Do you know if the SGP calculations you’ve done for your league are accurate?

Are you concerned that your home run SGP denominator is 8.87 and Larry Schechter’s book Winning Fantasy Baseball suggests using 5.93 for a 12-team league? Or that your RBI calculation shows 22.31 and the 12-team NFBC history I just calculated shows 19.55?

What does this all mean? Will your rankings be accurate? How can they be when your denominators seem significantly different than those you see elsewhere?

## Calm Down, These Numbers Are More Consistent Than You Realize

I know. You’re wondering how on Earth I can say that. How can a HR denominator of 8.87 be consistent with one of 5.93?

To be honest, I’ve had those same fears about SGP. I feel so scientific and strategic by calculating SGP. And then I look at my denominators in comparison to what I see published elsewhere and that confidence evaporates and is replaced with doubt.

In this post I’m going to share an important realization I just had about SGP (yep, I’m still learning too), show you how to properly compare your SGP denominators to different resources, and demonstrate that the dollar values calculated by different sets of denominators are more similar than you would believe. When we’re done here, I think we’ll all feel a lot more comfortable about things.

## Story Time

This story starts with me calculating the SGP for the last three seasons of NFBC leagues (which make their standings information publicly available).

I read *Winning Fantasy Baseball* a couple years ago (if you haven’t read it and you’re about to read 2,500 words on SGP denominators, you really should get the book), and I vaguely remembered the book giving SGP denominators for a variety of league types. I wanted to verify that my findings were similar to Schechter’s.

Here’s what I found:

Source | BA | R | HR | RBI | SB |
---|---|---|---|---|---|

2015 12-team NFBC Online Championship |
0.00180 | 19.92 | 8.43 | 19.55 | 7.59 |

Winning Fantasy Baseball 12-team League |
0.00165 | 15.52 | 5.93 | 16.30 | 5.93 |

Damn. WTF does this mean? Those don’t look close to me. Did I do something wrong?

## A Very Important Point

As I looked more closely at the book, I noticed I missed a very important point the first time I read it. Next to each number, Schechter had calculated a “relative SGP value”.

SGP Type | BA | R | HR | RBI | SB |
---|---|---|---|---|---|

Raw SGP Denominator |
0.00165 | 15.52 | 5.93 | 16.30 | 5.93 |

Relative SGP Denominator |
n/a | 1.05 | 2.75 | 1.00 | 2.75 |

And here’s the important point Schechter makes about these calculations:

… when you’re trying to adjust SGPs for leagues of various sizes, it’s important to realize that

the raw value of the SGP isn’t very important, but rather the ratio of the values.~ Larry Schechter,

Winning Fantasy Baseball

## Mine Is Bigger Than Yours Is

I glossed over that red bolded sentence on previous reads (because it’s not bolded red in the book…). But this small statement buried in the middle of the 350-page book is exactly the point I needed for the self-doubt I was experiencing.

So in order to hopefully save you the same trouble, take note! You can’t compare your SGP denominators to someone else’s. You have to convert them to a relative scale first.

## Raw Versus Relative – An Example

Let’s focus in on just the HR and RBI stats from the table above.

SGP Type | HR | RBI |
---|---|---|

Raw SGP Denominator |
5.93 | 16.30 |

Relative SGP Denominator |
2.75 | 1.00 |

If it takes 16.30 RBI and 5.93 homers to move up the standings, this essentially means that one home run is 2.75 times *more important* than one RBI (home runs are more scarce, so getting one of those is more valuable than the more common commodity, RBI).

**16.30 / 5.93 = 2.75**

Do you remember working with fractions in elementary school? I liken this practice to that whole “lowest common denominator” charade we had to go through. Dropping the SGPs to a relative scale is like converting them to a lowest common denominator. If you leave the SGP factors grossed up at these high numbers (like 5.93 and 16.30), it’s more difficult to see the relationships you can see when they’ve been translated into the relative scale.

## One More Math Concept

If you read *Using Standings Gain Points to Rank and Value Fantasy Baseball Players* or if you’re generally familiar with the SGP approach, you know that we would **divide** a player’s home run total by the home run “SGP denominator” to know how many SGP the player contributes due to his homers.

For example, if a player is projected by 30 home runs, an SGP denominator of 5.93 would indicate the player’s homers are worth 5.1 points in the standings (30/5.93=5.1). If the same player is projected for 83 RBI, an SGP denominator of 16.30 suggests the RBI are also worth 5.1 SGP (83/16.30=5.1). The 30 HR are worth the same as 83 RBI (5.1 SGP).

However, the way Larry Schechter has calculated his relative SGP would require you to **multiply** a player’s stats to achieve that same equality. For example, the 30 HR multiplied by 2.75 is 83 “points”. The 83 RBI multiplied by 1.00 is also 83 “points”. The 30 HR are worth the same as 83 RBI (83 relative SGP).

## For Consistency, I Will Calculate Relative SGP Another Way

If you look back at the big bolded numbers above, Larry Schechter used the largest statistic (RBI for hitters and K for pitchers) as the __numerator__ in his conversion. I will use it as the __denominator__.

**5.93 / 16.30 = 0.364**

I’m mostly doing this because everything I’ve written about SGP to this point tells you to **DIVIDE BY THE SGP DENOMINATOR** (heck, it’s called a denominator, meaning it’s on the bottom of the fraction). To now tell people to **MULTIPLY BY THE RELATIVE SGP DENOMINATOR** seems too confusing to me.

I’m sure I’ve confused the hell out of everyone at this point either way. And I apologize for this. But I think this topic is very important to understand. I’m giving it my best! Even if you’re confused, keep reading. I think this will all pull together very nicely in the end.

Going back to our example of a player with 30 HR and 83 RBI, if I divide by an SGP denominator of 0.364 I get that same 83 “points” (forgive the rounding), meaning the 30 HR are worth the same as the 83 RBI under this approach. So whether you use Schechter’s relative numerator and multiply or my relative denominator and divide, you get the same results.

## How to Calculate “Relative” SGP Denominators

I’ve talked a lot about multiplying and dividing. So just to be clear, **to put your SGP denominators on the same relative scale, choose the category with the largest numeric value, then divide each stat categories raw SGP denominator by that largest raw SGP denominator**.

The largest numeric denominator is typically RBI for the hitting categories (the 16.30 from above is the largest SGP denominator) and strikeouts for pitching.

For the rest of this post I will be using this calculation of relative SGP denominators and NOT the way suggested in *Winning Fantasy Baseball*.

## My NFBC Relative Versus Winning Fantasy Baseball’s 12-team Relative

Using the method described above, I calculated the relative denominators for Larry Schechter’s 12-team suggestions and my 2015 NFBC findings. Here are the results:

First look at the white lines. These give me that queasy feeling I was describing earlier. He’s saying 5.93 HR for a 12-team league? And I came up with 8.43? That’s 2.5 HR difference. How can these suggestions even be in the same ballpark?

Now look at the yellow-shaded lines. After everything is put on the same scale things look a lot more reasonable. When you look at all items on a relative scale, you can see many of the categories are strikingly similar (BA, R, RBI, SB, K, SV), but still show small variations. There is some variance in the other categories, but things don’t look as stark as with the raw denominators. This supports our beliefs about SGP being able to “tailor” to our league tendencies and preferences, but still leaves me feeling a lot more comfortable that my denominators are in fact closer to Larry Schechter’s than it appears on the surface.

Right around this time I’m starting to feel more comfortable with my analysis. But I’m also very curious about what happens if I start looking at SGP denominators from other sources. So I set out to find as many sources as I could find.

**NOTE:** After publishing this article, it came to my attention that there’s a typo in *Winning Fantasy Baseball* that makes this last segment somewhat less relevant. I’ve elected to keep it in despite this.

@smartfantasybb @MikeGianella @jeffwzimmerman @Razzball Reminder: Pg 141 my book column headers for 12+15-team lgs are inadvertently swapped

— Larry Schechter (@LarrySchechter) January 23, 2016

## Other SGP Denominator Sources

The reliable sources I was able to locate for this analysis are:

- My Own Home League (has varied from 11- to 13-teams)
- Winning Fantasy Baseball (12-team league)
- Winning Fantasy Baseball (15-team league)
- Information Published by Razzball in 2012 (12-team league)
- 2015 NFBC Online Championship (12-team league)
- 2015 NFBC Draft Championship (15-team league)
- 2014 NFBC Online Championship (12-team league)
- 2014 NFBC Draft Championships (15-team league)
- 2013 NFBC Online Championship (12-team league)
- 2013 NFBC Draft Championship (15-team league)
- Jeff Zimmerman’s 2015 Draft Prep Series (Actual, 15-team league)
- Jeff Zimmerman’s 2015 Draft Prep Series (Adjusted, 15-team league)
- Jeff Zimmerman’s 2014 Draft Prep Series (12-team league)

Not bad. I was able to scrape up 13 different resources for comparison. And I threw in the average of those 13 resources as my 14th.

## Here’s the Raw SGP Data

You can see things are all over the map. You can see general patterns, but the data fluctuates wildly. Some of the raw SGP denominators are almost double others. For example, Larry Schechter’s 12-team HR denominator is 5.93 while Razzball’s 2012 article calculated a 10.40!

## And Here’s the Relative SGP Data

I’m not qualified to speak about this statistically. But seeing these in relative form makes me think they really are similar sets of data. And this really drives home the point that you can’t directly compare the SGP denominators for your league directly to another. They might be wildly different numbers on the surface, but very similar when converted to a relative scale.

Let me cherry-pick a few examples.

Here are the raw denominators for my home league, Jeff Zimmerman’s 2015 draft prep, and the average of all 13 sources:

Those don’t look similar at all. The denominators are all over the damn place.

Here’s the information after converting to relative scale:

Wow. Before I came to this realization, I was looking at a 14.80 R SGP denominator against a 22.02 denominator and thinking, “that’s like a 50% error” ((22.02 – 14.80)/14.80). And now seeing them in relative scale you realize the difference is only about 7% ((0.987-0.919)/0.919).

## The Elephant in the Room

Humor me and let me play my favorite game, “Guess What My Most Skeptical Reader Must Be Thinking…”

This is all well and good, but none of this matters unless you show if and how this ultimately affects player dollar values.

How’d I do?!?! How’d I do?!?!?

## Conversion of SGP to Dollar Values

First, I ran both the raw and relative Razzball SGP denominators through my dollar value calculation process. I assumed a 12-team league using traditional rotisserie rosters (two catchers, that’s why you see Schwarber and Posey). And I downloaded Steamer’s 2016 projections as of January 17th. Here are the SGP denominators used:

SGP Type | BA | R | HR | RBI | SB | ERA | WHIP | W | K | SV |
---|---|---|---|---|---|---|---|---|---|---|

Raw | 0.00240 | 24.60 | 10.40 | 24.60 | 9.40 | (0.0760) | (0.0150) | 3.03 | 39.30 | 9.95 |

Relative | 0.00010 | 1.00 | 0.423 | 1.00 | 0.382 | (0.00193) | (0.00038) | 0.077 | 1.00 | 0.253 |

And here are the results (check out the “Razzball Raw” and “Razzball Relative” columns to the right):

Woohoo! It worked. The players came out with exactly the same dollar values (the differences are all less than $0.25, which I believe is just due to my rounding of the relative factors) and in exactly the same ranking order.

## Now Compare All 14 Sources of Relative Rankings

Ick. It was a workout to go through this 14 times. But gotta prove the point, right? Remember, the 2016 Steamer projections were used for all 14 variations of the test. So any difference in dollar value is due entirely to the differences in relative SGP denominators.

I took all 14 resources and converted them to dollar values following the process I outline in *Using SGP to Rank and Value Fantasy Baseball Players*.

I reevaluated replacement level after entering each system’s relative SGP denominators. Replacement level changed very little. It never changed for certain positions. And for others it would vacillate between two players, based on the system being used.

Here are the dollar values (click here to see the full spreadsheet or note the scroll bars along the bottom and right-hand side of the embedded sheet below)

The highest value for each player is shaded green and the lowest value is shaded red. My home league’s values come out the highest for the top end players. I think because it has the lowest AVG SGP denominator, meaning it’s placing more priority on the counting stats (which are highest for top players) than the other systems. It makes up for it by being the lowest on players toward replacement level.

And because it’s a bit hard to make sense of all those dollar values, here are those dollars converted to rankings:

## What Does This All Mean?

For the most part, there is very little fluctuation in dollar value (less than $1.50 on most players) or ranking. The players that do have noticeable fluctuations are extreme specialists of some sort, particularly stolen base artists (Dee Gordon, Billy Hamilton, Delino DeShields, etc.).

I was also glad to see that there’s less fluctuation in the rankings toward the top. This makes sense to me. The best players are separated from the field and from each other, to an extent. So mild fluctuations in their projected value don’t change the rankings that much. Whereas when you get to the tightly bunched players outside the top 30 you start to see ranges 10-20 ranking spots

I take away three main points from this exercise.

**First****, as long as your RELATIVE SGP denominators are in the ballpark of those shown below (the average of all 14 sources), you’re fine!** Don’t sweat it.

**Second****, despite similarity in relative denominators, SGP does have value in “tailoring” to your league preferences**. We see that above in the valuation of the “specialists”. Some leagues appear to value stolen bases and home runs differently. Batting average makes a different in some.

**Third****, if you’re in a new league or you don’t have at least several years of standings history to draw from, you can still use SGP. Just use these:**

BA | R | HR | RBI | SB | ERA | WHIP | W | K | SV |
---|---|---|---|---|---|---|---|---|---|

0.00009 | 0.961 | 0.384 | 1.00 | 0.377 | (0.00230) | (0.00042) | 0.091 | 1.00 | 0.213 |

These represent the average of the 14 systems’ relative denominators mentioned throughout this article. Because they all yield such similar results and are so close on a relative scale, this seems to be the best SGP starting point for anyone not sure where to begin or unable to obtain historical standings information.

You might be wondering, “Well won’t using these relative denominators screw up my dollar values?”. And the answer is no. Remember the demonstration above with the Razzball data. If you think about what we’ve done and the mechanics of the dollar value calculation, you can see this won’t affect things.

Yes, by using a HR factor of 0.384 instead of 10.40 (Razzball’s factor) the standings gain points for individual players are much higher. But they’re higher for every player! And they’re higher in the exact same ratio as the raw SGP values would have been. Because of this the league’s total hitting budget is allocated to each player in the exact same way and dollar values are unaffected.

## What Do You Think?

What kind of numbers do you see after you convert your denominators to a relative format? How about if you’re in an AL- or NL-only league? Do they hold similar to this too?

I’d love to see some of your data in the comments below. If you’d like to be notified of future articles like this, please be sure to follow me on Twitter.

I read that book at least 5 times and never saw that once. It certaintly makes me feel better about how I use other league data. Good read Tanner.

Thanks, Jason. I’ve only read it once… but now I’m thinking I need to give it a reread.

[…] The relative calculation seeks to put all the different SGP denominators onto the same scale. It’s helpful for comparing SGP denominators from different leagues (it can be misleading to just compare your own denominators to those above without converting them first). I won’t go down the rabbit hole now, but you can read more about relative standings gain points here. […]

[…] Relative SGP is a concept I uncovered reading Larry Schechter’s book, Winning Fantasy Baseball. It’s the practice of placing all the hitting and pitching denominators on the same scale so they become more comparable to denominators for other leagues and formats. You can see a bigger demonstration of the concept here. […]

I used the same approach but my $values for a mixed league are really low. Like $12 for Trout kind of low. I agree the relative approach is better but I can’t figure out why my values are so low.

I use Numbers for Mac so I haven’t followed everything from this blog but I followed the method from Larry Schechter’s book. Do you do the same thing? (Make $values for AL/NL separate and adjust in auction for looking for increasing discounts)

My rankings end up correct (using marginal SGPs for catcher/non-catcher) my money adds up to $260x Number of teams. My $ is split 67/33% but I’m getting like 1/3 of the prices

Hi, Zach. Thanks for the question. Yes, for the most part I follow the same approach Schechter outline’s. Or maybe one of the approaches. Because I don’t make separate ones for AL and NL. If memory serves he outlined several different methods to get to an end dollar value.

As I read your comment I thought for sure you must have some kind of an error where your total $ values are not correct. But then you mention that you’ve double checked that.

How many teams? How may players on the roster?

Are the last players that will be drafted valued at approximately $1 or something close to $0?

When I see odd dollar values, it’s usually an issue with where and or how replacement level was set. You may want to double-check that. The process of determining replacement level can be confusing and difficult to explain, so I actually made a video of an approach I use now that’s pretty fool-proof for me: https://www.youtube.com/watch?v=j0ztlMHwPiU

If you’re still having trouble, let me know.

I Used the marginal SGP method. Basically I set $ to:

Catcher: round(((SGP-last catcher SGP)*$/SGP amount)+1),1

Non-Catcher:round(((SGP-last non-catcher SGP)*$/SGP)+1),1

Pitchers, same as non-catcher but use last pitcher.

Rounding was to get one decimal place.

The only thing I could figure was that I messed up SGPs but the top guys check out on a ratio basis and the order makes sense.

What is the value of your last catcher? Do you have large benches?

I end up with a SGP value of 380 for trout and 104 for Jason Castro (last catcher) 178 for Colby Rasmus (last non catcher)

I’ve been trying everything I can think of to get correcting numbers. 12 team NFBC roster sizes.

The only thing I can think of now is that I actually messed up the SGP formula somehow and I’m not getting enough credit for a base hit in the average formula or something like that. My home league dissolved from the commish moving away so I don’t have league history for SGPs anymore so I was using your relative ones for an NFBC draft.

Unfortunately I end up doing a lot of my stuff during dead time at work on my iPad so I don’t have excel to work with, only numbers which no one seems to ever use (for good reason I think) so I’m not even sure how much help you can give. If you’d like to move it to email on any help you could give that’s fine

Hi Zach. Yes, I’ll try to e-mail you. Hopefully I can get the Numbers file to work.