Introduction:

A fairly common situation that I experience on-line is getting donkbet into. Is he trying to steal the pot from me, trying to find out where he's at, or trying to build a pot? Does the donkbet amount provide any clues?

To help figure this out, I wrote a program that parsed all the hand histories that I have, which are all from 25NL and 50NL.

Any flop or turn donkbets (DB) that went to showdown were analyzed. I'm not sure how many hands that covers, but I do have history on 34,000+ players.

I did not set a maximum bet amount for a bet to be considered a donkbet. Any flop bet before the pre-flop raiser acted, and any turn bet before the flop bettor/raiser acted qualified as a donkbet. The data is for hands when there were two and only two players that saw the flop. Stop-n-go's (bet/call flop, bet turn) are included in the data.


Data:

Total DB: 1078
Total DB saw showdowns: 397
Total DB won money at showdowns: 177
Total DB folded: 218
Total others folded: 463

so 37% of the time the donkbettor went to showdown, 20% of the time the donkbettor folded at some point in the hand, and 43% of the time the other player folded. It's impossible to know how often the donkbettor would have folded had the other player raised, but the data seems to show the donkbettor will go to showdown about twice as often as he'll fold.

Approximately 45% of the time the donkbettor won money at showdown. I would be interested to hear what others think a good number is for this, but I'm guessing winning players are close to or above 50%.

Straight flushes: 0
Four of a kinds: 1
Full houses: 27
Flushes: 23
Straights: 18
Three of a kinds: 26
Two pairs: 110
One pairs: 131
High cards: 25

These are the hands that the donkbettor had at showdown. One thing to keep in mind is that these are not the hands that the donkbettor had at the time of the actual donkbet (flop or turn). But it still shows some interesting things.

57% of the min-raisers had 2-pair or better hand.
93% had 1 pair or better.

It's probably safe to say that slightly more than half of the time the donkbettor has one pair or less or a draw.


Flop versus Turn Donkbets:

Are there differences between flop and turn donkbets? Of course!

DB by street: flop DB = 760, turn DB = 318.
DB folded: flop DB folds = 185, turn DB folds = 33.
DB saw showdown: flop DB saw showdown = 257, turn DB saw showdown = 140.
DB won $ at showdown: flop DB = 108, turn DB = 69.

One key thing here is flop donkbettors folded 24% of the time, while turn donkbettors folded only 10% of the time. So you have little fold equity against a turn donkbettor, but great implied odds.

Also note that W$atSD was 42% for flop donkbettor versus 49% for turn donkbettor. So turn donkbettors usually have a better hand.

Lastly, the data doesn't show this here, but the turn stop-n-go W$atSD is less than both the flop and turn W$atSD. However, there is not enough data (40 hands) to state this as a trend with any confidence.


Min versus Normal-sized Donkbets:

I had been assuming that min (<2BB) donkbets are a lot weaker than normal-sized donkbets, and had been treating them basically as a check. This was somewhat verified by Mike Kelleys post Flop min-bet Theorem. His theorem was "A minbet (1bb or 2bb) on the flop by OOP villain after Hero is the preflop aggressor = weak hand, straight or flush draw."

To verify those assumptions I broke out min-donkbets as well.

Min DB by street: flop DB = 155, turn DB = 56.
Min DB saw showdown: flop DB saw showdown = 59, turn DB saw showdown = 28.
Min DB won $ at showdown: flop DB = 26, turn DB = 13.

The W$atSD is 45%, which is higher than I expected. I assumed it would be lower than for normal-sized donkbets. One thing that may be an issue is the small sample size (87 hands). I looked through the W$atSD hands, and the vast majority of the time the donkbettor had only 1 pair. There were a few draws and a few 2-pair hands.

Running this with a larger sample size would be benefitial, but it appears that most min-donkbets are in fact done with weak hands and draws.


Summary:

1. Over half of the time the donkbettor has one pair or less, or a draw. Your top pair/overpair hand is probably still good until told otherwise (donkbettor 3-bets).

2. A larger donkbet generally means a bigger hand.

3. A turn donkbet generally means a bigger hand than a flop donkbet.

4. You have little fold equity against a turn donkbet, which means higher implied odds.

5. A turn stop-n-go doesn't necessarily imply a big hand trying to get all in.

Some of these follow common assumptions, I think it's good to see the data shows that they are true. Let me know if you can think of some other data that would be interesting, I will try to fulfill any reasonable requests.

Nice work.

The min donkbet gets a standard sized CBet over the top from me.... everytime. I haven't done the analysis, but I suspect this gets folded at least 70% of the time.
Min-donk says : "Please don't forget to CBet me, sir"

The more properly sized donkbet is the more interesting. I currently use it myself with a TP type hand on the flop. Or, and probably the most exasperating time, when I have an underpair on the flop. I have 99 and the flop is K73r. Flop has only hit a select few of villain's range hands and will likely fold out JJ,TT,AQ and the like.

Thx for sharing. Quality post ! But i think you have to use more HHs.

Your posts are gold sir... very nice.

This is one of the things that I'm really unsure of how to deal with and this post will help me alot in dealing with donkbets.

THANKS!

This is the same for me. I feel like they are just making a bigger pot for me to take with my CBet when they do this. After the success I've had playing it this way, I'm convinced that this is a very profitable scenario that is surprisingly common at micro stakes.

I think one of the best ways to play a flopped set against a GOOD player is to donk bet if you are oop. I've found that often someone with an overpair will will not raise a PSB donk bet on the flop. I've also found from experience that 3 barrelling is achievable in this situation and they will call down each street. Of course, this is villain dependent but even some super aggro villains will not raise but call each street.

I don't really know how to play it right when someone dbets into me. If you use realtime HUD it provides dbet stats, which I have found helpful, although I rarely use any of those programs anyways.

Nice analysis. In my experience, a min-donk bet is almost always a draw.

Donking sets is situation dependent, but I think it is important that you don't narrow your range by only donking sets. It's important to be careful that you don't announce your range to a good player.

+1, I like to donk bet good opponnents on boards that I believe missed his range, like 9xxr, or on sets, or semibluffs, or two pair that missed his range.

LOL...nice....

One of the few times I myself like to donk bet is when I am in one of the blinds and it gets folded around to the button who open raises. In those rare spots where I actually choose to defend my blind HU against the button, I will often donk bet the flop. Typically in this spot, the button raiser is just trying to steal the blinds and doesn't want to get into a contested pot with what is most likely a mediocre hand at best. On the flop (especially when it comes all junk) I'll lead out with a 2/3 pot donk bet. This seems to be fairly affective.