3-Bet Stats Misleading?

I guess I'm not sure how database programs like HEM calculate some of their stats. Tonight I just realized that the 3-bet stat is not the percentage of starting hands that you 3-bet. It's actually the percentage of times that you 3-bet when it was possible for you to 3-bet, i.e. when facing a raise.

So if your opponent is 3-betting 5% of the time, that does not mean his 3-bet range is 5% of starting hands, or, assuming he only does it with good hands, 5% of the top of his range.

Has anyone else been misled by this stat?

Preflop you still haven't made any bet when you get the chance to 3bet, so it's still 5% of possible starting hands.

On the other hand, people who are good enough to 3bet 5% are usually good enough for their 3bet range not to be strictly the top 5%, i.e. strictly {99+,AJs+,KQs,AKo}

fwiw, my 3bet range at 100nl+ is the table below... I 3bet about 4% overall, so that would be {99+,AQs+,AKo} if i was strictly playing the top 4%.

Attached Images

3bet.jpg (144.7 KB, 20 views)

I'm confused. The way you explain it is what I had previously thought. But tonight I played a short session and looked at my 3-bets -- they were 9.1% but only actually 5 times in 159 hands. What I discovered is that it was only possible for me to 3-bet 55 times. So by doing it 5 times out of 55, it came out to about 9%. But that doesn't represent a 9% range. To be honest, it was only an actual 4.2% range, 99+, AQs+,AKo.

Under Filters, I chose Could 3BET = True and Did 3BET = True to get this info. That's why I ask is the 3Bet stat misleading.

What I mean is that if you run "Could 3BET = True" over your whole database, you will have a distribution that is essentially the original one: .3% for suited cards, .45% for pairs and .9% for unsuited cards.

that's not 100% true, there will be a little less aces in there than usual, but that's insignificant overall.

it only means you ran hot.

As far as I can tell, the 3-bet stat has nothing to do with your range or how hot you are running.

You could get AA every time and not 3-bet it once if your opponents didn't give you a chance to do so.

Maybe if I looked at the entire database, what you are saying would be clearer to me, but that computer has gone to bed for the night.

This is why it takes longer for 3bet stats to be meaningful

Like you say, in an individual session, perhaps you don't get a chance to 3bet. But that just means your 3bet stat didn't change at all today. It only cares about those times in which a 3bet is possible.

This isn't mine, but a great mind once said:

In the same vein your V$IP and PFR are not necessarily representative of your range.

You could play a weird 15/10 style where by you played random hands like 720 23s etc. V$IP measures the % of times you put money in the pot, not the actual strength of your range.

PFR similarly measures the % of times you raise preflop when you have oppertunity to do so it doesn't have to reflect the top 15% of hands (although it tends to).

When looking at 3-bet stats you really aren't overly concerned with the exact value. You are looking for information that your opponent could be 3-betting light and then evaluating that information in light of the current situation.

Even if you think he is 3-betting light, your default play is still to fold most hands, but throw in the occasional 4 bet bluff and 4-bet value raise.