THE BOOK--Playing The Percentages In Baseball

Yahoo lets you choose positional roster requirements and allows for LF/CF/RF to be separate positions.

It’s blocked here at work.  Can you cut/paste the list of available scoring categories?

From what I could gather at http://help.yahoo.com/l/us/yahoo/baseballplus/rules/rules-05.html

Default Point Categories:
Point Modifiers, Hitters
Runs Scored = 2 pts.,
Hits = 0.5 pts.,
Home Runs = 4 pts.,
Runs Batted In = 2 pts.,
Stolen Bases = 2 pts.

Point Modifiers, Pitchers
Innings Pitched = 1 pt.,
Wins = 5 pts.,
Saves = 5 pts.,
Earned Runs = -0.5 pts.,
Strikeouts = 2 pts.

From http://help.yahoo.com/l/us/yahoo/baseballplus/custom/configuring/baseballp-31.html

Allowed Categories:
*Batting*
Games Played;
Games Started;
Batting Average (Rotisserie, Head-to-Head—Standard, One Win);
Slugging Percentage (Rotisserie, Head-to-Head—Standard, One Win);
On-Base Percentage (Rotisserie, Head-to-Head—Standard, One Win);
On-base + Slugging Percentage (Rotisserie, Head-to-Head—Standard, One Win);
Fielding Percentage (Rotisserie, Head-to-Head—Standard, One Win);
At Bats;
Runs Scored;
Hits;
Singles;
Doubles;
Triples;
Home Runs;
Runs Batted In;
Sacrifice Hits;
Sacrifice Flys;
Stolen Bases;
Caught Stealing;
Walks;
Intentional Walks;
Strikeouts;
Ground Into Double Play;
Total Bases;
Putouts;
Assists;
Errors

*Pitching*
Pitching Appearances;
Games Started;
Earned Run Average (Rotisserie, Head-to-Head—Standard, One Win);
WHIP Ratio (Rotisserie, Head-to-Head—Standard, One Win);
Strikeouts per Nine Innings (Rotisserie, Head-to-Head—Standard, One Win);
Strikeout to Walk Ratio (Rotisserie, Head-to-Head—Standard, One Win);
Wins;
Losses;
Complete Games;
Shutouts;
Saves;
Outs;
Hits;
Total Batters Faced;
Runs;
Earned Runs;
Home Runs;
Walks;
Intentional Walks;
Hit Batters;
Strikeouts;
Wild Pitches;
Balks;
Stolen Bases Allowed;
Batters Grounded Into Double Plays;
Save Chances;
Holds;
Total Bases Allowed;
Innings Pitched

I am not sure if there is a limit to number of categories.  Hope that helps.

That’s good.  Ok, then:
SV +2
Holds +1

If someone wants to run that through last year’s stats.  Ideally, you want the scale as follows:

average hitter in 650 PA
= average starter in 171 IP
= average closer in 72 IP
= single best setup guy in the league in 90 IP

All of these guys are worth 2 WAR, and would get around 8MM in free agent money (though the setup guy is a bit of a stretch if only that he’d graduate to a closer soon enough).

The “last player drafted” in the starter, relief and hitting categories should all be worth close to zero points.

If you’ve got that lined up, then the top-end scale should take care of itself.  But, it should be fairly reasonable.

Few things:

1) Yes, we can use decimals. And, you’re right, Yahoo lets us assign values to counting stats. I’m not sure what you meant regarding fielding positions, but each player is assigned a position. We don’t use fielding statistics, if that’s what you were asking.

2) We don’t use ER - just the components: 1B allowed, 2B allowed, etc. (including WP and HBP).

3) We don’t use saves (or wins for that matter).

5) In short, our categories are:

Hitters: AB, K, BB, 1B, 2B, 3B, HR, HBP, SF, GIDP, SB, CS
Pitchers: IP, 1B, 2B, 3B and HR allowed, K, BB, HBP, WP, GIDP

I may be forgetting some, but I don’t think so.

It’s not really fielding independent (because of the hit categories) but it’s also not really ERA-based. Do you know if there are weights available for those specific measures?

For fielding, I was asking if you can assign points for fielding games (or putouts, assists).  For example, can I give a certain number of points for putouts, assists, or innings played (specifically, by position)?

Is there a reason that you are precluding ER, or necessarily including the other categories?  That is, why are you married to those categories?

You can’t assign position-specific points for defensive stats on Yahoo. I don’t think there’s any good way to include defense on Yahoo.

Tango, great system you came up with. I’ve been using something relatively similar (but not quite as “accurate") since I started playing fantasy in the mid-90s.

I’m surprised that nobody’s tried to popularize a fantasy system like this. There seem to be tons of sabermetrically-inclined people who still play 4x4 or 5x5 even though those systems don’t correlate very well with reality. Why?

Generally, we’ve tried to get away from anything team-dependent, like RBIs, saves, etc.

We decided, last year, to move away from ERs because we wanted to boil the randomness out. Ideally, we’d just assign values for each batted-ball type allowed but Yahoo, of course, doesn’t let you get that granular. So, we figured the next best thing would be to use 1B allowed, 2B allowed, etc. Of course, those still depend on defense but they’re one step closer to batted-ball type than ERA (i.e., I think that, year-to-year, 1B allowed, etc. correlates better than does ERA).

In sum, we’re trying to get down to the most basic batting events possible since, the more basic you get, the less noise. We want our league to be about each manager’s predictive abilities and, to make that really work, we are trying to get rid of chance as much as possible.

And, finally, no I don’t think we can do anything with defensive stats on Yahoo. The best you can do is assign points for errors, but we think that would be nuts so we’ve stayed away all fielding stats.

If we had Putouts and assists, I would have been happy.  Ok, let’s move on.

***

You have to include saves.  While it is heavily team (and manager) dependent, you absolutely need to value Mariano Rivera *somehow*.

***

Ok, let me redo the pitching ones to include 1b,2b,3b, and exclude ER.

I don’t think we need saves to make Rivera come out strong. In fact, last year, using a modified version of what we’re now talking about, relievers went like hotcakes. The reason being that even non-headline guys like Broxton or George Sherrill generated far higher points/IP ratios than did the best starters - this is due to their very high K/IP rates. So players had an incentive to shift innings away from starters and toward relievers - all this without using saves.

Our league is set up so that you can start 8 pitchers (starters or relievers) per day and the max number of innings you can pitch (for the season) is something like 1300 or 1400. So, even though starters are worse per inning, you’d fall way behind on volume were you to play only relievers. On the other hand, if you played only starters, you simply wouldn’t generate enough points.

It worked out really well except (1) we wanted to improve the accuracy of our values and (2) we set the IP limit a little too low so, in fact, relievers were a little overvalued.

Here are the new weights:
IP 2
H -0.5
BB -2
K 1
ER 0
HR -3
2b -0.5
3b -0.5

Note that 2B and 3B are given the same weight.  After all, what is a triple?  It’s hit alot by speedy players, and by hitters to RF.  Given the very lack of triples allowed to begin with, I’d be happier calling a triple and a double the same thing.

So, try these out, and see what it gives you.

I don’t understand the K/IP you are talking about.  Are you doing a pure “points”, where you give points to counting stats, and that’s your total?

Or, are you doing a 5x5 type of structure, where your rankings in various categories (counting and/or rates) determines your points?

Sorry, what I meant was that, in real life, relief pitchers strike batters out at a higher per inning rate than do starters. Because of this phenomena, relief pitchers in our league accrue points at a higher per inning rate than do starters.

That’s fine that they accrue at a higher per inning rate.  But, what matters is how much they accrue per GAME.  And, if Wagner is getting 4 strikeouts per game, while Jeff Weaver is getting 5....

In any case, as I have laid out the points, the reliever gets killed without the leverage categories (saves, holds).

In your case, it’s almost certain that last year you gave very little points to the IP category (and you probably had plenty of starters with negative points).  That’s the only way you could have relievers turn out so well.

Tango,

You’re sort of missing the point. There’s a limit to the number of innings you can have your pitchers throw, so the per inning rate DOES matter. You’d rather have a staff that pitches 1,300 innings (if that’s the maximum) and gets 1,000 strikeouts than one that pitches 1,700 innings and gets 1,200.

Tango,
Yahoo does allow assigning points for putouts, assists, and errors, but doesn’t let you change the weightings by position. How would you incorporate them into this lwts scheme?

David: ok, I think I’m getting it.  The top 210 pitchers (presumed 10 pitchers x 21 teams) using my points system totalled 24209 forecasted IP in 2007.  That would make the average per team as 1153 IP.

So, even if the cap is 1400 IP, you simply won’t be in a position to reach those levels.

I think what is really happening with FantasyGuru is that the way they allocate points, it’s much closer to a “league average” than the “replacement level” that I am basing my points on.  From that standpoint, relievers will look pretty good, simply because they won’t rack up all those negative points that his league allows for their starters. 

To put it plainly, it’s very difficult in my points scheme for a pitcher to get negative points, and it’s probably really easy in his.  And, the more a starter pitches, the more negative points he’ll get.

***

Robert: ah, well that’s good.  Partly.  If it was only 2b,ss,3b, and OF, then the total number of PO+A would do a decent job as a stand-in for fielding+position.  But, 1B get tons of “chances” by virtue of their putouts, and same for catchers and putting out strikeouts.

The above points, given to the 2007 guys I forecasted, gives 59% of the points to the nonpitchers and 41% to the pitchers.  That’s an excellent split.

Pujols led all forecasts with 640 points in 2007.  Santana led pitchers with 440 points.

Closers really get the short-shrift, even with 2 points per save.  You have to give 4 or 5 points to do them justice.

Tom,

Is it me or do you have BB, SB, and CS weighted too high for hitters?

On your “linear weights by men on base” chart, it seems the average stolen base is worth about 1/7th of what a HR is worth. But here you have it worth 1/4th.

-Sunny

It’s you!

A HR gets you:
-2 AB
+ 6 H
+ 8 HR

That’s 12 pts for a HR.

I have 2 for SB and 3 for BB.  If you divide everything by 100, you get the win value.  Sneaky right?

So, the “200 points” for the avergae player, if you divide by 100, is 2 WAR.

As for the IP issue, let’s try to explain that better.

Each fantasy team gets 8 pitching slots. Every day, 8 pitchers can pitch for that team. At 26 weeks * 7 you get 1456 total pitcher games (assuming that you have 8 pitchers able to start every day which is unlikely). That means that even if each one of your pitchers only pitches once per day, you’ll be over the innings limitation.

Obviously having all starters will rapidly eat up the innings, while posting poorer points per inning. However, your counting stats will suffer if you don’t pitch the maximum, so picking stud relievers will leave you behind even if your rate is higher.

Mariano Rivera would be a decent player because he would post great points over 70 of your innings.

Personally, I hate IP maximums.  Just set the transactions to weekly or assign some RP slots if you want to prevent the “all starters K’s and W’s” strategy.

Sal/21: perhaps you can explain further.  Are you suggesting that if you have Santana and Bedard pitching the same day, and they each pitch 7 innings, then you only get nine-fourteenths of the total points accrued that day?

would giving points for pitching appearances be a way to help give relievers credit without having to use saves?

I’m not sure about that.  It might proffer up the mopup guy too much.  The ace reliever is about 100 points behind the #3 starter, which is right around where you’d like to have him.  With the reliever having say 30 more games than the starter, you have to give 3 points per game in order to align the ace reliever to the #3 starter.  However, the bottom-end reliever ends up getting a ton of points as well.  The bottom-end reliever was already properly setup in the original system.  You have to find a way to bring up the ace, without bringing up the mopup guy, and Saves is the easiest thing.

Right on...thanks for explaining that so clearly, I wish I was smart enough to realize those things!

just out of curiosity, regarding pitching weights, are er really 0?
“IP 2
H -0.5
BB -2
K 1
ER 0
HR -3
2b -0.5
3b -0.5”

I was surprised that Harang had more points than Johan, though he did pitch a ton of innings…

post 11 was in response to post 9

Just thought some of you guys might be interested to hear that we (a few sports analytics clubs) will be holding a fantasy competition using these fantasy settings.  Click my name to see the league site, though not much has been updated yet.

I ran a points league last year with the following weights:

Hitters

AB -.15
H .78
2B .24
3B .56
HR .96
SB .18
CS -.5
BB .45
HBP .45
GIDP -.54

Pitchers
IP 1.35
SV .12
H -.56
HR -.88
BB -.33
HBP -.33

It worked very well.  With a minute’s worth of arithmetic you could calculate a theoretical won-lost record.  I would resuscitate it this year but I don’t have the time.

I have an open free yahoo league using the honorable Tom Tango’s weights from this thread, and am hoping some baseball fans (who are more sabermetricly oriented that the patrons at my local watering holes) will want to join. All are most welcome. Email me for info on how to join the league

http://baseball.fantasysports.yahoo.com/b1?lid=303517
password: tomtango