When to move up levels in poker.

Your poker bankroll should be a prized tool in your growth in online poker, but for new players there is often the eagerness to move up a level and most want to do it before it's time. Here is a player in the forum however who has been diligently battling it out in the .02/.05 NL hold'em cash games.

As you can see his graph is definitely going in the right direction, but he could also be experiencing a good run with a relatively mediocre data sampling. At least he had the sense to ask some more experienced players in the forum as to how to know when to move up a level. Here is his graph and question:

Q: How do you decide to move up to the next level in cash games? I have about $140 playing .2/.5, do I have the funds to play .5/.10? I have been running really well at .2/.5, making just slightly less than $100 in 7,500 hands.

JJSquared tells it like it is.

You basically need two things and that is the funds to support the stake AND an edge.

Moving from 5nl to 10nl is going to be a very small change in opponent skill. You will probably retain your edge. I don't think 14 buy ins is enough funds to support $10nl though. It will be scared money. I think you should grind it out a little longer at 5nl for two reasons. To earn more money to take a shot at 10nl and to also gain more experience.

I think I have oversimplified a pretty complex topic. When moving up through the micros, you need to try and get to the next stake as quickly as possible to increase WR and also escape rake, but you can't rush it. If you spend 25% of your BR trying to take a shot at a level, and have to drop down to a stake that is going to earn you $4/hour.... it is going to take you a long time to move up and also damage your ego.

Phaedrus75 gives the mathematical explanation.

There are Two ways to do it.

1 - The aggressive way.

When you have the required BR for the next level, you move up to the next level. The required BR again depends on how aggressively you want to move up. Sun137 recommends a minimum of 15 Buy-ins for FR Micro stakes NLHE. I think you would be better working to 30 BIs for FR and 50 for SH, particularly when you aren't too sure where you stand at the moment.

If you are playing $5 NL, then you need $250 by my measure. Moving up to $10 NL you would need $500 before moving up.

For the sake of the record, once you hit NL50 and above, you really need to respect the fact that people mostly don't just give you their stacks. Because the games are tougher, a bigger BR is required. Both Ed Miller and Dusty Schmidt have published books recommending a minimum BR of 100 Buy Ins. I completely agree with them on that.

2. - The safe(er) way

Use a program like HEM to compute your standard deviation. If you don't know what your SD is then you could adopt about 90bb/100 as a reasonable estimate.... NLHE tends to run between 80-90bb/10 SD until you hit mega high volume.

A rough 95% confidence interval for your true win rate can be calculated as

SD / SQRT (Hands / 100) +/- WR

So using your PTR stats, you have played 5595 hands at NL5 for an actual WR of 11.62bb/100 (very nice indeed).

So with 95% confidence, and assuming you are running a best case scenario Standard Deviation of 80bb/100, your true WR is:

11.6bb +/- 90 / (SQRT 56) bb/100
= 11.6 +/- 12 bb/100

So based on the sample size presently available, we can say with 95% confidence, that your true Win Rate at NL5 is between -0.4bb/100 and + 23.6 bb/100

Accordingly, we cannot yet say that you are beating NL5 with statistical reliability simply because the sample size is too small - although the early indications are as encouraging as they possibly could be!

The top and bottom of the above numbers are extremely far apart, and again, this is due to small sample size.

To give you an idea what happens with bigger numbers, my WR at NL50 over 331,903 hands is 4.34bb/100 with a SD of 80.12

This means than my true WR (95% certainty) is between +2.95 and + 5.73 bb/100

Whilst I don't have any particular logic to it, I tend to think that a lower range of 3bb/100 at 95% confidence interval is a good target to hit before moving up.