Level of looseness in cash games
Just thought I'd share a graph I put together from my database, showing how win rate in Big Bets per 100 hands varies with the VP$IP (voluntarily contributed to pot) level. VP$IP is a good measure of how loose or tight a player is, low being tight, high being loose. As a guide, 20% or below is tight, and 30% or above is loose. You might think tight players would do better, but that's not what I found.
This first graph I put together was all the data I had, but since I'd been playing at beginners' tables a lot (I only started playing a year ago) and these tend to distort the data (lots of bad, loose players), I've taken these out of the dataset and just included $25 buy-in NL tables and higher. I also eliminated players with VP$IP > 50%, as I'd consider this outside the normal range.
First graph is for all players I'd observed for 60 hands or more:
Light blue dots are the data points. The dark blue points are the regression line, i.e. the best fit to the data. Yellow and Magenta are 95% error bands.
You can see from this
a) There's a lot more variance in winnings for looser players (as you'd expect)
b) There's no correlation between tightness (VP$IP) and win rate (BB per 100 hands)
Also produced a graph for players where I had >120 hands observed. Similar story, no correlation - if anything, suggests a slightly higher average win rate for looser players.
The regression was completed using the Excel Analysis Toolpack and assumes a straight-line fit.
Conclusion: one style, loose or tight is no better in terms of win rates than the other. Playing loose allows you more scope for winning or losing large amounts, although this data cannot determine whether loose or tight play is correlated with variance in session winnings on an individual basis.