Wilson Software

[Easy E]

Mason,

You have been an opponent of computerized simulation/play as a learning tool for a long time. Over ten years ago you wrote an essay that was included in Poker Essays vol 1 ("Computer games- Why they are so bad") where you dissect the effects of hot-n-cold simulators and you have included other essays in Poker Essays I and II that also commented on the lack of value in hot-n-cold simulations when used to formulate playing strategy.

I agree with you but even more strongly than you state. When you run what is called a "hot and cold" simulation, that is a simulation with no betting considered and all cards dealt out, you know what you get. A good player should be able to roughly adjust these results to something worthwhile or conclude that they have little value.

Strangely enough, however, you then use a hot-n-cold simulator (presumably Mike Caro's Poker Probe, since you mention it by name in various places) to do some simulation work for your essays. You also include data in Appendix A of 21st Century 7CSFAP where you run some hot-n-cold simulations of 2 hands against one another (primarily 3rd street, but some 4th and 5th street situations) along with some 3-way hands (all 3rd street situations). I'm assuming you used Poker Probe for these as well, since you list the dead cards missing.
In that appendix, though you caution that these hot-n-cold results are not poker (correctly), you still wrote "these results can give some important insights into whether and how a hand should be played"

So, you don't seem dead-set against simulations, but you are not very fond of hot-n-cold... which makes sense, because except for all-in situations, they are not poker as you said.

But TTH takes into account the play of the hands. That is betting, raising, checking, and folding. Now at first that might seem pretty good, and it would be good if and only if TTH does a very good job of simulating actual play.

Now to be fair, I haven't looked at the program in quite some time, but there are two things about it which I'm told and which we see posted on these forums all the time. First, it's very easy to beat even the toughest line ups, and second, some of the advice that the advisor gives is very poor.

This tells me that the program probably makes a fair number of errors in its algorithms, and these errors may get compounded throughout the play of the hand. Put another way, when you get a simulation result, you have very little idea how to adjust it to represent real play.

That's why I think TTH has value only for someone relatively new to hold 'em who just wants to get a feel for the game. I would never use if as a serious research tool.

Isn't this a fairly large leap to make, conclusion-wise, especially if you haven't used the software to any great extent AND you are evidently several releases behind?

You caution people in your books (in response to some criticism in the past, if I remember the posts correctly) not to take a "set in stone" approach to the strategy and play that you recommend or discuss through specific examples. Yet it seems that you set a different standard for software that tries to simulate actual game conditions and reactions.

I think that, for someone who credits himself with accurate analysis of the value of various poker books out there, you would want to do a more direct analysis of a software simulator before making such a sweeping condemnation.

Why don't you pick up a copy of the latest TTH release 5.0 (heck, I'm sure Chuck would give you a copy!), set up a lineup that comes close to what you are used to seeing in your $20/40 and $30/60 games, and do some playing and evaluating? Maybe 500-1000 hands, maybe customizing new player Profiles to more closely match some of your known opponents? You'll definately want to set some of the Toughness settings, so the players can adjust to your play.
Then compare the results to your actual real-world records, which you know are good, and see how close or far away they are. That, I think, would be the best evaluation of the quality of the algorithms and advice that TTH or another poker simulator provides, that is NOT a hot-n-cold simulator.


----
By the way, I ran some simulations tonight using Turbo 7-card Stud (ver. 4), based on the hands that you mentioned in your "Which Way is Up?" essay in PE II. You said you hadn't run any simulations- you "didn't bother to do any since their results will be exactly identical" in a hot-n-cold simulation.
Using a prepackaged low-limit Turbo 7stud passive lineup with 7 opponent players, and using Repeatable Deal functions to insure that the cards dealt were the same in every test, I ran 5 tests of 100,000 hands each where I stacked player 8 (Natasha)'s hand with the following starting cards:

(sorry about the charts- they aren't coming out very well and it's too late for me to make it pretty)

Starting Cards--- Win Rate-- $ Net/hand-- Avg # see 4th-- Avg # see 7th-- Avg # at showdown
(7cJh) 7d-- 28.4%-- $0.12-- 4.6-- 3.3-- 2.3
(7c7d) Jh 30.1% $0.26 4.5 3.2 2.3
(7c7d) Jh7s 76.2% $13.51 4.6 2.5 1.9
(7cJh) 7dJs 53.0% $3.06 4.6 2.6 2.1
(7cJh) 7d7s 78.2% $10.13 4.7 2.4 1.8

Note that this was a pretty loose game with little raising, with the other players seeing 4th on average 50%, seeing 5th about 30%, 6th about 22% and 7th about 16% in one test, with about 80% of hands going to showdown

Here are the results of the same starting hands, with the same Repeatable Deals, with a mix of 7 tight aggressive and average/more aggressive opponents (the "Tight2" lineup) and a player in seat 8 that likes to check-raise too much (Sandi Bagg):

Win Rate $ Net/hand Avg # see 4th Avg # see 7th Avg # at showdown
(7cJh) 7d 35.0% -$0.74 3.0 2.2 1.5
(7c7d) Jh 34.2% -$0.59 3.0 2.2 1.5
(7c7d) Jh7s 80.9% $9.36 3.2 2.1 1.2
(7cJh) 7dJs 64.6% $1.83 3.2 2.1 1.7
(7cJh) 7d7s 82.4% $5.39 3.2 2.1 1.7

Note the differences, with the hands making it to showdown averaging about 43% and multiple bets on many streets in the hidden hand situations.

Last, I changed the Lineup, putting the same tough tight Profile (Rosetta) in the 7 other seats and let Sandi take her hands up against that table

Win Rate $ Net/hand Avg # see 4th Avg # see 7th Avg # at showdown
(7cJh) 7d 36.8% -$0.72 3.0 2.2 1.5
(7c7d) Jh 35.6% -$0.62 3.0 2.2 1.5
(7c7d) Jh7s 81.6% $8.86 3.0 2.1 1.7
(7cJh) 7dJs 66.2% $1.70 3.2 2.1 1.7
(7cJh) 7d7s 82.9% $4.87 3.0 2.1 1.7

(By the way, all of these simulations, $1-5 stud with $1 bring in and no ante, included the effects of a 10% to $4 rake and a $0.75 tip for $19 or higher pots)

Now, whether these numbers are exactly right is not that important (If these numbers are wildly off, that WOULD be important!) But assuming they are not off significantly, would you agree that these simulation results could possibly be used by someone (not beginners, but more experienced players) to formulate some ideas on how to play the hands given, in various situations, WITH THE UNDERSTANDING that the simulation's "advice" should not be taken carte blanche, any more than the advice from a poker book should be?
In fact, with the wide range of factors that could be influencing these numbers, the simulation numbers would have to be looked at very hard to determine the reasons for the results. BUT this would also stimulate some thought as to how and why things could happen. <ul type="square">[/list]

[Mason Malmuth]

Hi Ed:

I only skimed through your long post, but I think I can answer it "easy."

First off, "hot-and-cold" simulations make far more sense for stud than they do for hold 'em since it is often correct to chase in stud while that is not the case in hold 'em.

In the past, I have looked at three different releases of TTH. I plan to look at the current release of TTH in the near future. I plan to look at how the "tough" players play, and at what the strategy advice is. If my opinions change, I will post them here.

Best wishes,
Mason

[Blackjackgod]

Personally I don't think the tough players are that tough. When I am bored and I am in the first few rounds or so I will rasise the blinds all-in with aces and i usually get called by at least one player who will have hands like 10-6...I would like to think most people wouldn't call with that. And the game is set on "tough".

[Dentist]

With all due respect, I am very confident that if you evaluate the game in terms of the quality of the opposing players and the strategy advice given I think you will be disappointed...

However, if you evaluate the game from a standpoint of: Hey, this game is a computer simulation of playing hold 'em, shows the player the devastating effects of the rake (at LL), and allows you to practice and put yourself in game situations, it is valuble.

I am confident that it is infinitely more useful to a beginner/advanced beginner than it would be to a poker scholar like yourself..

I don't know anything about the simulations because that's not why I bought it.

But I thoroughly enjoy the stud/hold 'em/Omaha8/and tournament versions because I can play 200-300 hands in an hour and the competition I can program it to face is very similar to the low to mid-limit games I play in.. those people make mistakes, and so does the computer.

I don't use their advice either, I use my own thought process gained from 2+2 books...

It's a fun game, and a useful practice tool.
The tournament mode is valuble for putting you in those "decisions" over and over again.

[Mason Malmuth]

Hi Dentist:

Your comments on the program are essentially my opinions as well. I think it is a very good product for someone relatively new to poker who wants to get a feel for the game, and this probably includes tournaments even though I have never looked at their tournament software.

I do question its value as a research tool and for teaching someone how to play well.

Best wishes,
Mason

[Gabe]

The biggest problem with the TTH program in the past was it's players' inability to properly take in to account the betting on the previous round. I don't know if he fixed this or not.