Anyone checked out Guesstimate?

[octothorp]

So this new spreadsheet tool has been making the rounds in the last day or two. It's still really Beta, but I think there's a lot of stuff in it that's interesting to stat enthusiasts.

The difference in this is that rather than dealing in absolute values, you put in range, it applies a bell curve to the values, running a Monte Carlo method on it, which means running a large number (5000) of simulations. I believe this is the sort of system that those stat sites that calculate probability of teams making the playoffs use, although this is on a much more basic level.

I'm actually not sure of the best way to make use of the tool, but I thought I'd put it out there and see if some of the stats junkies can make use of it. I did this simple shooting calculator, where you can put in a player's attempts and percentages.

http://getguesstimate.com/models/2996

These estimates are for DeRozan, and the distributions seem pretty good: his 95th percentile in the simulation is around 33, and the season so far it's around 34 in real life. His 5th percentile is around 16 in the simulation, and around 15 in real life. (Click on the arrows on the Points Per Game box to see details of the distribution).

It's only going to produce an accurate projection where a player's numbers mimic a bell-curve. So, like, not Bebe.

Feel free to put in any values you want here, you can change values on this spreadsheet, but not save your changes, I think.

[KHD]

So this new spreadsheet tool has been making the rounds in the last day or two. It's still really Beta, but I think there's a lot of stuff in it that's interesting to stat enthusiasts.

The difference in this is that rather than dealing in absolute values, you put in range, it applies a bell curve to the values, running a Monte Carlo method on it, which means running a large number (5000) of simulations. I believe this is the sort of system that those stat sites that calculate probability of teams making the playoffs use, although this is on a much more basic level.

I'm actually not sure of the best way to make use of the tool, but I thought I'd put it out there and see if some of the stats junkies can make use of it. I did this simple shooting calculator, where you can put in a player's attempts and percentages.

http://getguesstimate.com/models/2996

These estimates are for DeRozan, and the distributions seem pretty good: his 95th percentile in the simulation is around 33, and the season so far it's around 34 in real life. His 5th percentile is around 16 in the simulation, and around 15 in real life. (Click on the arrows on the Points Per Game box to see details of the distribution).

It's only going to produce an accurate projection where a player's numbers mimic a bell-curve. So, like, not Bebe.

Feel free to put in any values you want here, you can change values on this spreadsheet, but not save your changes, I think.

where are you getting your ranges from?

Monte Carlo is great (in my line of work we use it a lot) but as always the outputs are only as good as the inputs (GI-GO, as they say). Not to say your values are bad just a general word of warning.

For interest's sake you can actually put together your own monte-carlo sims fairly easily using things like excel (here), but this site looks like it could be a compact way to do it which is kinda nice.

[octothorp]

I got my ranges by starting with his per game averages, and then looking at game-logs to try and come up with ranges that covered most games, and kept the per game average about the same.