Do you agree with Mike Sexton on probability of win?

[mdlm]

In the latest CardPlayer, Mike Sexton writes:
Wake Up and Smell the Coffee – Part II

"Do you consider yourself a winning poker player? If so, you should be on the plus side of the ledger after 10 sessions of play. Basically, this means you should win six sessions out of 10, assuming your sessions last about eight hours apiece. Professionals should win seven eight-hour sessions out of 10. If you’re not doing this, it’s time for you to wake up and smell the coffee."

Are his numbers correct? Also, how many players regularly play 8 hour sessions?

[Billy LTL]

I'd say that's about right...oops. Mike Sexton forgot to add a zero or two.

70/100 is closer, 700 wins out of a 1000 eight-hour sessions would be a lot more accurate I think for a good player.

Billy

[Clarkmeister]

I thought his numbers were ridiculous and had no basis in fact.

[mdlm]

Clarkmeister,

If you think Sexton's numbers are wrong, which numbers are right?

Thx.

[Jimbo]

Even though my session win/loss ratio falls within the expert range of Sexton's guidelines I see little relevance or value in that alone. If he were to couple that with you being a total winner over all these cumulative sessions and preferably quantify that in some manner then it may have value. As it stands if you lose $5000 in 2 sessions and win a total of $1600 in 8 winning sessions you are an expert!!! What is up with a measurement like that?

Jimbo

[Noo Yawk]

I could shoot holes in this all day, but why bother? Only over the course of several years are you going to know if you are a consistant winner as long as you keep accurate records. And BTW, if you don't continue to learn and adjust your game, that may not last. I saw the biggest fish in my cardroom reading HPFAP the other day. He obviously wants to get better, so I'd better not rest on my laurels. 6/10 wins tells me nothing today if others are going to get better tommorrow.

I'm sure Mr. Sexton is a very good player, but that part of his article made me wince. Its symbolic of his generation's approach to poker - thinking in terms of sessions won/lost rather than hourly rate and SD.

Thinking in terms of sessions won or lost leads to all sorts of bad practice like "quitting when your ahead" and "money mangement."

I suspect the ratio of winning/losing sessions varies quite a bit between winning players. For example, maybe a real expert who's chasing every minor edge s/he can, actually experiences more losing sessions than a "solid" player. But the expert wins more money because his wins are bigger.

I'm pleased that someone raised this issue, as I've become a lot more experienced as a poker player over the past few years, I've begun to realise why so many of David and Mason's essays voice concern about the amount of bad advice out there. In every issue of Card Player, you'll find at least one article with a factual mistake or bad concept...

- roGER

[Mason Malmuth]

I haven't read the article so what I'm about to say could be off. A good player, at limit poker, should be a winner in the majority of sessions that he plays providing that he is playing a reasonable length of time on average. However, is over your poker career. You can certainly have losing streaks where winning 6 out of 10 times is only something that you dream about.

While I like Mike Sexton, and respect him as a poker player, he is not a statistician. When this is the case, and when it comes to this type of statistics, the intuition of players like this is usually off, and sometimes substantially. So even though I haven't read the numbers that he gives in his article, I would be skeptical at best.

An example of this are the bankroll requirements that I have published. These first appeared in a Gambling Times article in the mid-eighties. Over the years, I have been told by many good players that they are way too high, only to be retold by them at some future time that I was right after all.

What this indicates is that most reasonably good players who do not understand well the underlying statistics of the game, tend to underestimate how often extreme events will occur. That's why we, for instance, see people like Phil Helmuth kicking down doors. The so called statistical outlyers are far more common, especially if you play a lot of poker over a fairly long period of time, than they realize.

/forums/images/icons/blush.gif So to answer your question, you now know why I only drink green tea or decaf coffee.
/forums/images/icons/tongue.gif

Best wishes,
Mason

[Mason Malmuth]

I agree completely except for one very important fact. If you are a winning player, actually expert would be a better term, you should be able to routinely see some errors in the play of your opponents that

1. You know for sure is an error, and

2. You of course do not make.

When this is the case, you should be able to tell relatively quickly if you are a winning player. Of course, if your idea of what is correct/incorrect play is flawed, then this method won't work very well.
/forums/images/icons/grin.gif

Best wishes,
Mason

[Noo Yawk]

As my old college teacher used to say:
"If the phone rings, and you answer the bannana, there is probably a flaw in your education"!! /forums/images/icons/grin.gif
I always loved that line!! /forums/images/icons/cool.gif

[AceHigh]

Being a winning poker player is about winning money not winning sessions. Players should try to maximize $$$ not number of winning sessions.

How often do we see stuck players staying in a game when the are not playing there best or the game is tough because they are "trying to get even"?

How does Mike Sexton know how many sessions out of ten he wins? I am 99% certain that I read an interview in card player with him by that Roth Slatboom (sp?) where sexton said that he doesn't keep records and joked that he knows its been a bad month if he can't pay the rent.

[Bozeman]

Actually, as long as you are playing consistent sessions as Mike hypothesizes (8 hours), this approach is precisely consistent with hourly rate and standard deviation.

If you win 60% of your sessions, your session SD is 32 times your hourly rate.

If you win 70% of your sessions, your session SD is 15 times your hourly rate.

Thus if an expert (professional) wins 1-1.5 BB/hr, Mike's claim is equivalent to saying that he believes that a professional session SD is about 15-23 BB.

This seems a bit small to me, but I would guess that anyone who has played as long as Mike has would have a reasonably accurate heuristic.

The rest of the complaints in this thread boil down to taking 6/10 to not be an average result over time, and issues about money management, which Mike at least implicitly excepted in his article.

Craig

PS I am mostly a tournament player, so could someone give me data on what their ring SD is over a reasonable amount of time? BTW, 8 hour SD ought to be 1 hour SD*2.8.

[Jimbo]

if David and Mason are correct, and I have no reason to believe otherwise, then it takes a thousand or so hours to be 95% sure you know your correct SD. Multiply that by 8 (to convert sessions into hours) and now you need 4 years worth of data on your sessions to believe it is an accurate measurement and have a useful ststistical sample. Long before then you will either be a cab driver or know you are a winning player without the session measurement. Even if you assume that is incorrect why would you believe you could not book less bigger winning and more smaller losing sessions for a long period of time due to other
circumstances?

In summary I still believe his measurement has no value to either a winning or losing poker player unless you have a very large side bet on this statistic.

Jimbo

[Bozeman]

Jimbo wrote:

"if David and Mason are correct, and I have no reason to believe otherwise, then it takes a thousand or so hours to be 95% sure you know your correct SD."

Not true. SD should be well known in ~30 measurements.

Also, I believe that SD is relatively constant among winning players. Do you have good reason to believe this is false?

"you need 4 years worth of data on your sessions to believe it is an accurate measurement and have a useful ststistical sample."

The mathematics of SD depend on the number of measurements, not the amount of time. If you measure by the hour (even if you play sessions), 30 hours (4 sessions) is enough, if you only record after each session, then you need 30 sessions (~ 1 month). Also, this analysis can be predicated on using someone else's measurement of SD and presuming your SD is not very different.

Again, the constraints of sufficient time to show that you are averaging 60% winning sessions apply, but they are the same as those that apply to computing hourly rate. Ok, so a slightly better measure could be obtained by keeping track of hourly session wins, but only marginally.

"Even if you assume that is incorrect why would you believe you could not book" FEWER "bigger winning and more smaller losing sessions for a long period of time due to other
circumstances?"

Because in the long term, your sessions better be normally distributed, or SD does not make mathematical sense. (Again, we are hypothesizing exactly 8 hour sessions, so money management anomalies are not present (stop losses, for example)).

"In summary I still believe his measurement has no value to either a winning or losing poker player unless you have a very large side bet on this statistic. "

On the contrary, show me a winning player who does not win a majority of his sessions, and I will show you a severe statistical outlier, given he has played 30+ sessions.

Craig

PS After 30 sessions, the chance that a break-even player will have won 60% or more is 18%, and the chance he will have won 70% or more is only 2%.

[Jimbo]

Great explanation Craig!!! I see much more value now than before. Do you think Sexton understands the underlying math as well as you or intuitively understood this to be the case?

I suppose I was simply using 20 years of playing with one particular fellow who falls so far outside your guidelines that I did not think it possible without Sexton's supposition being incorrect. As far as I can tell now it would be more likely that my friend won the lottery 3 times and had septuplets on Good Friday than achieve his profit by losing significantly more often than he wins over this 20 year period. He will not limit his profit in one session but limits his losses to our minimum $500 buy-in per session. Does this stop-limit provide the difference in SD?


Jimbo

[Bozeman]

Yes, a stop loss or a stop win can easily change the shape of your result distribution, so that a non-normal distrib. of session results occurs. However, if you consolidated the data into 8 hour chunks, normality ought to be preserved.

I suspect that Mike has only an intuitive understanding, and I would not be surprised if his 6/10 number is not nearly as good an estimate as his 7/10 since he is a very good player.

Overall, I was trying to make the point that, while having all the mathematics is nice, there can be meaningful heuristics that work for the less mathematically inclined (I think I failed).

Craig

[Erdnase]

Hi all,

Yes, I was also thinking of this article (good thing I read all the posts...lol).

The author is Rolf Slotbloom (dutch) and it appeared in PokerDigest, I think.

Isnt standard deviation, hourly rate and such very much dependend on the type of poker game and your opposition??

Greets, Erd.