While No-Limit Texas Hold’em is still the king of tournament poker and the lower stakes, it has begun to find itself a bit played out in the world of high-stakes cash games in recent years. It’s a very well-studied game due to its popularity, and that means that it’s increasingly hard to find an edge at the high levels of play.

At the same time, the most common mixed games – things like Razz, 2-7 Triple Draw and so forth – are also pretty well-understood, and also simpler overall than Hold’em. As a result, top cash players are always on the lookout for something new, or else inventing it themselves.

One of the latest innovations is something called Duck Flush. Nate Meyvis brought it to my attention yesterday while I was doing an interview with him and Andrew Brokos for an upcoming instalment of his ThinkingPoker podcast. He didn’t tell me what it was called, but this morning, I saw it mentioned as part of a mixed game lineup and figured out with some help from Google that it was the same game we’d been talking about.

You know you’re at the bleeding edge of poker when Badacy and Baducy are already losing their luster. Personally, I’m still waiting for a chance to play those in the first place.


Conceptually, Duck Flush is pretty simple, albeit bizarre. The one-sentence description is that it’s a five-card triple draw game which plays high provided the best hand is a flush or better, and according to 2-7 lowball rules otherwise. If that makes sense to you, feel free to skip ahead, but for those who need a more detailed explanation:

  • Players are dealt five down cards to begin with. The players to the left of the dealer post small and big blinds. There may also be antes.
  • There are four rounds of betting – two with small bets, two with big bets – with three drawing rounds in between.
  • During the drawing rounds, players may exchange any number of cards from their hand or stand pat.
  • At showdown, a player’s hand must be a flush or better in order to qualify for high.
  • The best qualifying high hand wins the entire pot. If there is no qualifying high hand, the best low hand wins the entire pot instead.
  • Straights count against low hands, so 7-5-4-3-2 is the nuts when there is no qualifying high. (Technically, flushes would count against low as well, but that’s irrelevant in this game, as if you have a flush, you have a qualifying high and your low doesn’t matter anyway.)


During our interview, Brokos asked me for my initial reactions to the game, which I had never heard of before. I guessed that the pot would often be won by a high hand, because in a straight highball game with three draws, big hands are pretty common. As it turns out, however, this is incorrect; the article I found on the game (which is in French) quotes Allen Kessler as saying that most pots are in fact won by a low hand. Now that I’ve had a chance to play around with the game a bit, I can see why this is the case.

It’s a funny game because it plays a bit like a split pot variant even though the pot is never actually split. Whereas in Hi/Lo variants, you want to be drawing to both ends of the pot at once, in Duck Flush what you want – assuming you don’t already have a qualifying high, of course – is to be drawing such that if you hit, you will have a qualifying high, while if you miss, you will at least have a decent low.

This means that you rarely want to be playing two pair and trying to draw to a full house, for instance, except when the pot odds are very good or you have some other plan for winning the pot. Your odds of filling up are less than 25%, and if you miss, you are more or less dead for low. Much better is to be drawing to a small flush, because you can easily miss your flush yet still win the pot with your low, provided that everyone else misses their draws as well.

On the other hand, trips – small trips in particular – are an interesting starting hand. You’re closer to 30% to make a full house or better by drawing two the whole way, which is a lot better than your odds with two pair, and there’s a more subtle advantage as well: if you manage to make quads on the first or second draw (about an 8.5% chance), you can continue to draw one while holding an almost unbeatable made hand, which gives you excellent concealment and makes it very likely that you’ll be paid off. Finally, if your trips are sevens or lower, you have one further edge in that you’re heavily blocking your opponents’ chances of making a strong low, which in turn makes it more likely that you’ll pick up the pot at some point without a showdown.

It’s when you consider these aspects of starting hand selection that you can see why flushes and better won’t actually come up as much in practice as they do in a highball game. Single pairs will almost always be broken and two pair hands folded, while getting trips in one’s starting hand is fairly rare, so you won’t see full houses coming up all that often. On the other hand, keeping three low flush cards and drawing two is pretty viable, something you would never see in a highball game. Finally, if a player does make a very strong low at some point, they are pretty likely to decide to keep it rather than continuing to try drawing to their high.

Once you’ve recognized these facts, you can start to see how this becomes a crazy and agonizing game. Let’s say you start with 732xx with those low cards all being spades. You draw two, of course, but get the Queen of Spades and Four of Diamonds. Now what? If you pitch the Queen, you’ve got a fairly high probability of making a good low on the next two draws, but your reverse implied odds are awful if anyone else manages to make a qualifying high. On the other hand, if you discard the Four, you’re only 36.4% likely to make the flush, but can be a lot more confident of winning when you hit (and can fold much more readily when you miss). So, what do you do?

Obviously, that decision is going to depend heavily on how many people are involved in the hand, and to a lesser extent on the current pot size. It also depends on your expectations of your opponents, however. In a multi-way pot, if you think everyone else is prioritizing their low hand, you probably want to gamble on the high. Conversely, if you think they’re all likely to be going for low-probability high draws, you might want to stick with your low and hope they all miss.

You’re also going to get a lot of interesting situations coming up at the end of the hand, because generally speaking, the better high draws make for worse lows when they miss and vice versa. For instance, in a heads-up situation where both players are drawing to flushes, the player with the higher cards is likely to have the better flush if both hit, but also likely to have the worst low when both miss. This can make it pretty tricky to figure out when it’s worth raising with a draw and when it’s better just to call.

Finally, snowing (pat bluffing) seems like it would be a riskier and more complicated proposition in this game than in Badugi or 2-7 Triple Draw. It’s very hard to be dealt a pat flush or better (less than 0.4%), so a player standing pat from the start is essentially representing a strong low. If another player has four cards to a flush, therefore, they’re likely to call, since they’re approximately even money to hit on three draws. That means that the bluffing player will likely need to fire every street, and then there’s a danger that the drawing player backs into a reasonable low and feels they need to call off on the end because the pot is so big by that point. Meanwhile, typical snowing hands for 2-7 Triple Draw will often have some high potential in this game; if you’re dealt 4477K in Triple Draw, for instance, you’ve got a good hand to try snowing with. Here, though, you might want to take one or two shots at the full house before starting to bluff; it gives you two possible ways of winning the pot, for one thing, but also, a player who draws one and then starts patting is way more likely to have a qualifying high than someone who was patting from the start, so it makes your bluff more credible as well.

Taking all this into account, you can see why the game is so popular with the cash game high-rollers. There are a lot of non-obvious decisions to be made and no publicly-available information on when, for instance, it’s correct to draw to a flush instead of a low, or how you should play a hand like 2277x. The guys who are playing the game, then, are doing their own analysis away from the table and keeping their mouths shut about it. Whoever is first to discover a killer strategy is therefore going to hold an edge like you’d never see in a well-understood game like Hold’em. Of course, once such a strategy has been used enough times, others will pick up on it… but that just means it’s time to invent the next, even crazier game.

Alex Weldon (@benefactumgames) is a freelance writer, game designer and semipro poker player from Montreal, Quebec, Canada.