expect players in the latter to be more savvy. The site with the lowest rake at that game is hard to gauge for a player. I looked at a dozen for this article. Without any promotions, loyalty scheme, or affiliate deals, that site is PokerStars. The second is PartyPoker, of Rob Yong fame. If you include RIO’s Splash The Pot (a promotion that runs all the time regardless of the amount you play, so it’s fair to consider it), it’s RIO – about 10% lower than PS. But, as much as I like about what RIO do (and I’ve done some work with them, full disclosure), they are far from having the lion’s share of online NL400 play.

You may be wondering how I can say x is 10% lower than y. If one site rakes 4% to 100c at NL50 and another 5% to 75c, which is lower? You have too little information to tell, because you need to know the distribution of pot sizes from each hand in that game. You then have to calculate it over tens or hundreds of thousands of hands. No player has the dataset to do that for every game, at all times of the day – but the sites do.

The answer, by the way, is the former, by about 2% net, using the samples I have seen. Even that is going to change depending which site it’s on – when it’s that close, it comes down to the kinds of players on the site in question, because they determine the pot size distributions.

Poker rake is obscured from players in many situations, and even with the best intentions from the site, players cannot easily gauge price. As Callum points out, NLHE isn’t getting new patches every month, but even if it did, it would need a fundamental restructure to have a transparent pricing structure. Another industry that does this – though I think entirely intentionally – is consumer banking, with its opaque pricing structures. People very rarely move banks for price considerations, and online poker is the same, as shown by Unibet’s NL4 games and RIO’s NL400.

At one point, we spent a month with lower rake at PL25, because I was worried the games were too expensive – I think it was the game with the highest rake in bb/100 from the site’s perspective, so I reduced it by about a third. We did it without announcing anywhere, and not one single player noticed. There was nothing sent to support, nothing on 2 + 2 and nothing from any of my friends who played those games. I ended up taking the decision to keep it low (then we did announce it – and nobody said ‘oh, I saw, but didn’t want to tell you in case it wasn’t on purpose’) because my analysis said it was paying for itself. Players were not notably price sensitive here either – if they were, they would have noticed, like you would if your favourite takeaway halved its prices without fanfare.

As an aside, in an ideal world, I would be able to say “if we reduce rake by x%, liquidity raises by y%, which is worth €z per month (possibly minus €z). When we ran this experiment, I wasn’t able to make that calculation, but now I’m fairly sure I can. This is part of what’s so interesting to me for liquidity calculations. Being able to optimise rake for liquidity would be pretty useful for a poker site, hint-hint-I’m-a-consultant.

But anyway, you may remember that I agree with Callum’s general point, it’s just that I disagree with the diagnosis. Let’s go back to NL400. I made these graphs four years ago for posting on 2 + 2 while we were discussing this very topic. I worked at Unibet at the time, so my data are from their pot size distribution at each stake, over a million hands at each.

This shows the Unibet rake strategy: much lower at low stakes, but mildly higher at high stakes. It’s relative to PokerStars, pre- and post- a change they made in 2016. This isn’t the data that PokerStars will see, because they’ll use their own pot size distribution at those stakes, data I don’t have access to. I imagine it’s quite close though. Notice, by the way, that higher stake games are already the lowest raked games, and that holds true on every site bar Unibet.

The site will still be perfectly capable of operating a profit without rake traps, with hopefully more high-volume players moving swiftly into higher stakes, where they pay more rake. This has actually been used for some time by Unibet, at the micro levels, and they’ve more recently made changes to also lower the rake at 25 and 50NL to extend it. Last I checked, they were doing pretty well.

@Callum_TH

We had higher rake in the highest games we ran, but it was not uncommon for us to have more NL400 tables running than PokerStars did, and I don’t think we ever had fewer than Party, despite their lower rake.

But I want to go back to my main point of agreement with Callum:

Here’s the issue for sites complaining about bumhunters — it’s your fucking fault, not ours.

@Callum_TH

This is entirely right. But the reason those games ran more on Unibet than (say) Party wasn’t due to rake, it was due to product decisions. No table selection, no HUDs, alias changes, promotions aimed at keeping weaker players alive for longer, no external software, etc. This is what led those games to run more often than on any other site.

I’ll get back to that point in a minute, but it’s important to state that I don’t particularly want to defend higher rake in games. It’s a negative from an ecological perspective (and maybe some of my analysis in this post persuades you that it’s something I can actually measure, though I have no idea if Party et al can), and that’s why I’m not a fan. A lot of what we did at Unibet was for internal political reasons, and I think I’d prefer to just match PS at NL400 if I were doing it again today. When you have these internal conversations, if you say you are going to lower rake in the smallest stakes, it’s much easier to get everyone on your side by saying you’ll raise it in the highest. Even if you have the final say, you need other departments on your side, so you prioritise what you want to change and you have to be prepared to concede ground elsewhere.

It’s also important to state that you can’t reduce rake far. The reason that all subscription poker sites have failed is that they made too little money per player. If PokerStars have a poker lifetime value of (say) €300 but you have €200, PS can outbid you for new player acquisition. This is a big part of what liquidity actually means – that your players can play whatever games they want, whenever they want. That increases their lifetime value. Rake presumably follows some kind of laffer curve, where 100% rake leads to the site making no money, 0% likewise, and somewhere between there’s a sweet spot. If you reduce rake too far, you can no longer acquire new players (or buy food to eat), and you die anyway. Unlike applying the laffer curve to real world tax, there are rake experiments that you can do, and need I remind you at this point that I’m a gambling consultant?

Anyway, back to my actual point. Some people may think I’m saying we should trick people into playing NL400. If you had full anonymity, no table selection, and no tracking, aren’t you just fooling people into playing the games? Let’s put to one side that it’s kind of insulting to those players (of me, to put those words into your mouth). They can tell fairly quickly that the games suck. Fortunately, there’s a much better objection.

In a lot of these conversations, there’s an unspoken assumption that poker is a zero sum game. Either the best players get the money or the site does. But this is wrong, because the games are not static. Perhaps it can feel this way from a player’s perspective, but player acquisition and churn rates are a site’s two most important metrics.

A few years ago, I saw my boss deposit $500 on Party and sit at an empty NL400 table. He just played for fun and wasn’t a winner on the site. The table immediately filled – and I mean within two seconds – and as soon as he lost his buyin, every other player sat out. It took about 10 minutes. Now, can I blame those other players, with seating scripts and predatory instincts? Of course not! They’re acting in their own best interests. But can I blame my boss for wanting to never play on the site again?

This is, absolutely, the site’s own fucking fault. With no table selection, no fixed aliases, no way to see who is seated right now, and no external programs, my boss would have at least felt he was fairly losing his money, as he inevitably would. It would have taken a bit longer, but he would have felt a lot better about it. And the key point: he would have felt a lot better about making another deposit on another day. He has a better chance of becoming intrigued by the game.

This is why poker sites are not zero sum. He is an extreme example, but he would happily deposit a couple hundred euros and play casino games on his lunch break (he’s C-level in the gambling industry, it’s a pretty positive trait to have). A more normal example might be the person who sometimes buys a new video game to play that month, but sometimes deposits €50 and plays NL10 for their entertainment instead. These are the people we want coming in to poker, not Starburst or Big Bad Wolf (or Persona 4) or whatever. Lower rake has very little impact on them compared to product changes.

At Unibet, pre-relaunch in early 2014, a nightmare for the poker team was if one of the casino whales – some of these people are worth hundreds of thousands to the site – would sit at NL2k, lose 40k in a day, and never return to Unibet. Poker could effectively lose its monthly revenue from something like that, and even worse, it meant casino were wary of cross-selling its players into poker. This is for sure happening at other sites too, and product changes are the only real solution.

When we discuss this, we of course focus on the extremes, but that’s just because the points apply most strongly there. There are as many extreme losers as there are extreme winners, but most of a site’s ecology is made of slight winners and losers. Slight losers who just enjoy playing the game also benefit from softer games, and from people not being able to easily prey on them.

This is why Unibet succeeded with their relaunch. Not only did NL400 run more often than e.g. Party, who had lower rake, they were some of the softest games around. Players acted in their own best interests, they made more money, and the site made more money. This is the solution to bumhunting.

2 thoughts on “Poker: Rake and Bumhunting

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s