Stat-reads vs. real reads for multitablers

[TimovieMan]

The following post is specifically addressed to the multitablers among us. The ones that play only one single table and have all their attention on that table obviously use only real reads since stat-reads would take too long for them to gather (and stat reads are nowhere near as accurate).
But the ones that play 3-4 or even more tables at a time are too busy playing cards to be able to notice what goes on at the table during the hands we folded. For them (or should I say “us”) the stat reads become more important. We’ll play more different opponents, we don’t see half the hands they show down, we don’t see any betting patterns since we’re busy following other tables, etc.
In some cases we even have ADD, so we wouldn’t even WANT to follow the table in folded hands.

For us (or at least ME) the stat-reads are important. They give us a clear idea of the looseness of our opponents, of their aggression, etc. It’s our first step to playing the players instead of just the cards.


But the real question (and reason of this post) is the following:

At what point do the stat-reads become irrelevant (or at the very least less relevant), and at what point should we start to use PokerTracker to really analyse how our opponents play their hands?


Personally I’m VERY careful with players I have less than 50 hands on (unless I watch them openlimp UTG with 75o and real obvious stuff like that), and as of 50 hands the stat-reads become more accurate. It helps me divide my opponents in the basic loose-tight categories, passive-aggressive, solid-maniac, overplayers, rocks, etc. and I adjust my play against these opponents accordingly.
Then I basically *update* my stat-reads from PokerTracker (I’m currently playing at Everest where PAHud doesn’t work) every time I have another 50 hands on them (so every 100th, 150th, 200th etc. hand). Note that this is just an approximation. I’m not checking PT every two minutes to see how many hands I have on my opponents so I can update the stats.
That’s basically my stat-reading.

When I have 200 hands on my opponents, it becomes interesting to know how they play OTHER than just looking at some simple stats (that are inaccurate as reads). Then I use PT to look at the specific hands they VP$IP with, the ones I’ve seen them PFR with, etc.
Basically that’s just analyzing the stats again, but this time a bit more in detail (more or less the same analysis people use to try and identify the biggest leaks in their own game).

But when I have 500 hands on an opponent, then comes the fun part. I now have enough hands on my opponent to 1) assume I’m likely going to meet him at the tables again in the near future and 2) have some good solid reads on him.
And that’s when I start to replay the hands I have on this specific opponent in an attempt to figure out how he reasons, so I’ll know what I’m facing if he’s in the hand…
And I repeat this process for every other 500 hands (although I then tend to replay them ALL to have a better chance of spotting a real pattern in his play – since 500 hands still is way too small a sample)…


How do you guys incorporate your reads in your game? Is this a good method of getting an accurate read on any opponent? When is a good time to analyse your opponent by replaying all his hands from his point of view? Do I start that too soon, or way too late?
The reason I wait till I have 500 hands is because I also don’t want to spend 5 hours analyzing things with PT for every 1 hour of actual play, whereas 1 hour of analyzing for every 4 hours of play is imho in good proportion.

Do most of you solely use real reads, even when multitabling? Do some of you solely use stat-reads?
And most importantly: how do you guys feel about using stat-reads vs. real reads?

Don’t know if this is a good topic, but I think there’s plenty to be talked and discussed about on this topic…

[Dankenstein]

I think this is an interesting topic. I multitable on 4 tables at a time (FR) when I play. I don't look at a persons stats without at least 100 hands on them.I treat most of the players the same until they prove to me they aren't a "normal" player. Once they prove to me otherwise I take notes on the player and I use these notes for reads much more than stats. I do put a good amount of stock in AF though. If an opponents AF is less than 1 and he wakes up in a hand I incorporate that into my read. Similarly if an opponent has a high VPIP coupled with a high PFR and is raising preflop I might increase my 3-bet range against him. Or if his VPIP is extremely small and his PFR is small I adjust my range accordingly. To me stats are only good for reads when they are EXTREMELY deviant from a normal range. All in all Stats can be decieving and I think putting too much stock in them is -EV so I trust my notes much more than player stats.

[neurotiq]

[ QUOTE ]

Personally I’m VERY careful with players I have less than 50 hands on (unless I watch them openlimp UTG with 75o and real obvious stuff like that), and as of 50 hands the stat-reads become more accurate.

[/ QUOTE ]

I don't think that 50 is a magic number at which your stats suddenly become "more accurate." Is there really a significant difference between 49 and 50 hands?

More to the point, though, it's very possible for me to get a hot run of cards over only a 50-hand sample. You might conclude that I'm really loose and really aggressive when, in reality, I'm just getting lucky. It's also possible for me to be incredibly card dead over a 50-hand sample. You might conclude that I'm a nit.

This probably speaks volumes to your point about about how reads are more important than stats. Small sample sizes can be misleading, and lead you to conclude something totally incorrect about a villain.

[bellatrix]

[ QUOTE ]

I don't think that 50 is a magic number at which your stats suddenly become "more accurate." Is there really a significant difference between 49 and 50 hands?


[/ QUOTE ]

Yes, but we have to start to assign labels at some point. The 50 is an arbitrary, but statistically significant boundary. Since we have so many opponents, it's just a filter for villains that have only like 20 hands played. i agree, that there is no difference between 49, 50 or 51. This is why the #hands tag is important. I will take the label/note at 75 hands less serious than a label/note at 750 hands, but just as PT I start doing a quick analysis around 50 hands. The VP$IP converges fairly quickly that will give you an estimate of at least that number. At around 100-200, I start narrowing down the PFR. Around 200-500 the AF. Above that I don't change my label per se, but start noting down common threads of play. I 2-table while playing, so I have no advice for more-tablers.

[neurotiq]

[ QUOTE ]
The 50 is an arbitrary, but statistically significant

[/ QUOTE ]

I can feel my former stat professor cringing over 50 being called a "statistically significant" sample size. [img]/images/graemlins/smile.gif[/img]

[Bona]

stats prof=nit
100>50>25>10>0

[BadBigBabar]

yes but 50 is insignificantly small

i go with 300 as when i actually start to pay attention but you could argue 500 or 1000 is actually meaningful

[Xylocain]

I havent read the posts but as for the number 50 ... or whatever.

A villan that donks an AKT two tone flop after your pfr and c/c two streets with T9o 1 hand is enough.
A villan that raises 5 hands in a row ... 5 is a significant number.
A villan that folds 50 hands in a row ... we have no additional info more than he is probably getting cold carded and he is not loose.

so ... it depends.
my 2.

[bellatrix]

[ QUOTE ]
[ QUOTE ]
The 50 is an arbitrary, but statistically significant

[/ QUOTE ]

I can feel my former stat professor cringing over 50 being called a "statistically significant" sample size. [img]/images/graemlins/smile.gif[/img]

[/ QUOTE ]

Yes it is, because for the purposes we only want to know a TREND, an error of +/-(10-15%) is ok for our purposes, which would not be correct in political polls, for example.

neurotiq said just like 2 days ago:

[ QUOTE ]

Maybe fold this guy preflop. A 4% PFR is icky for our AQ.


[/ QUOTE ]

This was on a 46 hand stat... [img]/images/graemlins/smirk.gif[/img]

Look, at 50 hands (or 40 or 60, who cares about the exact number), you start to see a TREND of VP$IP. I'm not talking about other stats, just that of VP$IP. True the statndard deviation might still be around 10% points, but if VPIP=60 it is a 3 sigma result that the villain is a rock (opposite also true). You will not know with a VPIP=20 if you have a very tight player or a sLAG, it can vary wildly withing the typical PT numbers, but you have a trend that this is a player that is not a maniac that will play 80% of all hands.

Around 100-200 hands you will see TRENDS converging on the PFR. So if at 200 hands the PFR standard deviation is perhaps around 3%, so if you see someone with 2% PFR, well then you know he only raises the absolute premium and you can give him leeway to top 5%, but not at 50 hands, of course.

And then around 400 hands, you can start to BEGIN to see trends in AF.

If you have played over 3-5k hands with somebody, you can start to look at W2SD,WaSD,FAF,TAF,RAF and even some individual hands...

And then notes like "loves to slowplay flopped sets","plays big PP fast","called fown AK UI", etc. are much more valuable than stats.... (but as above, these are real reads, and not stats reads).

[Smurph64]

I use stats as a yardstick to determine if my read was correct not the other way around.

There should be convergence at some point but I have found stats misleading so many times I just use them as broad point guidelines.

It is much more relevant to me whether a player calls to showdown with bottom pair vs calling to showdown with an ace or what he defends or steals with.

I gain more insight if I know the guy reraises with draws or needs two pair.

Typically I can tell a lot about a player by marking down what he cold calls with and it is on those players that I expect to see some convergence of stats at some point.

The other day I had an interesting situation. I was playing at a table when a guy made an interesting comment:

He was acting in front of me and said: " I am limping with any thing because this fish (me) raises preflop 45% of the time"

To which another guy stated, "you do that, he is a fish"

The one guy who made the first comment just looked at my stats, the other guy who replied avoided me or 3 bet because he knew I had raised a lot in the last few hands because I had pocket AA's twice AK 4 times and KQ suited a few times.

It took that guy who called me a fish about 100 hands to figure out I was killing him before he left.

That session I had indeed raised 45% pfr and played 48% at that time but it was only 40 hands in. After 140 hands I was still high but much more in line with my usual stats. I was still high because I iso raised the limper whenever I was in late position and he was first in.

It's really important to try and put notes on your players that explain situations. One good note can make or save you a lot of money a lot more than a sample of stats.