99 Responses

1. Tanner Bell at |

1. John czajkowski at |

How would you calculate era and whip? And say a specific ratio like k/bb? Thanks for the great work.

1. Tanner Bell at |

The calculations for ERA and WHIP are documented at the very end of the article under the section “* Note About Pitching Rate Calculations”.

K/BB is interesting. Remember there are two things you will need to calculate:

1. The increase in K/BB required to move up one spot in the standings. To calculate that, take the average ratio of the category winner for the last several years and subtract the average ratio of the last place team in the K/BB ratio category for the same period. Then divide that difference by 11 (assuming a 12 team league).

2. You need to weight each player’s K/BB performance and figure out their effect on an average team. I’m wildly speculating because I don’t have data to draw from, but assume the average team in your league had 1,300 Ks and 400 BBs, or a ratio of 3.25 K/BB.

For a 9 pitcher staff that’s 144 K per pitcher and 44 BB. So an 8 pitcher staff would have had 1,152 K (144 * 8) and 352 BB (44 * 8). If the pitcher you are evaluating has 200 K and 60 BB, he would bump the ratio from 3.25 up to 3.28 (1,152 + 200)/(352 + 60).

Then divide this increase by the answer from the first bullet in this comment. (.03 / XX)

2. john czajkowski at |

This may seem like a rather simple question, but how do you add the formulas so that they automatically calculate the SGP for say wins, saves, etc.? I added the formula in the column, but it didn’t do anything, the column is blank still. Please help.

3. john czajkowski at |

Tanner,
Also, looking at the Excel chart posted with this page. I cannot find the SGQ ERA based on your ERA formula. Give me an example please of how you end up with a specific SGP for a P.

2. Don Mallo at |

I have been Commissioner and founder of the Rockland County NY, Roto League since 1981 and have to congratulate you on the most clear and precise explanations of the SGP method. Well done. The average seniority in our league is 24 years and no one has come close to explaining SGP. in fact, I am going to translate our historical stats based upon your method. By the way , we are a 6×6 non keeper ultra league rostering 12 hitters, 9 pitchers and 2 reserve spots for \$250. Have you compared SGP with PVM? If so, what’s the variation, if any?

Against, thanks and I look forward to future articles.

Regards,
Don Mallo

3. john czajkowski at |

How’d you calculate .0024 above for AVG?

4. Don Mallo at |

Looking forward to your article on PVM. Will be interesting to compare the rankings. Getting back to SGP for a moment. Have you posted an article on translating the SGP scores into dollar values? I have created a sheet using our 16 team 12 category league but am unsure if my method is correct. Our total pool is \$4000 (250 x 16) and I have allocated this pool between hitting and pitchers on a .70/.30 ration for simple math. This leaves \$2800 for hitters and \$1200 for pitchers.

I then calculated the total number of points achievable by league members for the 6 hitting and 6 pitching categories in our 16 team league (1644 total points with 822 for hitters and 822 for pitchers) and divided these results into the appropriate dollar allocation to obtain a dollar value per point. (Our league uses an extra point for 1st place in a category so that the leader gets 17 points and the 2nd place team gets 15).

I then multiplied the dollar value per point by the appropriate SGP number using your methodology to obtain a dollar value.

Thanks again for a great and well-presented series.

5. KG at |

How do you handle outlier historical data? I’m in a 10 team NL only 4×4 (OBP instead of Avg) and its hard to find reliable values for that setup so this year I’m giving SGPs a shot this time around. Our league has historical data going back to 2001 (the league has actually been around since 1988) so getting the data for the counting stats is very easy. Unfortunately for the rate stats our website has only recorded the final number and not the components (PAs, IP, Hits, etc) that went into those final numbers. As this is not a common league finding a compilation of data like Razzball has on a 12 team 5×5 mixed is a little hard to do.

I think I have found a decent way around this. On the site I can get the player stats for each season and so I pulled that data and came up with a yearly average OBP from the Top 140 players that the site lists (exactly how their rankings are calculated I’m not sure). I’m looking at data from 2009-2013 and for three of those seasons my average OBP is a perfect median number and one of the years its only off by one (four above, 6 below).

However 2009 is super funky. When I calculate the average OBP my result would have place between 7th and 8th which isn’t that close. The interesting thing is that this average is the highest average of the five year period I am looking at, the recorded OBPs for that season were apparently insane. Looking at the data I noticed that season’s AB’s and Hits were about 10% less than the 2010-2013 data (which were all remarkably consistent). I’m having a hard time reconciling how my “average” OBP (once again I’m not certain on the actual results each team had) is so much lower than what the median should be. I know data is data but seeing how the odd this is all seeming I’m thinking about just throwing out 2009 all together (and will therefore have to throw out the 2009 counting stats too).

Also, has anyone given any thoughts to the effect of the Astros moving to the AL is having on historical data in mono leagues? Comparing the 2013 data my regression lines are not as steep but I’m not sure how much that has to do with the Astros based on one year of data. If there is an effect I’m not exactly sure how to quantify and adjust for it.

1. Tanner Bell at |

KG, you ask a number of really good questions. I’ve always played in standard mixed leagues, so I’ve never run into these issues myself before. But these questions have me giving long looks at the other ranking approaches (PVM and z Score), because they’re not so dependent upon league history. They are inherently “forward looking”, whereas SGP is looking at the past and assuming history will repeat itself.

I hope to eventually write up how to do all three approaches. But I can’t say when I’ll have something like that completed.

I have been thinking about if it makes sense to include league history from 5-10 years ago in SGP calculations because the offensive environment has changed so much. I don’t have any science to back me up, but I am thinking it would be best to eliminate anything prior to 2009 from your calculations. I think it will just muddy the water too much and possibly give you inaccurate results.

1. KG at |

Unfortunately when i was working through the pitching data my workaround solution went out the window. When I go and pull player stats from previous years, our site (CBS) didn’t lock/archive what players were actually present. I noticed this while going through last year’s pitcher data and saw Fister show up despite being in the AL last year. I was also noticing that my top 90 pitchers also included several middle relievers who were not rostered during the season which was also throwing off the data.

Since I don’t have reliable historical data (I’m going to make a better attempt this season to keep track of the rate stats), I’m going to have to move on from SGP even though my gut tells me that for established leagues it might be the better way to go. I’m not convinced on PVM on my limited study of it so far, it seems like you need to do too much adjusting by guesswork to balance the individual stats.

6. AT at |

Apologies if this was covered somewhere and I missed it…seems like a bit of an obvious question. I’m in a 20 team H2H re-draft league and I can’t seem to find the magic place where it shows past stat totals. Would the SGP for a 12 team league be similar, or do I have to somehow obtain that information (which I’m having trouble finding.)

Sorry if that’s a dumb question, never really got into the nitty gritty world of spreadsheets before and I’m not really skilled with excel. Thanks for this guide, it’s really helped a huge amount.

1. Tanner Bell at |

I think the fact that there are 20 teams will affect the SGPs. Being a H2H league also complicates things because it is a legitimate strategy to punt one or more categories in H2H leagues. But I don’t suspect it will affect things too much. The players you establish as “replacement level” will be the biggest difference. What site is your league run from?

1. AT at |

Yahoo. Some guys are using z-scores instead and I’m not sure if that’s a preferable method for H2H. I think I’m going to try running with this and see how it works out and also since it’s easier to understand. Would you be able to take an educated guess at the SGP for K/BB?
Thanks.

1. Tanner Bell at |

AT, I can’t give an educated guess. But if you have your historic league standings/results available to you in Yahoo, you can perform a calculation yourself using the tool here. You could just fill it out for the one category you need.

1. Tanner Bell at |

And while I can’t give the SGP calculation, I do discuss an example of how you would calculate it above in this comment.

7. Ryan B at |

Tanner,

Just found your site recently and have been enjoying what I’ve seen so far! Nice writeup on SGP, I’m personally a bigger fan of using z-scores (and using historical stats for std devs), but the concepts seem essentially the same.

Question time: I’ve been wrestling with an issue lately that I’ve been personally referring to as the “Hanley Ramirez problem” – certain players, especially elite ones with injury histories, tend to be projected for ~550 PA or less while still managing to be well above replacement level. But are these various methods correctly valuing these players? As you point out in another of your articles, winning teams maximize their opportunities to get more PAs. If Hanley manages to put up 550 PA of elite production, that still leaves you ~100 PA of some other bum rep lvl SS to accumulate more counting stats. Meanwhile Elvis Andrus is projected (Steamer 2014) to put up almost as much value as Hanley, but over a full 650+ PA, so there’s no bonus playing time value to be had.

So, have you come across any methodologies that account for this in some way? Certainly we wouldn’t want to go in the direction of Oliver and try to just normalize every player to the same # of PA’s, but maybe some amount of replacement level performance should be added to a player’s statline, weighted based on something like (650-PA)/650? (Maybe 650 PA would not be the number to go with, but something like avg. PA of above-replacement players at each position)

8. Tanner Bell at |

Hi Ryan, I’ve thought about adding this to my process to see how much it will affect things. I envision it would work similar to how the “Replacement Level” lookup works right now. I think you would have to create some kind of a table that calculates the stat line of the replacement level player ON A PER PLATE APPEARANCE BASIS.

Hanley’s calculation for runs would be:

Hanley’s own baseline calculation for Runs + (650 – 550) * Per PA Runs from the table

Where 650 is the “standard cieling” you adjust everyone to and 550 is Hanley’s actual plate appearances. The difficult part is that i don’t think you can give credit for the full 100 games Hanley misses. Some of those he might not be active, but not on the DL. Some will be normal days of rest, etc. You can’t predict some of that accurately. So maybe multiply the gap by 80% or some other measure.

9. direwolfdc at |

Great stuff here! Thank you. How would the step by step calculations change for calculating a SGP formula/value for OBP instead of AVG? More variables than just hits and at bats.

1. Tanner Bell at |
10. Sean at |

“I know that it takes about 200 SB to win the category. So I draft players to reach 200 SB. Not 12 SB SGPs”

I would like to know how to compare the SGP of the players you’ve drafted with the SGPs needed to win the league. I feel this would let me know when I’ve over-drafted a stat category and where my needs are to make trades.

For instance, in my fantasy hockey league I calculated the SGPs of the offensive categories for the season thus far.
G P HIT BLK +/-
4.0 11.0 68.7 8.3 5.3

If I total my teams SGPs up I get 65.3 53.6 16.4 59.3 21.4 in those same categories, respectfully. I know I need 49 126 223 121 14 in each to finish just ahead of the 2nd place team. Basically, I’ve got a feeling that my team has overkill in one stat (Hits) while lagging in another (blocks). How can I verify my hypothesis with SGP math? How can I used SGP mid draft to stop drafting a stat I have accumulated enough of and focus on category holes on my team? MUCH THANKS! Feel free to email me.

1. Tanner Bell at |

Hi Sean, for SGP to work well you must have information about the statistics it takes to win a given category. I don’t play fantasy hockey, but let’s just say you know that over the last 5 seasons in your league it has taken an average of 500 hits to win the category. If your team is projected to exceed 500 hits, then you are overkill in the category. There really is no need to overcomplicate things with SGP math.

Your sentence “If I total my teams SGPs up I get…” is worrisome to me because this is exactly what I suggest not to do. It’s much safer to simply know the statistics necessary to finish first in a category and see if your projected stats meet those requirements. When you start evaluating on the basis of SGPs there is a greater room for error and noise in the results.

11. Seth at |

Maybe I’m just not excel-savvy enough to figure this out, but how could I change things around here to fit the scoring system in my h2h points league? Any help would be great thanks!

12. Tanner Bell at |

Hi Seth. Points leagues do not need to use a system like SGP to rank players. It’s much simpler for a points league. You simply need a set of projections and then multiply those projections by the scoring settings for your league.

For example, I’m in a points league where doubles are 3 points, HR are 9, SB are 3, H are 5, 3B are 5, BB are 3, and AB are -1.

So if a player is projected for 40 2B, 30 HR, 10 SB, 200 H, 5 3B, 80 BB, and 600 ABs, their projected points would be 40*3+30*9+10*3+200*5+5*5+80*3+600*-1 = 1,085 points.

There’s no need to determine SGPs or z Scores or anything like that. You might consider an adjustment for replacement level players though. So if my example player above is an OF and the replacement level OF is projected for 400 points, then I would subtract 400 from every outfielder. The player mentioned above would then have 685 points in my ranking system.

1. Seth at |

Thanks for the response! Do I just need to make a separate sheet, and do the formula, for example HR * scoring?

Also, how would I figure out a replacement level player? Sorry if these are simple questions, just now getting into the depths of all I can do with fantasy statistics!

13. Tanner Bell at |

I do prefer to do my rankings on a separate tab from the projections, but you don’t have to. I do it on a different tab because it makes it easier to download a new set of projections and drop them into the file without having to redo all the formulas again. But it does take a lot longer to set up initially.

Replacement level is discussed in Part 6, if you haven’t read that yet. But basically you need to know your league size and about how many players will be drafted at each position. If you think 24 SS will be drafted in your league, you can use that 24th SS as replacement level.

14. Seth at |

Oh duh, that makes perfect sense for replacement level players. Thanks! Love the excel layout!

15. help plz at |

What would the formula be for k/9 if you calculated the average k/9 was 8.0 for your pitcher pool?

16. mizzoudavis at |

Tanner,

Love the site and your demonstration of SGP. I have a question regarding the example you present in the article. When you mention 14 batter positions, are you counting bench spots? Or just offense position spots?

Mike

1. Tanner Bell at |

Hi Mike, thanks for the kind words. Are you referring to the calculation of BA SGP? For this calculation, I think it is important to only use the number of starting batter positions you have. If you start 14 hitters, then 14 hitters need to be included in the calculation. The reason for this is that a single player can have a larger effect on smaller starting lineups. The same hitter will have a smaller effect on a large starting lineup.

1. mizzoudavisMike Davis at |

Awesome. Thanks!!

17. Drew at |

When calculating the SPG for Batting Average you say “In the Steamer projections, the top 168 major leaguers..” Who are these “top” players? The players with the most ABs? Great site, keep up the good work.

1. KG at |

Hi Drew,

This is something I struggled with at first until I figured it out (at least for hitters). Note that I’m using z scores instead of SGPs (I do not have enough historical data for my league setup to do SGPs) I think the method should work just as well.

1) Bring up your entire projection list for hitters and run the calculations for the counting stats (R, RBI, HR, SB) just the same as Tanner has listed above for all of the hitters you have projected.
2) When it’s time to do batting average (or OBP or whatever else you might use), just take the first 168 players on that list, it doesn’t matter what order they are in.
3) Calculate your average ABs and Hits based on those first 168 and apply the calculation to your entire set of players.
4) Calculate your total SGPs for each player.
5) Once you have done that, sort your project list by your SGP totals. This should re-sort the list based on your original averages and give you a new set of 168 to work with.
6) If you setup your formulas so that they update automatically you’ll just need to re-sort your list a few more times until the list stops changing and you should have your 168 (or however many you need with your league setup) hitters. If you’re in a 2 catcher league you may need to add a catcher or two at the bottom of your list.

I tried to do the same thing for pitchers but it hasn’t come out very well, it keeps trying to put a lot of middle relievers in my pitcher set that have a rare chance of getting drafted. I’m trying to find a method that works there but I haven’t been happy with my results so far.

2. Tanner Bell at |

Drew, great question. I just chose the 168 hitters with the most projected ABs. But you made me second guess myself. And upon further review, that’s probably not the best means to identify the top 168 hitters.

The method KG describes is probably the best way to identify the top 168. But I also thought it would be helpful to quantify the difference between the two methods.

Above I mention the formula for BA is “=(([@H]+1768)/([@AB]+6617)-0.267)/0.0024”, where 1,768 represents the average number of hits for a team, and where 6,617 represents the number of at bats the average team hits.

That is based upon the average hitter having 509 ABs and 136 H (for a .267 average).

If I look at the top 168 players after I’ve ranked them (and not just the 168 with the most ABs), the average player is projected for 495 ABs (instead of 509).

A .267 BA on 495 ABs would assume 132.2 H (.267 * 495).

That means the average team would have 1,851 H (14 * 132.2) and 6,930 AB (14 * 495). Adjusting that to the 13 players necessary to calculate one player’s impact is 1,719 H (13 * 132.2) and 6,435 AB (13 * 495)..

Running Trout’s 2013 Steamer projection (184 H, 619 AB) through yields the following:

(184 + 1,719)/(619 + 6,435) = 1,903 / 7,054 = .26977 New Team BA.

.26977 – .267 = .00277

.00277 / .0024 = 1.154

So Trout’s SGP on the top 168 players is 1.154. When I ran it on the 168 with the most ABs, it came out to 1.151. So the difference is negligible. But in a perfect world, you should do an analysis similar to what KG suggests.

18. Redbird at |

Would you adjust the SPG formulas posted above if the league is a 10 person league rather than a 12 person league? I’d imagine it would shift the values some…just not sure if it’s significant enough to need to recalculate. Thanks!

1. Sean at |

There’s only 9 hitters per team in my 12 teamer. Since I don’t have access to past years’ data, would you expect using your/or other public SGP values to be in the realm of correctness?

2. Tanner Bell at |

I think using the SGP method is still sound. But you need to make an important adjustment in the calculations. You need to adjust replacement level for your specific league.

For example, if you’re in a 10-team league instead of 12-team, you won’t use the 24th catcher as replacement level (assuming a 2 catcher league). It might instead be the 20th catcher.

Or if it’s a 12-team league with 9 hitters, you might use the 12th catcher.

You can use the rough SGP calculations, but you HAVE to adjust replacement level for your own league’s roster sizes.

Once you get some final standings for the league, then you could refine the calculations by calculating the SGP factors.

19. Matthew Scott Martin at |

Tanner, I am in a 10 team, mixed, standard 5×5 league. How can I change the formulas in the SFBB-Rankings-Part-4.xlsx for use with my team in this 10-team league? Thanks. I appreciate how you freely share your information. Love this site. Unfortunately for you, I won’t be sharing this site with my league-mates, but perhaps others who play in different leagues.

20. Tanner Bell at |

Hi Matt. Unfortunately, I don’t have an established history for 10-team leagues to draw from. If you have an established league history where you can access prior season standings, you can calculate these things yourself using the “What It Takes To Win Your League” tool.

But even if you don’t have past standings to draw from, I think you will still get meaningful results from using the same formulas I have listed on the site here.

As I mentioned right above here, the very important adjustment you must make is in determining replacement level for the 10-team league. Use your actual roster composition and the fact that you’re a 10-team league to figure out who the replacement level C, 1B, 2B, etc. are in your league.

21. Jimmy at |

Hey my league is a 10-team mixed league but only in our second year. Added new GMs as two didn’t even pay attention to their teams last year. So don’t trust the differences in points. You know a place to find 10-team league SPG numbers?

22. Jimmy at |

Thx

23. Jimmy at |

I’m just wondering if I should use different formulas for Starters and Relievers. I’m in a 10 team league with 5 starters and 3 relievers per team. Our categories are W, L, HR against, ERA, K/9, BB/9, H/9, SV. As you can see there are many ratio stats there, so relievers have more value. After doing this exercise however, the discrepancy is too large (all relievers are worth more than any starter).
So let’s assume all my numbers are without error (doubtful), I still wouldn’t pick Kimbrel ahead of Kershaw, no one would, but if I separated out the RP and SP, I could wait til people start picking relievers and use my RP list a that point.
I could use the IP and ER averages for the top 50 starters, and do the same for the top 30 relievers and make separate ranks. Maybe I’m way off track.
Thoughts?

1. AT at |

Maybe not my place, but my way of getting around this issue was simply converting ratio stats into counting stats. My excel formula looks like this: =([@ER]*9-([@IP]*AVERAGE(\$Q\$2:\$Q\$264)))*-1 where Q is ERA. This works for any stat, take the numerator and multiply by the denominator*the league average for that ratio stat. I went with z-scores over SGP in the end (lack of league history plus I think SGP works better for roto leagues, I only play 20 team H2H) so I am not sure how you would calculate SGP from that.

1. AT at |

*subtract the denominator, not multiply, sorry.

2. Drew at |

You will want to treat those categories like Tanner has treated the other ratio categories. “Does the fact that Trout will hit .297 over nearly 200 more at bats make him more valuable? Or does Tulowitzki’s .300 average make him the better “batting average contributor”?” Think 200+ IP of Kershaws’ K/9 vs 60 IP of a Kimbrel.

2. Tanner Bell at |

Hi Jimmy. If your categories really are W, L, HR against, ERA, K/9, BB/9, H/9, and SV, then I think it makes sense for relievers to have more value than starters. A reliever will have fewer losses, fewer HR against, a lower ERA, better ratio stats, and more saves. The only benefit to having starters is the W category.

I do not think it’s necessary to treat SP and RP as different groups. But you should calculate replacement level separately for the two groups. The calculations will all be the same, but starters will be compared to replacement level starters and relievers compared to replacement level relievers.

If you follow the formulas for weighting the stats, you’ll be fine.

24. Jimmy at |

Hey thanks AT and Drew,
AT – Where do you find z-scores? I must have missed something.
Drew – I agree, I’m just thinking from a strategy point of view; with our cats, my numbers say RPs are worth the most, but strategically it doesn’t make sense for me to pick RPs in the 3rd round for e.g. So I’m just looking for relative value between SPs and RPs, I understand the contribution aspect, but I will just use my RP list when guys start drafting RPs.

25. AT at |

Z-scores are different from SGP and simpler to integrate. You just standardize all the stat columns your league uses and add them together. For example, my z-scores for K/BB: =STANDARDIZE([@XKBB],AVERAGE(\$T\$2:\$T\$180),STDEV(\$T\$2:\$T\$180)) Where T is XKBB, which is K/BB converted to a counting stat. Then you just mash them together. My rankings passed the sniff test, so I’m happy with it. It’s just an alternate to SGP.

26. PEDex at |

My league is switching over from AVG to OBP. Any advice on adjusting ranking calculations for OBP without inflating its overall weight?

27. Matthew Scott Martin at |

Tanner, being a neophyte at ;Excel, I do not know how to do this one thing other than manual. I would like to grab the PlayerID, LN, FN, POS, and TTLSGP columns from both the Hitter Ranks and Pitcher Ranks tables and merge them into one worksheet and table so I can compare the pitcher and hitter Totals at a glance w/o having to toggle back and forth between sheets. How do I do this? I will have enough to do on draft day (two draft day software pkgs. [I might buy a 3rd b4 DD], and at least three spreadsheets besides your SGP sheets). So, I don’t want to be toggling any more than necessary. BTW, just bought the entire 10 chapters so I could read up on the impact of keepers; I am in a keeper league.

Thanks

28. Bert at |

When calculating the SGP value of AVG, for example, what’s the advantage of using the Razzball average AVG (or even the average AVG from one’s own fantasy league) as opposed to using the projected average of those top 168 hitters per the projections? Given the yearly changes in the MLB run-scoring environment, wouldn’t using the projected averages of those players provide a better baseline?

I’m assuming, of course, that projections consider those trends, which they do, b/c they consider the trends in each individual player’s production. If there’s a league-wide decline in AVG, that’s the culmination of individuals’ AVGs declining.

But previous year’s data, taken statically, doesn’t account for that.

29. Brett at |

Hi Tanner, This is my first time trying to do my own dollar rankings, and I’ve found your guide to be very helpful and straight forward. I am in a 22 team mixed league though. We start 9 hitters and 9 pitchers, (6 SP and 3 RP). I’m just wondering how I can calculate the AB’s, Inn’s etc. for the ratio categories. Based on the above answer, could I just figure out the avg # of AB’s for the top 198 Hitters (9×22), 132 SP’s, and 66 RP’s?

30. Brett at |

I worked through it this way, and find AVG to very under represented. The top producers in R, HR, RBI, and SB’s all earned 8-10 SGP’s, whereas the top in AVG only earned about 2-4 SGP. The numbers I used were 4500 AB’s, a .262 avg,and 1205 total hits. Any suggestions on what I can do to fix this? Thanks for your help…

31. Anthony at |

Having trouble with calculating K/9 and BB/9. I have the Denominator for K/9=.09 and BB/9= -.09. 14 team league 10 pitchers (5 sp 5 rp) calculate the top 140 pitchers projected K/9 using ERA under 4.2 and top K/9 that would be rostered. Obviously average out my leagues average for both stats, would the formula look something like (TOP140K’s with 9 pitchers + proj K’s)/9/(Top140IP + Proj IP)-LeagueaverageK/9/.09 for K’s ? and then (TOP140BB/9 with 9 pitchers + Proj BB’s)/9/(Top140IP + Proj IP)-LeagueaverageBB/9/-.09 ?? Thanks, let me know if any one else can help out with calculating these two.

32. Dan at |

Tanner, do you notice that your top end players (hitters mostly) don’t have the nearly the same high end values that you see on other value sites? For example, in my 14 team 2 catcher 5 OF league \$260 cap 67/33 split, I have only 3 hitters above \$30. When I know Trout, Miggy, etc. have values over \$40. I feel like when calculating using historic SG it creates a ton of players in the \$10-20 range and not as many in the higher or lower ranges.

I’ve had success the past two years just with basic knowledge that I’ll have to spend more early, well above my projected values, but can’t seem to get it balanced like I’d prefer. I use steamer for my projections.

Thanks

1. Tanner Bell at |

Hi, Dan. I just ran a test on the league settings you described and I do get Trout over \$40 and then Cabrera and Stanton right around \$38, then 6 more players also over \$30.

Do you have a lot of bench players in the league that you’re trying to reflect in the values too? They could be draining some of the money from the top players.

33. Chris at |

Hi Tanner,

Working on my preseason. I’m in a 5×5 AL only roto league with 14 hitters and 9 pitchers. We have a bunch of keepers, but I’m trying to evaluate who is worth keeping based on my preseason work. So far I’ve done well, but now I’m on figuring ERA and WHIP. Based on my leagues history the Razzball averages are too low. To figure average ERA and WHIP do I simply take the average of our 12 teams over our history and use that as the basis for our average ERA and WHIP to create my own formulas?

34. Chris at |

Did some math, over the last 4 years, our leagues average ERA is 3.97 and our average WHIP is 1.29. Are those the numbers I should use in your formulas for calculating TTLSGP or should I use different numbers?

1. Tanner Bell at |

Hi Chris,

You could use historical averages, but I think I would prefer using the projected stats for the coming season. I looked at my projections for the top 108 AL-only pitchers and it looks like they will pitch 13,850 innings, giving up 5,750 earned runs for an ERA of 3.74 and a WHIP of 1.2574.

I think this is best because we are trying to figure out a pitcher’s impact on our ERA for this season, and our best guess of what those numbers will be are included in the projection

1. Chris at |

Hi Tanner,

Thanks for answering my question! One more, without using TTLSGP how did you determine the top 108 pitchers? Did you use the same formula you use in your book, the top 108 pitchers based on strikeouts for pitchers under 4.20 ERA?

1. Tanner at |

Hi Chris, this is a “circular calculation”. I take a guess at what the average ERA and WHIP will be and then figure out all of my SGP using that (including TTLSGP). Then once I have that I can “fine tune” the ERA and WHIP SGPs for the top 108.

After you change your ERA and WHIP formulas, the top 108 will probably change slightly. We could go into this continual loop. “Edit the ERA”, “redetermine the top 108”, “Edit the ERA”, etc.

35. Chris at |

Hi Tanner

Another question. I used steamer projections for 2015 and ran into the same problem as dan. Only trout and Bautista are ranked above \$30. I think I may need to run through the projections more carefully from steamer to be sure I have the correct info. I followed each step in your book to the letter. The only exception is I calculated slope numbers from my leagues historical data, used the numbers you gave me yesterday for whip and era and figured my own numbers for batting average based on the average of our leagues history. I could change the average to match the numbers from the top 168 in the projections that is my next step.

I also did a 70-30 split and noticed the top pitchers are valued lower than in my league but the middle level pitchers are valued about right. most likely my league just over values the top 6-10 pitchers. Do you see that often?

36. Tanner at |

Hi Chris, if you think your dollar values look odd, the first thing I would do is to double check your replacement level information. I found myself making a lot of mistakes with this part and found that it takes a long time to do, so I created a video to demonstrate the approach I’m using now. You can see it here.

If you see that your pitchers in your rankings are undervalued (in comparison to what you actually see on draft day), you could adjust your split downward. Try 65-35 and see if that aligns pitcher values better with your league.

37. Matt at |

First of all just want to say thanks for all the great work. I think I have run into a little problem though. I’m stuck at part 6 because after calculating SGP my rankings are crazy. A lot of closers are ranked at the top in pitching categories. I have 3 pitchers ranked higher than Kershaw. I have Altuve, Jacoby Ellsbury, Brett Gardner, Starling Marte, Billy Hamilton, and Dee Gordon ranked in the top 10 hitters. While I think s little error is normal this just seems crazy. Here is my league settings:
H2H Points League 12 Teams
1 C, 1 1B, 1 2B, 1 3B, 1 SS, 1 MI, 1 CI, 4 OF, 1 DH, 7 P and 6 bench spots

Scoring:
Batting- singles 3, doubles 5, triples 8, home runs 10, walks 2, runs scored 2, runs batted in 2, stolen base 5, hit by pitch 2, caught stealing -2.
Pitching- innings pitched 2.25, earned runs -2, wins 4, saves 10, strikeouts 2, hits allowed -0.6, walks issued -0.6, shutouts 2.5, hit batsmen -0.6, complete games 2.5, no hitters 5.

I Tried to calculate SGP for R, HR, RBI, SB, H, 2B, 3B and W, SV, IP, BB, SO, and then I did era because CBS would not give me our leagues ER.

I don’t know if this is wrong or not but it is producing crazy results like I said earlier.
Should I tailor part 5 to my league and if so, how?

Once again thanks for all the help!

1. Matt at |

And when I say I have crazy pitcher rankings, I mean that all closers are 1-26 and then I have 5 more pitchers ahead of kershaw, 3 of which are SP.

1. Tanner at |

Hi Matt, it sounds like you’re in a points league. Standings Gain Points should be used only for rotisserie format. I do have a separate set of instructions for points leagues here.

1. Matt at |

Okay awesome thank you! And is there a way to convert those rankings to \$ value form auction drafts? Will purchasing the book and using parts 7-10 work?

1. Tanner at |

The “Using Standings Gain Points” book will also only help you calculate dollar values for rotisserie leagues. I’m working on a dollar value book for points leagues. Stay tuned on that one!

38. Zach at |

Tanner,

Is SGP a useful concept for head-to-head leagues? If not, what’s the best approach to converting projections to ranks or dollar values for such a league?

Zach

1. Tanner at |

Hi Zach,

SGP exactly as I write about it does not apply to head-to-head, but with a slight tweak you could use it. Instead of determining SGP on the season ending statistics, you could grab several (or all) weekly standings data and run the SLOPE function on those weekly standings.

I think this is still a little imperfect for H2H because many managers don’t try for a balance strategy in that format. You’re incentive is to win the week, not perform well across the board in all categories. You may want to research the Percentage Valuation Method or z Score approach for something that applies better to H2H.

Thanks,

Tanner

39. Brendan at |

HI,

Thanks very much for this. Can I use this for an 8 team league? I know it is a pretty major difference, but won’t it still yield similar results (who is undervalued, where each ranks relatively, etc.)?

Thanks!

Brendan

40. […] for player valuation for your own fantasy baseball league, using a model based either on Standing Points Gained above replacement, or Fantasy Points above replacement, depending upon the scoring system of the […]

41. Mark at |

Tanner:

I play in a 20-team league with 28 man rosters. At a minimum, you must start 23 players, comprised of either 14 hitters, and 9 pitchers OR 13 hitters, and 10 pitchers (the difference is the utility can be a hitter or a pitcher). Historically, teams draft 16 hitters and 12 pitchers.

I am struggling to decide what numbers to use as a basis for my calculaiton of Standing Points Gain for the ratio stats. A couple of options:

Go with average number of players drafted as follows:
20 teams * 16 hitters = 320 hitters
20 teams * 12 pitchers = 240 pitchers

OR

Go with active roster sizes as follows:
20 teams * 13 or 14 hitters = 260 to 280 hitters
20 teams * 9 or 10 pitchers = 180 to 200 pitchers

Any thoughts? Obviously, this impacts how I arrive at the total expected ABs, Hs, IPs, ERs, etc.

Thanks!
Mark

1. Tanner Bell at |

Hi Mark,

I don’t have hard evidence to support this, but I can give you my line of thinking. It is only players in the starting lineup that contribute to team statistics, so I would use the “active roster” option. I would try to estimate how the teams in your league do this in practice. Does it break down that about half the teams start 14 hitters and half start 13? I’m sure it even fluctuates during the season, but give your best estimate.

So maybe it’s 20 teams * 13.5 hitters = 270 hitters

Tanner

42. mtnaesthetic at |

Are there any historical category resources for QS and OBP? We just switched this year and I feel like it’s a major inefficiency to exploit in my league, but I have no idea what my targets should be.

1. Tanner Bell at |

Hi mtnaesthetic, I don’t have this currently. I just put out a request on Twitter to see if any followers might be willing to share. Stay tuned…

43. White Sock at |

I’m new to the SGP conversation and am prepared to learn that I’m way off base here. Disclaimer aside…

I am uneasy with the potential for players to be given a negative SGP value in the ratio stat categories. It really shouldn’t be treated differently than the counting stats, should it? A player below league average does not have a negative value in the home run category.

I do understand that player A’s .230 average “drags down” player B’s .300 average, whereas player C’s 1 home run has no effect on the 30 that player D hits. However, the points should be considered at the TEAM level because that’s what we’re after. In this sense, hitting only 1 home run is just as harmful as having a .230 average.

What if I draft 14 players with averages ranging from .230-.250? The model would give me a negative SGP. If I draft 14 players who hit 0 homers, I end up with 1 “point” for that category and the model will reflect that.

My solution would be to change the comparison you’re doing to value one player:

from this:
the average 13 man team + PlayerX
(minus)
the average 14 man team

to this:
the average 13 man team + PlayerX
(minus)
the average 13 man team + Player0

To find Player0:
1 – calculate the projected team BA for the average 13 man team + PlayerX for every player in your universe. Might want to exclude anyone who isn’t projected for 450 AB or more. You’re probably not drafting someone who won’t anyway.
2 – Find the lowest…this is Player0

It’s not different than finding value for the counting stats. You’re effectively assigning the player a numeric value based on what he adds over the worst possible (and practical) outcome in each category.

In the end, if I draft a bunch of .230-.250 players, the calculations should tell me I’m going to gain zero in the AVG category. Same as 0 HR.

But what’s the true impact?

Honestly, NONE, in practical terms. The deltas between players are what matter. And this change shouldn’t affect the order players are ranked in within the category or their relative worth overall. It just puts the ratio stats on the same scale as the other categories.

If there IS a danger, it would be that of perception while drafting. That people might under-prioritize the ratio stats because their scale, while relatively correct between players, is between -W and Y, rather than 0 and Z like the counting stats.

In fact, you could probably get pretty close to the same answer by calculating the difference between the current SGPs and the worst SGP in the player universe for that category. Doesn’t address the player 0 gripe, but should be directionally correct.

Thoughts?

1. Tanner Bell at |

Hi White Sock,

I do want to start by emphasizing one point you make, “What’s the true impact? Honestly, none, in practical terms… This change shouldn’t affect the order the players are ranked in within the category or their relative worth overall.”

I think that’s very well said. It was the thought I had in mind as I started to read your comment, but you said it better than I could have myself.

I do see a lot of merit in the change in calculation method that you propose. Have you tried this? I’d be curious to see what you come up with if you compare the method I wrote about to your proposal. What kind of change in dollar values for players might we see?

If you have not had an opportunity to try this, I will try to keep it on my radar for something to try next preseason.

Thanks for the suggestion.

Tanner

44. Zack at |

Quick question as this is my first time calculating SGPs: How do you define “top” batters when determining the # of at bats for the “top” 168 batters? Is this by the top 168 batting averages or the “top” fantasy players by earnings (and if it’s the latter, is that determined by the 2015 actuals or 2016 projections?)

Trying to avoid the old saying: Garbage in, Garbage out! (when it comes to modeling)

Same applies to ERA and WHIP too.

Many thanks!

45. Tanner at |

Hi Zack,

I figure this out in a pretty simplistic manner. I just sort the 2016 projections by at bats. For the most part, we end up drafting any player that plays full-time in the top 168, so it’s not that far from who the top 168 actually are.

You could get more involved here and make this a circular calculation. You can run the numbers first using the approach I just mentioned. And then when you’re all done, figure out who the top 168 actually are. Then take the numbers from that 168 and adjust your formulas to these new inputs.

You will probably find that doing that does not have much of an impact on the rankings.You can apply these same principles to the pitching categories. Just sort by IP to get the top 90 and go from there.

Hope that helps.

46. […] hundreds of thousands of words about this topic, and most try to quantify it via something called Standings Gain Points, which I only have a loose grasp on. Basically, it tells you how many Rotisserie points your team […]

47. […] I calculated a basic 12-team Standings Gain Points (SGP) value for each position. The formula helps to standardize different stats to a similar reference level. […]