The simplest way to understand what a semi-bluff is is to think of it as a bluff with a backup plan.  It is a bluff,  insofar as you would like your opponent to think you are stronger than you really are  --  and fold as a result of your bet.  But if it doesn’t succeed as a bluff,  it still has a way to win  --  by having some potential to get better on the next card  (or in the case of a pre-flop semi-bluff,  on the next cards).

Here’s an example of a semi-bluff and an explanation of how it would work.  You have just seen the Turn.  You are in late position holding the As-Kh.  You raise preflop after one player called the blinds.  You,  the caller,  the Big Blind and the Small Blind saw the flop of Qd Ts 2s.  All three of them checked to you.  You checked as well.  The Turn was the 8s.  Again,  everyone checked to you.  You bet the pot.  That bet was a semi-bluff.

Here’s the explanation of that bet.  On the one hand, you’d be very happy winning the pot right there.  You have some reason to believe you might. Everyone checked to you,  showing weakness.  For all you know,  they might fold to your pot-sized bet,  fearing that you made a Flush or even some lesser hand that was still stronger than their hand.  But even if they don’t fold and you get called,  your hand may improve on the next card.  If,  for example,  the River is a Jack or any spade then you’ll have either the Nut Straight or the Nut Flush.  Of course if your Flush card pairs the board you could give someone a Full House,  but that’s unlikely since no one bet their hand.  The bottom line is that you have two ways to win:  either by inducing everyone to fold when you bet without a made hand or,  if they called that bet,  by hitting your hand on the River.

A semi-bluff cannot be made on the River.  On the River,  of course,  there are no more cards to come.  Hence you can’t improve on the next card.  On the River,  a bluff is just a bluff.  It must succeed or fail on its own with no chance of improvement.

Here are some Things To Consider when deciding whether or not to semi-bluff:

First of all,  is there really some reasonable chance that your opponents will fold?  If there isn’t then you aren’t making a semi-bluff,  you’re just betting on the come.  There are times when you want to do this,  but they are much fewer than the opportunities for semi-bluffing,  since they do not have the added advantage of winning the hand on the bluff alone.

For example,  if the blinds were called by seven players and you’re on the button,  a raise of double the blind is not likely to induce anyone to fold  --  since they just have to put in a small amount to see the flop.  Some players make this move with two medium suited cards in late position.  It’s not a semi-bluff  --  it’s just a bet meant to increase the size of the pot in case they hit a Flush or a Flush draw on the flop.  As such it isn’t a semi-bluff.

Similarly,  betting into an opponent or more than one opponent who is very loose generally isn’t a good way of semi-bluffing,  since based on their general style of play they are unlikely to fold.  If it’s the Turn and you have a four-Flush and someone raises in front of you for nearly his whole stack,  it’s not a semibluff for you to put them all in when you have a Flush draw.  There’s no chance that this loose player will fold for a small amount more.  He’ll call.  You may win by hitting your Flush on the River,  but don’t expect to win this bet as a bluff.  It’s just betting on the come.

Generally speaking,  attempt your semibluffs when there are few opponents  -- preferably one  --  if you read them for being relatively weak when they act or if they are otherwise likely to fold when you bet,  and when your hand has some potential to get better.

There are mathematical ways of figuring out whether your semi-bluff makes sense.  Think about the chances that your bet will succeed as a bluff.  Of course no one can know this precisely.  But do your best to estimate it.  If,  for example, you think there’s about as good a chance as not that your bet will get your opponent to fold then give it a 50% chance.  Think there’s only a very slim chance,  but a chance nevertheless,  that your bluff will succeed?  Maybe you have a 10% chance that you’ll succeed as a bluff.  That’s what I mean by estimating.

Add to that chance that your bet will succeed as a bluff the chance that your hand will improve to a hand likely to win on the next card.  Flush draw on the Turn?  There’s about a 20% chance that you’ll make your Flush and win the hand on the strength of the Flush.  Have Two Pair and need to hit that Full House to win?  That’s an 11.5% chance.

To figure out if semibluffing makes sense,  look at each possibility and then add them up.  Let’s look at a situation when you have one opponent in front of you,  you have a Flush draw on the Turn,  and the pot is $100.  Under this scenario,  your first consideration is the bluff part of your bet.  Let’s say you estimate that you have a 50% chance of winning as a bluff.  The pot on the Turn is $100 and your semi-bluff bet equals the pot.  50% of the time,  your $100 will win you the $100 in the pot,  when your opponent folds to your bet.  Half the time you win the pot;  half the time you lose your $100 bet.  That makes the bet revenue-neutral.  It has,  as they say in gambling lingo,  0 EV.  It neither helps nor hurts you.  Over the long run you get out of it exactly what you put in.

But there’s more to this bet.  20% of the time when the bluff doesn’t work you still win by improving to a Flush.  So 20% of the 50% of the time = 10% of the time you win that $100.  That’s a positive of $10  (10% of the $100 you bet).  That gives your bet extra value.

And your value doesn’t end there.  There are the implied odds of winning money when your opponent calls your bet on the River.  If you don’t hit your Flush you’ll surely fold if your opponent bets,  since you won’t have any hand at all.  But when you hit your hand your opponent will,  at least sometimes,  call your bet on the River.  Let’s just say for the sake of completing this estimate that you bet the pot on the River and your opponent calls your bet 10% of the time.  Well,  the pot will be $300.  So 10% of the time your opponent will call an additional $300 bet. 10% of $300 comes out to $30.  Add that to your expected value and you have your $100 bringing you,  in the long run,  $140  --  a 40% return on your investment.

Of course all of these numbers are estimates.  Your initial estimate of how often your bluff will be believed is only a guess.  But,  as you can see,  the addition of both the chances that your bluff will succeed and the chances that your hand will improve can come to much more than the chances that your hand will succeed as a bluff alone.  That’s why semi-bluffing is often more effective than a pure bluff alone.

A word of caution about this.  Players can become seduced by the potential power of a semi-bluff.  They can start using the notion of a semi-bluff to justify wild and otherwise indefensible aggression.  I can see you in a future game: Jh-7s in early position?  4s-5s in late position?  2h-2d under the gun?  No problem.  Raise all in!!!  It’s a semi-bluff!

The problem is that if you have a relatively tight and aggressive image at the table,  you can actually start winning pots by becoming this aggressive.  And every so often,  you’ll run into a streak where the combined power of your aggression and a few really lucky draws will start convincing you that this wildness actually makes sense.  Be warned.  Semi-bluffing works in the long run when it is done carefully and selectively.  When you’re against experienced players they are cautious at first and may respect your large raises.  But they eventually catch on and start calling you down,  knowing that the law of averages is against someone who raises all the time and with junk.