20-80 pt scale questions

by bk11 on Jan 28, 2008 1:53 AM EST Comment 15 comments

hey all,

i have a few questions on 20-80 scale ratings and would appreciate it if any of you could help further my understanding on this.
i know it's based on standard deviations, 50 +/- 3SD

so my first question is: i often hear a player has a plus pitch, or a plus plus pitch. what score would a plus pitch rate? how about a plus plus?

second question, since the pt scale is based on standard deviations then it should be possible (but rare) for a person to have tools or a pitch rated higher than 80 (1sd=68%, 2sd=95%, 3sd=99.8% if memory serves me right). who would you say has a tool or pitch higher than 80?

thanks everyone!

0 recs | Comment 15 comments

Story-email Email Printer Print

Comments

I think

plus is considered 60, and plus-plus is 70.

http://mvn.com/milb-yankees/

by lemonjello on Jan 28, 2008 2:07 AM EST reply actions 0 recs

You lost me

but I'm interested.

Where does it say that the 20-80 scale for prospect tools is based on standard deviations? You can't actually measure the player's "tool" (well, not THAT one, anyway), so I'm not sure how you could assign a grade based on an SD.

Now, you could grade a whole bunch of prospects and then compute their mean and SD after the fact, but that doesn't seem to be what you're after here.

gogotabata: "I'm like the biggest Walden fan around here (adult division)..."

by siddfynch on Jan 28, 2008 2:09 AM EST reply actions 0 recs

hmm

i don't remember where it is said, but i remember it being mentioned. i always wondered why 80 was the max score, and it made sense if avg was 50 and sd=10.
so a player with 80 power has more power than 99.8% of mlb baseball players. because there's still a 0.2% left it's possible for a player to have an 85 for power, etc...

by bk11 on Jan 28, 2008 2:15 AM EST up reply actions 0 recs

formal distribution

conceptually, i guess it's not a bad way to think of it -- that the distribution will turn into something of a bell curve -- but, if you thought about it, you'd realize why it's obviously not actually going to fit this distribution.

in order for something to meet a bell curve, the data would have to be grouped and compared, then assigned numbers AFTER it was all together. in reality, scouts assign numbers on the spot.

sure -- you could tell a scout "only assign 2% of players a 70 or above" or something like that, but: 1) i don't believe humans have the capacity to follow a guideline like that (which is why you'd need to group the data BEFORE assigning the grades), and 2) you wouldn't necessarily want every scout to keep to a curve himself, since some scouts will consistently be looking at better players than others.

anyway, i'm pretty positive the numbers don't signify any type of distribution. though it would be interesting to see what distribution of grades certain scouts did end up giving.

by bleedjaxblue on Jan 28, 2008 2:31 AM EST up reply actions 0 recs

There cannot be a standard deviation.

hypothetically speaking the SD 10. Average=50

(because it is in incremental jumps of 10 anyway)

But like you said, you cannot actually measure the player's tool, and there's some error associated with the score.

Say you developed a test in order to measure for this tool, BA as a measure for "Hitting for average"

X ( their test score) (For example .300)
T= (= their true ability, maybe they are only a .280 hitter)
E= (= error in your measure, in this case it would be .2)

Unfortunately you break a bunch of assumptions in classical test theory in constructing this measure. C'est la vie.

by cubsfan2883 on Jan 28, 2008 3:21 PM EST up reply actions 0 recs

I think

30 = below average
40 = fringe
50 = average
60 = plus
70 = plus plus
80 = "off the charts"

by phuturephillies on Jan 28, 2008 2:20 AM EST reply actions 0 recs

Quote from BA handbook 2007

When grading a player's tools, scouts use a standard 20-80 scale. When you read that a pitcher throws an above-average slider, it can be interpreted as a 60 pitch, or a plus pitch. Plus-plus is 70, or well-above-average, and so on. Scouts don't throw 80s around very freely. Here's what each grade means:

80 - Outstanding [plus-plus-plus?]
70 - Well-above average [plus-plus]
60 - Above average [plus]
50 - Major league average
40 - Below-average
30 - Well-below-average
20 - Poor

I think the original usage of the scientific 20-80 scale may have been based on standard deviations, but that's never been the case with scouting (at least to the best of my knowledge).

by aCone419 on Jan 28, 2008 2:35 AM EST reply actions 0 recs

LOL

"but that's never been the case with scouting (at least to the best of my knowledge)."

Bingo!

I realize we are a community of statheads and roto geeks but if you think scouts give two strokes and a squirt about "standard deviation" you are losing touch.

Baseball Instructor - www.frozenropes.com

by HuskerBob on Jan 28, 2008 2:35 PM EST up reply actions 0 recs

Standard deviations

Pretty sure it doesn;t have anything to do with that, because if you think about it, scouts see a LOT more guys with scores below 50 than scores above fifty. It isn't like 50 is the high point on the bell curve, because major league average is far better than what most professional ballplayers would rate.

by fargocraig1971 on Jan 28, 2008 10:05 AM EST reply actions 0 recs

50

50 means average for a major leaguer. It's standard deviations from an average major leaguer. Minor leaguers aren't a part of the sample from which the standard deviations are derived.

Also, with respect to prospects, usually there are two scales that are graded - current and projected. Not that the site has been active for a couple of years, but calleaguers.com had good examples of this in its scouting reports.

Vogt early, Vogt often.

by Brickhaus on Jan 28, 2008 11:20 AM EST up reply actions 0 recs

I stumbled across the calleague site

a couple months back. I understand that the guy is working for the "A"s now so we won't get that kind of analysis from him anymore. Do you know anyone else who is doing that kind of work that I/we can access?

For laughs, everyone else guess the prospect .."Power arm was best on a position player in league gives him a legit chance on bump if bat doesn't work out."

by Yoda on Jan 28, 2008 1:43 PM EST up reply actions 0 recs

Its based off a Normal Distribution

If you have ever taken a probability course everything (pretty much) ends up being a normal distribution. And thus in a normal distribution almost nothing should fall outside of 3 standard deviations. Supposedly Branch Rickey came up with it:

It's generally accepted that the legendary Branch Rickey devised the universally used 20-80 scouting scale. The actual origin of the scale is a bit of a mystery, but some believe it's based on a scientific scale used in some fields that revolves around a median (in this case 50), and three standard deviations above (60, 70, 80) and below (40, 30, 20). What does that have to do with organizational rankings? Not much, unless you're a scouting dork like me. As opposed to being completely subjective here, I went about the task of trying to actually quantify the rankings.

by Kanst42 on Jan 29, 2008 11:05 AM EST up reply actions 0 recs

Actually

I have read the normal distribution thing and I think it is right on. Gotta remember it's a normal distribution for MLB players...so most minor leaguers will grade below.

by Dfarth on Jan 28, 2008 10:25 AM EST reply actions 0 recs

I think you all are trying to read...

to much into this as usual. I really don't think using standard deviation on tools which are largely subjective is the way the system works. It doesn't even make sense to me to be honest. McKamey put bench marks in his book for what he felt each level was at the MLB level.

It's my understanding when grading a prospect, you rate their tools against what the average MLB player does. Then you use OFP (overall future projection) to get a general idea of what he might end up at.

Here is a link I was discussing it with a scout on the Phillies board. I probably shouldn't have put McKamey's stuff up there but I can't edit them out now.

http://www.forums.mlb.com/n/pfx/forum.aspx?webtag=ml-phillies&nav=messages&msg=60367.39

by jfish26101 on Jan 28, 2008 3:05 PM EST reply actions 0 recs

Your thinking about it wrong

They arent rating the tools on a number and then scaling it based off a normal distribution that wouldnt work. The 20-80 number system is used often to grade things since ideally most samples will fall onto a standard normal. Most likely Branch Rickey (or whoever did it first) say something else graded on that scale and decided he liked it. At this point it really doesnt have a pure statistical meaning any more. Its more so that an 80 tool is a very very good best in majors tool, a 70 is very good, a 60 is above-average-plus a 50 is average and so on and so forth.

by Kanst42 on Jan 29, 2008 11:07 AM EST up reply actions 0 recs