Am I running hot/cold? Quantifying results

[Derek in NYC]

A lot of 2p2ers post stats questions and ask whether they are running hot/cold. Nobody ever answers these questions with empirical data. Instead, we look at VPIP, PFR, AF stats, or BB/100 stats, and the sample size, and make some assumptions about whether the person's stats are within the expected range for a TAG. For instance, if the VPIP is 20%, the PFR is 10%, and the post-flop AF numbers are all 2.0 or greater, in a full ring game, we'd all probably say, "You're playing fine." Now if the BB/100 stat was +4.2 over 25,000 hands, we'd add, hey your win rate is too high, so dont expect it to continue.

But can we quantify the hot/cold question differently? Suppose instead, for the same 25,000 hands, we knew that the hero had received AA 270 times. Well statistically, he should get AA once in 221 hands, so since 25000/221 = 113, we can say for certain that he has received AA more often than we would expect. Moreover (although I cannot do so myself), we should be able to determine the degree to which hero's AA incidence exceeds the statistically expected value.

Now suppose that we created a series of metrics. For example: prevalance of big pairs, percent of pairs that flopped a set, percent of suited cards that eventually went on to make a flush, etc. Would it be possible to calculate a variety of these metrics and empirically state that a person is running hot or cold?

[uuDevil]

I think EV Tester (http://www.evtester.com/) goes partway towards what you're looking for. It's a computer program that takes data from Party HH's and compiles a few stats and evaluates your "luck."

For instance, it will tell you how often your pairs flopped sets and compare this to the expected frequency and also look at how good the starting hands you recieved were relative to a random distribution of hands (using Pokerroom's EV data). But it doesn't evaluate things like your exceptional lucky streak of sets getting beat by straights, flushes, and bigger sets.

[Orpheus]

That's a fine idea, but your dealt hands and "hand if shown down" are only part of the the picture, especially if you think you may be running hot/cold. Every metric I can think of fails to take into account most of your opponents hands. By the time you've factored in all the hands at the table, you're pretty close to a hand history.

There is a saying in applied mathematics (esp. cellular automata and chaos theory): "The universe is the most succinct possible complete description of itself." Mathematical models must be simplifications.

When I wonder if I'm running hot/cold, I look at the PT %ages that address my results, rather than *how* I play: W$SF, W$SD etc. If my VP$IP, aggr at each stage, and other "decision" stats are unchanged, then it seems likely that the thing that is changing isn't my decision, but the cards.

Of course, that's not an entirely satisfactory tool. A good or bad run of cards can change your short term stats, even if you're playing exactly the same strategy. Most players have a super-tight rock mode that we fall into during long stretched of dry cards. I find it very helpful, but boy do my stats change!

Some NL say that this is a suboptimal strategy, but I'm not good enough to bet and win for long without some help from the cards, and sometimes the cards just aren't very helpful. Is a rock strategy a cause of running cold or an effect? Are hot spell a result of increased aggression, oir is it just easy to be aggressive when the cards are hot?

My guess: if your W$SD is 80%, it's probably not because you've magically figured out how to make strong hands fold and weak hands come out to play (simultaneously!). Sometimes it signals that you are being too conservative-- but usually not when your bankroll is climbing like a little monkey who just stole the big monkey's banana.

Well, anyway, that's my take on it. Unlike most players, I can't create quads and FHs by force of will, and my reads don't have the moral weightof a fat pope and the rigor of physics postdoc on viagra. (At least that's the impression they give when they're fuming over a "bad beat" that the rest of the table saw coming back at the flop.)

[AaronBrown]

I addition to Orpheus' excellent and witty reply, I would add that it makes sense to focus on other metrics. It doesn't make as much difference as you might think to get good versus bad hole cards. You can turn bad cards into good just by playing tighter, which only costs you some antes or blinds.

The important luck factor is whether your good hands come when someone else has a slightly worse hand (you get rich fast) or slightly better hand (you get poor just as fast).

There are 1,352 possible starting hands in hold'em (the 178 you usually see treats all suits as equivalent; that's valid before the flop but not afterwards). When the board comes down, we can rank these hands from 1 (best) to 1,352 (worst). There will be some ties. It's not quite that simple, because some hands make others impossible. Hand 8 might still be the nuts because it contains cards that make 1 to 7 impossible.

You don't care about the average rank of your starting hand, because 90% of the time (more or less, depending on your aggressiveness and the table) you're not in a showdown. So only your top hands matter. If you can get a few 1's, you don't care if every other hand is 1,352. It doesn't cost any more to fold a 1,352 as a 100. It also doesn't matter that much how many top hands you get, as I said, you can just fold until you get the hands you want.

I think the most informative metric would be to see how many times your 30 gets beat by 1-29, versus beating 31-59. Beating 60 or higher is unlikely to make you rich because people won't bet much with hands like that. Those two counts should be equal in the long run, but even one more or less can determine whether you are ahead or behind for the session.