How to tell if you are running good or bad (Revised v2)

[krishan]

The idea behind this guide is too help people determine if they are running bad or playing bad/running well or a god at poker. There have been a few model designed to determine how well you have run. They range from checking your big pairs, formulas including W$SD and W$WSF, or a website that you import hand histories into.

My theory has it's base in the pokertracker misc. tab. All the information held here assumes you are only looking at hands that go to showdown. One hole in my theory is it doesn't take into account draws that you don't go to showdown with.

In it you can tell how each of your poker hands is performing. One pair, two pair, three of a kind, straights, flushes and full houses comprise over 93% of money earned. Here is the breakdown in terms of %.

One pair ---- 21.9%
Two pair ---- 24.7%
Three ------- 14.5%
Straight ---- 12.0%
Flush ------- 11.4%
Full House -- 10.7%

To tell if you are running bad, you compare the performance of these hand types with the long term quantities and/or win rates. For instance with top pair, you should win 44% of the time. If you are performing at 15%, it will be very tough for you to have a winning day. If top pair is performing at 77% you are getting lucky.

Quantities aren't important for top pair and two pair because happen so often. The deviation isn't as important. Here are the expected performance for top pair and two pair.

Top pair -- 44%
Two pair -- 57%

I came up with these numbers by looking at over 400K hands worth of Misc tab data. You can compare your own long term data to see if these numbers look right. If not, feel free to use your own.

Premiums (straights, flushes, fulls) are different from one pair and two pair because they come up so infrequently. Also you make much more per hand with premiums. Because of this if you are either not getting your share of premiums or if they are not winning at showdown the average amount, it will seriously impact your bottom line. As an example, flushes are worth about 4.7 BB in my database and each top pair is only worth 1.2 BB. Each premium occurs roughly every 200 hands. Straights happen a little more and full houses a little less. Premiums should perform at 80% at least.

I haven't looked into three of a kind that much but since it's worth 14% of your earn I should give it more attention. From a preliminary look at my database I'd say three of a kind should win at 75% or better.

Using this you can tell if a particular day was good or bad due to cards. It really won't help you become a better player. But somedays, the reassurance that you don't suck at poker can be worth quite a lot.

Krishan

[mungpo]

Nice work, krishan.

[Surf]

Great work Krishan! I'm anxious to take a look overall at how my hands have been performing.

I think one of the most important things to take from this is that 45% of our earn is from 1 + 2 pair hands - which actually is more 1 pair than 2pair, since pokertracker counts "2pair" as 1 on the board and 1 from your hole card(s) being used.

This staggeringly large % of our earn is from the most complicated hands to play. I'd recommend players that are running bad examine their 1pair and 2pair hands - are they missing value bets? You can check your hands that were not bet on the river in the misc tab. Calling down too much? You can look at hands that were called on the river and saw a showdown.

These types of filters are great for working on the most complicated part of our game - understanding how to play marginal situations.

Surf

[bds]

Thanks for all your hard work on this, Krishan. I’d like to post my own poker tracker information, compare it to this work and seek input on whether I’ve looked at the data correctly and whether my conclusions are flawed.

You say: [ QUOTE ]
All the information held here assumes you are only looking at hands that go to showdown.

[/ QUOTE ]

However, it seems to me that when I look at how many premiums I’ve had of each type hand, I need to base this calculation on the total hands dealt to me, not on the number that went to showdown. Is this a correct conclusion?

Here’s what my results show about the percentage wins for each of these hands and how they compare to your model.

Type Hand-----Model---My Data
1 pair----------44%-------38%
2 pair----------57%-------53%
3 of a kind----75%-------73%
straights ------80%+------86%
flushes---------80%+-----83%
full houses----80%+------87%

So, I’ve won less than my share with 1 pair, 2 pair and 3 of a kind hand and more than my share with the premium hands (straights, flushes and full houses).

Next, I looked at the percentage of my overall win for each of these categories of hands compared with the model.

Type Hand-----Model---My Data
1 pair----------22%-------15%
2 pair----------25%-------23%
3 of a kind----14%-------18%
straights ------12%-------15%
flushes---------11%------13%
full houses----11%-------14%

The results are similar, with less of my overall $$ won coming from the first three categories and more from the last three categories, consistent with the results when I looked at percentage of wins from each type hand.

Finally, I looked at how many of each type of premium hand I have had. I recall from the earlier thread that you said (I think) that for shorthanded play, the expected occurrence of each premium hand was 1 of 170 and for full ring games it was 1 of 200. I’m not sure if that is still your conclusion, so I calculated both ways. I play shorthanded. Either way I calculate, I have way more than I should of each premium hand (if I’m doing the calculation correctly.) I take total number of hands dealt to me (255,051) and divide that by either 170 or by 200. Correct?

So, at 1/170, I would have 1,500 of each and at 1/200 I would have 1,275 of each. I actually had:

straights----2,511
flushes------1,966
full houses--1,957

Now, the question is what can I learn from this data. All input appreciated here.

First, it seems like I’ve had many more than my share of premium hands and won much more than 80% of the time with them. Does this mean that I’ve run very well over the last 250,000 hands? If so, I guess it means I need to expecting the poker gods to provide me with many fewer of these of the next large number of hands. Correct? I’d like to think this was not so, but the data surely points that way. Seems to me like 250,000 hands is long term, but perhaps not?

The other conclusion I draw from looking at my win rate with 1 and 2 pair hands taken to showdown is that I show down too many of this category of hand. I already knew that I need to learn to fold these types of hands sooner yet I seem to continue to call down hoping that the opponent is bluffing me (which he/she often is), but apparently it is a real leak for me. Yet, I wonder if my tendency to stay too long with a 1 or 2 pair hand can result in my actually ending up hitting more of the premium hands? For example, could staying too long with a 2 pair hand result in more full houses than the model shows? Could staying too long with a 1 pair hand with a straight or flush draw create a similar result? I don’t really know how to interpret this.

I’m most interested in other observations as to how I can improve my game by analysis of these numbers in comparison to the model. Again, thanks for your hard work, Krishan. You are appreciated.

Cool!
How do I check "Top Pair %"? I only see "One Pair" in the list.

[Wynton]

Krishan,

Your analysis is chiefly aimed at finding out how often one receives and wins with various hands, right, as opposed to how many bbs one wins with those hands? I ask because it seems impossible to know whether an unexpectedly low number of BB won per hand is due to bad luck or just inability to squeeze that extra bet out of one's opponents.

[krishan]

[ QUOTE ]
Cool!
How do I check "Top Pair %"? I only see "One Pair" in the list.

[/ QUOTE ]

It's a typo. It should read one pair.

Krishan

[Entity]

[ QUOTE ]
[ QUOTE ]
Cool!
How do I check "Top Pair %"? I only see "One Pair" in the list.

[/ QUOTE ]

It's a typo. It should read one pair.

Krishan

[/ QUOTE ]


I'm curious about this. Here's some stats from a recent bad run of mine (-.79bb/100 over my last 6,856 hands -- two days of running pretty bad).

--Showing only hands that were not folded.

High card - 16.47% - $360.06
One pair - 38.79% - $2,992.14
Two pair - 51.61% - $2,377.43
Three of a Kind - 78.57% - $2,276.62
Straight - 86.96% - $2,030.75
Flush - 94.29% - $1,903.00
Full House - 77.14% - $1,393.5
Four of a Kind - 75% - $166

I'm not sure what that qualifies as. I feel like I can tell I'm running bad somewhat due to the relatively poor performances of my normally big hands:

AA - 69.23% - 1.31bb/hand
AKs - 52.94% - (.12bb/hand)
AKo - 52.05% - .03bb/hand
AQs - 55.56% - (.57bb/hand)
AQo - 59.38% - .39bb/hand

...

Anyway, thanks for the post. Interesting at least.

[nbake]

adding to favorites...oh yeah

[krishan]

[ QUOTE ]

Finally, I looked at how many of each type of premium hand I have had. I recall from the earlier thread that you said (I think) that for shorthanded play, the expected occurrence of each premium hand was 1 of 170 and for full ring games it was 1 of 200. I’m not sure if that is still your conclusion, so I calculated both ways. I play shorthanded. Either way I calculate, I have way more than I should of each premium hand (if I’m doing the calculation correctly.) I take total number of hands dealt to me (255,051) and divide that by either 170 or by 200. Correct?

So, at 1/170, I would have 1,500 of each and at 1/200 I would have 1,275 of each. I actually had:

straights----2,511
flushes------1,966
full houses--1,957

Now, the question is what can I learn from this data. All input appreciated here.

[/ QUOTE ]

BDS,

Just checking, are these numbers the number of premiums taken to showdown? or the number from the total times column? That might be the issues. You should be using the number of premiums taken to showdown. Let me know and I'll reanalyse.

Krishan