A Few Words on Using Statistics

pokeroff 0 1710
A Few Words on Using Statistics

Poker students often ask me which statistical values I use in my playing - even more often they ask if this or that value is a good statistic to use. I will try to answer these two questions in this article.

Maybe you already know, or maybe not, but as it turned out at the beginning of my journey into the world of poker I did not have the opportunity to use statistics - and now, looking back that time, I think it did my skills a world of good. Once the moment came when I decided to take the game seriously, I had a bankroll worth a couple of hundreds euros residing on B2B network accounts (since re-named Entraction and now closed). I did not bother to transfer the bankroll somewhere else, and I was more than satisfied with playing for euros and the high rakeback percentage.

This network was not supported by the HUDs that were popular at the time. Eventually I had the opportunity of uploading hands into Poker Tracker 2 for analysis, but did not have real time statistics. Then, PokerOffice with a built-in HUD appeared, but as I was playing well without it, I did not even try to use it. Moreover, by that time I was already playing middle buy-ins where the list of regular players was the same month after month, it being a small online network.

I support thoughtful playing and do not usually play more than four tables at once, and seldom did when I was just starting out as a serious poker player. I think that statistical value analysis would have devoured a large dose of the time I assigned to roll (by the way, there was no time bank button on Entraction network) and instead of training in poker by logical thinking, my brain would be busy with comparing numbers.

In addition, in my opinion, statistics have a bad habit of persuading you to do things that are not really worth doing (especially if you are a newbie). For example, you bet two streets with a low set, your opponent plays check/call, the flush draw is closed on the river and you are donkbet into. Probably your experience will tell you to fold, but your eye stops on the value river with aggression factor four, and you pay the flush saying “I’ll never fold against this” or “What do you have there?”

You may well have thought that I am against statistics as a matter of principle - not quite. In large networks, where new players appear and old ones disappear frequently, statistics are more of a necessity.

If I have the opportunity, I will put VPIP, PFR, 3-bet, fold for 3-bet, aggression factor and WTS all on the table. Sometimes I display more values thinking that they will be useful (for example, continuation bet) but in truth I always forget about the new values while playing, and don't use them. If I don't have the opportunity to play with statistics, it does not upset me, and I simply continue playing without them.

Generally speaking, I need statistics to visually divide the opponents into plain fish and regulars from the beginning; and, by the way, using statistics you cannot say a lot about how strong regular players are. The second function of statistics is to prevent yourself from getting into obviously marginal situations, especially before the flop. For example, not to 3-bet lightly with 67s into a guy who never gives up to 3-bets, or not to call a 3-bet with AQ on button against blinds with a 3-bet ratio of two, putting him on re-steal.

Present-day programs like Hold'em Manager or Poker Tracker 3 introduced a large amount of new statistical features. I will not evaluate their potential use, as for all their diversity only 3-bet and Fold for 3-bet appeal to me.

I see all the other numerous values either “over-specialized” or too connected with meta-games or requiring a respectable distance. For example, raise vs continuation bet OOP. It's obvious that if this value is equal to 20 it is way too much, i.e. the opponent cannot have a good hand every fifth time, although this value is strongly connected with his call pre-flop raise. If an opponent against you out of position will call only pocket pairs, in twelve percent of cases there will still be a set, and in eight percent of cases some kind of bluff or, rather, a semi-bluff with a draw hand. Rolling against such a player with TPTK or over-pair and further playing for the stack will be a big mistake. At the same time it will be correct against the guy who calls against you out of position with a significant amount of hands, protecting them from the steal.

Next, what distance is necessary for this value to have anything in common with reality? Obviously, the fewer amount of hands a player calls, the larger amount of data on him you should have.

Finally, are you sure that this opponent will play check/raise correctly against you in 20 percent of such situations? The difference between 15 percent against you and 25 percent against other players may play the devil with your stack.

What do we have as takeaway? If you realize all this on an instinctive level, and thinking over this information does not eat into your valuable time allotted for decision-making, or your general level of playing neutralizes the win-rate loss because of these imprecisions, use it for good.

If not, BBs are disappearing from your win-rate very slowly but surely, every time you make a decision based on the cold hard facts of statistics. For the sake of experimentation, try to get all these colorful numbers away from your screen and go and play a couple of thousand hands instead. I am almost sure that you at least will not do worse than normal, and if there is no difference, why should you overload your brain with unnecessary information? Let the free neurons work on logical poker thinking and memory - anyway, you will not lose anything and you're always able to get back to your beloved numbers.

Personally, I get much more out of my own notes or “float” on my opponent and the two previous playing rounds than some bet turn versus missed bet value equal to eighty percent for 357 hands. When you are playing four tables and are focused on them, it is not too much of a complication to remember which opponent never folds to your continuation bet when in position. By the way, make sure you are playing four tables, and not four tables and Skype and forums and an episode of the Simpsons.

I will now give you one more point that's worth thinking about. If you can make right decisions purely based on statistics, why has the world of online poker still not been taken over by bots? Who better than a computer to multiply all these percentages with sufficient accuracy?

Why do you need all these values if you cannot trust them? I suspect that they can be used only for your own playing analysis. For example, you can compare them with the similar values on your limit of the successful players in your room - however it is only an assumption, as I have not done so.

That last paragraph raises one more issue: some students aim to match their values with others - the ones they saw a “cool” guy on the forum had.

Reality shows that different players can stay in the black by playing in totally different ways, as long as you don't make logical mistakes. It's not important whether you play, for example, 31/25 (VPIP/PFR) or 16/14, you can win either way. It's difficult to do in first instance as there are lots of complicated situations that will appear, and in the second case you are easily read. That is why at the moment I recommend that all my students to play a middling style - something like 21/17. But this does not mean at all that it is a necessary requirement to beat the field. It's just some kind of “balance” between “unreadability” and “high infrequency of complex decisions”.

Why am I saying all this? Your statistics are not worth dwelling on, if you, like most of my students, possess some large logical “holes” in your understanding of the game. Don't search for the black cat in an unlit room - all the more so because it is not there. It's better you spend this time watching my videos.

“Won without showdown” or, more popularly, “red line” generates a special buzz. In my opinion there is no big difference as to where this line goes – up, down or parallel to equator - the main thing is where the profit line is heading.

Say one player selects a table but does not play if there is not at least one fish present. He plays as straight as possible against fish, without over-bluff attempts, which in general is correct. In this case the red line will naturally head downhill, as weak hands will be folded and strong hands will often go to a showdown. Another player on this limit plays nine tables with the same pre-flop statistics and does not bother too much about which tables he selects. With relative success he uses squeezes, floats, light 4-bets, double and triple barrels, bluff bets on rivers etc. against regulars. His “red” is accordingly heading, or trying to head upwards; at the same time the loss on showdown decreases when he misses with the bluff.

What is the point for the first player trying to lead his “red” to the same point where the second has, if they play against different types of opponents? Not long ago nobody ever worried where their “red line” was heading to as there was no place to check it, and now the profit graph is not considered a graph anymore if there is no “red line” upon it.

To be fair, I will say that you can - on-the-fly - calculate some systematic holes by the statistical values in your play. For example, a low fold to a 3-bet together with high VPIP will immediately say that you will find a systematical leak of money in 3-bet pots.

The main message I am trying to get to you to understand in all I have written is this: trust yourself, your perception, your memory, and your logical conclusions more than cold and hard statistical facts – and your probability of being “lucky” will be equal to 88.73 percent for 1965 hands.