Turn raises

[mchilger]

Terry, I've been taking a look at your book. I started at the turn chapter . Some of the tips and notes are quite good.

On page 151, the note shows you did an analysis of turn raises in a $5-$10 game and found that 48 of 51 were top pair or better. I find that really eye-opening and valuable for many players here.

I was wondering if you've done a similar analysis of the $30-$60 games at Stars and FTP?

Matthew

[TerryBorer]

Thanks Matthew. Lawrence calculated that number for the book and we didn't do it for the sites you mentioned.

That's a pretty lousy answer, and since I'm making big bucks this month I decided to count now. So I wrote an SQL query into my PokerTracker database to find hands that met the following conditions:
$25/$50
5 or 6 players
raise or c/r on turn but not a bet/re-raise
went to a showdown

For all players who were not me (all hands have showdowns)
37 top-pair or better
4 value raises (judgment call by me - example is 99 on T622 board)
4 semi-bluffs or moves (remember that I won't see semi-bluffs if either players folds)
5 pure-bluffs

Just for fun I did it for 5/10 as well (using 50 random hands meeting that criteria)
43 top-pair or better
1 value raise
5 semi-bluffs or moves (again, some judgment calls)
1 pure-bluff

And again at 3/6
45 top-pair or better
2 value raise
2 semi-bluff
1 pure bluff

I did this for my own hands (with or without a showdown) with the following results:
32 top-pair or better
4 value bets
6 semi-bluffs or other moves
8 pure bluffs

Again, we can only make calls on hands we see, so the actual rate of bluffing will be somewhat higher and is influenced by how the opponents play. So this isn't definitive but there is a clear trend.

I'd be happy to answer more questions like this if people would like.

-- Terry

[mchilger]

This analysis is quite interesting. I wrote in ITH that you must respect turn raises. However, I think some players start to somewhat discount their strength in short-handed games. Some players also think that in today's game people are always raising on bluffs. Yet in this analysis we're looking at a 3 to 1 ratio even at higher limits in short-handed games.

You guys talk a lot about how calling a turn raise often means calling a river bet, so you're really risking 2 big bets hoping to catch your opponent with a weak hand. 3 to 1 odds can be difficult to overcome.

One of my goals is to get back to Poker Tracker this year as I just haven't played a lot of cash games to do this sort of analysis the last couple of years. Appreciate the stats.

Matthew

[osmosis21]

I really think that the numbers are skewed since you arent seeing hands that didnt go to showdown. Im absolutely positive that in the last year raising the turn by a typical tag/lag doesnt not necessarily mean tp or better. In fact I do find that too many players are making poor decisions to semi bluff when they have no folding equity.


Do you have the stats which say out of those turn raises how many times did the opponent fold to the raise or the river bet and how many times did the raiser fold to a three bet or a river raise?

[jeffnc]

[osmosis21]

What I meant to say is that in the games I play ppl are not raising with tp or better as much as these results would suggest.
Partly due to the flush draw semi bluff and free showdown raise (both which are often used in situations where there is no folding equity, unfortunately my games have a number of thinking lags and tags who are often way too aggressive on the turn, which just means bloated pots...and helps explaing why many players are showdown bound now.

From todays session I can think of 3 spots where someone threebet the turn on a draw (different players) and a couple of turns that were capped with less than top pair.

[TerryBorer]

Interesting discussion.

I agree the numbers over-estimate the number of times the raise is TP or better. I think jeffnc made a good point about thinking about the data in the context of his game.

Osmosis - why don't you share some data with us? Take a look back at 50 hands in succession that meet that requirement (which will span hundreds of hands) and share with us your numbers (and the stakes you play). If you use a SQL PT database I can share the query with you if you'd like. Another way to do this is write down the results as you play your next session.

I don't believe humans can accurately estimate frequencies (I've seen data on this). I also think people will tend to remember bluffs more. I was surprised to see how little bluffing I measured too.

Looking at my numbers I had the good 36 times out of 50 and I'm very aggressive up against other loose, aggressive players.

-- Terry

[deNile]

[osmosis21]

Ill take a look later and count them up..I agree that humans are poor judges of frequency, however I know I had a lot of tough decisions today and I was calling down or three betting the turn more often than not and had a very profitable session....but I will do a count...cant do a sql queary yet, but I will be switching over soon.

[nsidestrate]

Using your categories ar 15/30 short-handed:

top pair or better -- 37
value bet -- 3
pure bluff -- 7
semibluff -- 3

This method is a bit misleading as "top pair" included several situations where the raise was essentially a bluff. There was one, for instance, where the guy had A4clubs on an Ace high board with four diamonds. He was obviously trying to represent the flush and lost the hand. At least of the value bets was stupid (22 on a KKT4 board).

There is a pretty huge difference between the initial stat (48 of 51) and some of the other numbers we are finding here (37 of 50). In the first case, the raise means top pair 95% of the time and you could just fold to that raise pretty safely. In the latter case, with 20% bluffs or semi-bluffs, the combination of that 20% chance you are good combined with outs when you aren't will usually make calling down reasonable with a range of hands.

It was interesting to me that several players on whom I have a ton of data almost never raise the turn without the goods. I added to a few of my player notes while reviewing this data, so it was good exercise.