*** VD 10 Commentary Thread ***

[heyelander]

not sure which iteration I'm using, but the one I have looks fine. 1s in 90s, 00s, 10s for Rivera.

[johnnya24]

No, it's something in the way you imported the data into the spreadsheet. If you look in the GDrive folder (again, that's http://tinyurl.com/neutralmilkhotelsucks) and look in the CH.csv and CP.csv files, those columns are correct and don't match the spreadsheet version.

I'll have a look again tomorrow. I don't see why it would only be the Decade data that is out of sync? It was all imported at once ... surely everything should be out of sync? Bizarre.

[The Feral Slasher]

Using the spreadsheet I created since we started doing these things 10+ years ago! Frankly, I'm surprised anyone who's done more than a handful of these doesn't have their own spreadsheets/rankings using the career and BY stats.

Heye told me that was cheating

[heyelander]

I'll have a look again tomorrow. I don't see why it would only be the Decade data that is out of sync? It was all imported at once ... surely everything should be out of sync?

since mine looks good, my guess is you botched a sort at some point, didn't realize it, added Frae's BY or something and saved it.

[heyelander]

Heye told me that was cheating

way to go revo...

[revo]

Instead of adding hundreds or thousands of lines to an old spreadsheet that has the wrong career stats and tons of missing best years, I just rolled my own calculations and rankings into the new sheet. I'm surprised that anyone who's done more than a handful of these would use an old spreadsheet to do that, since you're essentially ensuring that something will be wrong. I've always downloaded the new sheet after others (thanks to all you others out there!) have added the new years and updated career information and then calculated my rankings on the new sheet. I usually then use two spreadsheets per draft --- one for the "real" draft and one for projections.

Adding new career #s and new BYs to a career/BY spreadsheet took about a solid 30-40 minutes, give or take a bathroom break.

[heyelander]

Adding new career #s and new BYs to a career/BY spreadsheet took about a solid 30-40 minutes, give or take a bathroom break.

cripes, that's way more time than I put into thinking about my entire draft.

[heyelander]

Inputting my rankings into a new sheet takes about 3 minutes, tops.

that's more like it...

but what are these ranking things you all keep talking about?

[revo]

Inputting my rankings into a new sheet takes about 3 minutes, tops. I literally have to copy/paste 6 things and drag the formula and that's it. Plus, I don't have to identify which BYs are new, which career #s need updating, etc. In fact, I was using the VD 9 spreadsheet for this draft since I already had rankings in there but got annoyed at having to do the new BYs and update all the career #s every time a damn error popped up. So I put my rankings in the new sheet and that's when I noticed that all the decades stuff is wrong on the new one.

All you really have to do is filter for the last year and see who kept playing -- unless you have hundreds of players listed, which shouldn't happen if you have your own sheets.

And after all this time, all these gazillions of drafts, I still don't have a very good ranking system for pitchers. I just am not clever enough to come up with weighted averages, where a 2.50 ERA over 2000 IP is worth much more than a 2.50 ERA over 200 IP. I end up with really random relievers very near the top of my list.

You have to come up with a point system that weights IP, H, BB, & ER rather than ERA & WHIP.

[johnnya24]

And after all this time, all these gazillions of drafts, I still don't have a very good ranking system for pitchers. I just am not clever enough to come up with weighted averages, where a 2.50 ERA over 2000 IP is worth much more than a 2.50 ERA over 200 IP. I end up with really random relievers very near the top of my list.

I set a value for ERA and WHIP without any IP considerations. Then I add a multiplier column ... "1" for the estimated average number of IP (say 180 IP). If the IP is below the average I reduce the multiplier. If above, I increase the multiplier. I can't remember the precise details of how I worked things out.

It's not perfect, but it works OK.

So for instance:

Walter Johnson has a very low ERA and WHIP in his 1913 year.

ERA = 51
WHIP = 71

The IP is considerably higher than the average, so the multiplier is 1.8 (based on 346 IP)

So his final ERA and WHIP scores are:

ERA = 51x1.8= 91.8
WHIP = 71x1.8= 127.8

Mariano Rivera also has a very low ERA and WHIP for his 2008

ERA = 33
WHIP = 75

But his IP is much lower. His multiplier is 0.4

So his final ERA and WHIP scores are:

ERA = 33x0.4= 13.2
WHIP = 75x0.4= 30

George Bradley with his 548 IP has a 2.8 multiplier:

ERA = 45x2.8 = 126
WHIP = 33x2.8 = 92.4

By contrast:

Hardie Henderson's 1885 had 539 IP of garbage:

ERA = -72x2.8 = -201.6
WHIP = -60x2.8 = -168

Low Ratio, high IP guys get murdered in the rankings.

[DMT]

that's more like it...

but what are these ranking things you all keep talking about?

Exactly, VDing is an art not a science!

[Controller Jacobs]

For multiplicative scoring, I would suggest
1. Find the expected ratio of the below average (but not worst) teams.
For example, in vd2, an average career era would be around 2.92, a "below average" era would be around 2.94 (good enough to stay out of the bottom four)
2. Find the expected average ip per pitcher
In vd2, this average career ip would be around 2450 ip.

Then for a given pitcher, the scoring of their career era would be (cip/2450)*(2.92-cera)/([delta cera])
[delta cera] is a value you need to come up with - perhaps it is related to the difference between quartiles (e.g. in this case abs(2.92-2.94) = .02). I personally like to work the standard deviation in there.

[The Feral Slasher]

Exactly, VDing is an art not a science!

That would look great on a t shirt

[heyelander]

That would look great on a t shirt

here's the t-shirt I wore to the PONY (hacko, frae, joncarlos) draft a few years back

[heyelander]

http://www.vistaprint.com/studio/rev...LargerPage1048

very nice!