It’s Time To Ditch The Quality Start
Yesterday, I sat down to write this article only to discover there was another topic I had to cover first. Classic 5×5 Roto is one of the original forms of fantasy sports. And, despite attempts to tinker with the inputs, it remains the king of fantasy baseball.
A common change to the classic 5×5 categories is to swap out pitcher wins for quality starts. Although all agree a quality start doesn’t exactly measure its name – after all, a six inning, three run performance is pretty mediocre – everybody also agrees the stat is less capricious than pitcher wins. This season, Jacob deGrom won 10 games with a 1.70 ERA over 217 innings. He led the league in quality starts. Ryan Yarbrough won 16 games with a 3.91 ERA in 147.1 innings. He only made six starts, none of which were a quality start. Yeah, wins are sloppy.
Despite the flaws with wins, they’ve become vastly preferable to their number one alternative. It’s time to ditch the quality start. Don’t believe me? Behold, a table!
Das Maths
| Year | W in GS | L in GS | Decisions | QS |
|---|---|---|---|---|
| 2018 | 1515 | 1612 | 3127 | 1996 |
| 2017 | 1640 | 1521 | 3161 | 2121 |
| 2016 | 1628 | 1706 | 3334 | 2262 |
| 2015 | 1673 | 1705 | 3378 | 2432 |
| 2014 | 1706 | 1719 | 3425 | 2623 |
The above data are taken from BaseballReference.com. We’re looking purely at decisions and quality starts by starting pitchers over the last five seasons. Here we can clearly witness how the latest real world pitcher trends have affected our fantasy games. Wins by starting pitchers have declined by nearly 200 per season leaguewide. During the same period, over 600 quality starts have vanished.
Both trends are liable to worsen over the next season. The Rays’ success with the Opener-Follower strategy will encourage other clubs to copy them. Although I don’t expect any team to adopt openers as completely as Tampa Bay, we may see nearly every franchise dabble with the strategy.
The supply of starting pitchers who offer meaningful win and/or quality start totals is roughly equal – about 40 pitchers. Replacement level for wins is around seven to nine while quality starts is usually in the low teens. That the replacement level is higher for quality starts does not mean the supply is bigger.
Quality starts are more predictable. As our above anecdote about deGrom and Yarbrough demonstrates, you can count on good players to post a big totals. Relievers are auto-zeroes. Wins are wild. Anything goes.
As we discussed in yesterday’s celebration of the classic 5×5 roto format (first link above), it’s not a good thing that quality starts correlate heavily to ERA and WHIP. In this way, the rich get richer. When I see a staunch advocate for using quality starts over wins, I immediate suspect them of preying upon their weaker leaguemates by controlling the supply of a stat. They’re not necessarily conscious of this manipulation. After all, there is only one way to accrue a large quantity of quality starts. There are many ways to skin the wins cat.
Although the starting pitcher supply of both stats is comparable once adjusted for replacement level, wins don’t exclude relievers. In the past, this has been used as an argument both for and against quality starts. From one perspective, it’s nice that elite non-closers typically win games at the same rate per inning as mid-tier starters. This gives them another category of value, thus making it all the more important to roster the Josh Haders of the world. The other perspective views this as an unnecessary evil. Relievers have saves, starters have quality starts. Each has four categories. Balance.
Let’s review the facts. More and more innings and decisions are going to relievers. After all, there are 2,430 wins every season, give or take a couple for rainouts or Game 163s. If starters are winning fewer games, that means relievers are winning more. Quality starts are simply evaporating into nothingness. The new Opener and Follower meta is only going to exacerbate this trend by turning most back end starters into relievers. Under these conditions, it’s time to embrace the non-closing reliever as a source of wins.
Now What?
Perhaps it’s time for an alternative – a stat that captures the spirit of quality starts without restricting the supply to only starting pitchers. After all, there is merit to having an elite reliever-only stat like saves (or saves+holds) paired with a volume-only category. Maybe it’s time to discard the concept of starters.
How about a Quality Outing? We could define it as any appearance of four or more innings with a 3.50 ERA or better. We can quibble over the specifics; they’re not important right now. Instead, we need to focus on fixing what’s broken. Quality starts are dying as a viable category. They already only work in 10 team mixed and shallower. However, people still want an alternative to wins, and it’s not like that stat hasn’t also become deeply weird in recent years. It’s time to make an alternative.
You can follow me on twitter @BaseballATeam
¿porque no los dos?
Why not both? It’s an option, roughly as sloppy as wins alone. There also aren’t many platforms that offer W+QS. Not that any offer my made up QO.
Yep, kind of thinking IP might just be the answer. Bad IP are already hurting ratios. It’s all about propping up the players that pitch the most.
My primary league runs out IP, (W+QS-L), K/9, HR/9, and WHIP. It works a pretty nice blend of accumulations and skills. You can always stream to win IP, but unless you pick real lucky it’s going to hurt you everywhere else. We actually had to institute an innings floor after one enterprising owner decided to roll out an all-reliever lineup to dominate the ratios.
I actually always argue to ditch QS in favor of Wins because pitcher wins encourage better baseball fans by way of forcing people to follow a game after their pitcher has left the game. You could define better as “better informed” but what I want is a league full of people who want to follow baseball games beyond just the stat lines.
It is entirely possible that this is an overreaction as much as the new normal. Part of the reason that QS are in decline is the cost efficiency of scumming a game together out of the bullpen. I don’t know that QS are any less valuable or that successful teams will have that many less QS. There are certainly less teams trying to win games – which means that fantasy baseball players are generally in trouble. Hard to imagine some kind of quality game becoming particularly relevant. I feel weird that I have to caution people about ERA, but people seems to be paying more attention to what was becoming less valued (correctly) just a few years ago – we are fully against errors, but still excited about ERA? WHIP was a revolution that changed the game for the better not long ago. I think the best solution is just increasing the weight of innings pitched, but that doesn’t lend itself as well to the narrative of a bold new future.
This is why I like points leagues, where you can easily ditch W/L/QS altogether because the most value is in inning totals.
I’m not sure I see a problem with evaporating QS. Back end starters become elite middle relievers. It is easier to obtain elite ratios when allocated a few innings vs 5 or 6. They produce ratios that 99% of the starters can’t produce. There is an opportunity cost here that I think makes it fair. Forgo some QS/W for elite ratios? It’s strategy. I worry further fantasy category reward for middle relief will too greatly shift fantasy value from starters to middle relievers.
I think we should double down: quality start and qualitier start
As said by smada above, Total IP is the answer. It is not complicated and it serves the purpose of driving the value of quantity which what is what the W category does now except the W is more arbitrary. And often high IP means at least reasonable quality still. It also still allows relievers to contribute to it as well.
That only works for H2H unless you remove the IP cap for Roto. I consider no IP cap roto to be an experimental format.
Why not use a Games Started as a cap, and use saves+holds? Adding holds give relievers the bump they need in counting categories, and a GS cap while using IP as a category places emphasis on starters who pitch deeper into games and long relievers. Granted, this is done in a 6×6 format (IP, QS. SvH, ERA, K, Whip) but we made the switch a couple years ago and have really enjoyed it.
I’ve like this. Only downside I see is that this makes the Opener a toxic asset, as he torpedoes your IP/GS and has no chance at QS.
I like it. What are your offensive categories? We use QS and OPS now but people are talking about shifting something because QS aren’t as meaningful anymore
There is no innings cap in NFBC (only a very easily reachable 1000 IP minimum) and hasn’t been since its inception. It isn’t experimental. It is standard for high stakes fantasy baseball. Only allowing weekly lineups for pitching eliminates streaming issues which also allows for multiple viable strategies. I have won leagues with no closers and all starters and won leagues with multiple closers.
> after all, a six inning, three run performance is pretty mediocre
What is the average number of runs scored in an MLB game? Oh.
A 6 IP 3ER start gives your team a chance to win every single game.
So does 7 and 4, and 5 and 2. What is your point?
The entire point of using QS over Wins is so that you can utilize a predictable and consistant stat that is always deserved by the player attributed it. Wins are not always deserved, but QS’s always are… and they are earned the same exact way every time.
QS = Consistent and Repeatable = Fair.
Wins = random dumb luck = not fun.
Do we need maybe a better quality start metric? Sure. That doesn’t mean we “ditch the quality start”, though. That means we must *improve* it.
Given enough iterations, mono-strategic categories like QS are less fun than a difficult to manage, volatile stat.
Can’t agree with that at all. You know what isn’t fun? Inconsistency in application of statistics between managers.
Wins aren’t “difficult to manage”, they are impossible. There is absolutely no repeatable strategy involved with them at all. When 2 pitchers play giving identical stat lines but one gets a W and the other gets an L… how were you supposed to “strategize” around that?
When you target pitchers with good rates and the ability to go deep in games you are rewarded with QS’s. This is a positive action by me resulting in a positive action in the standings. If I targeted that pitcher this year and he was named deGrom, how was I supposed to mitigate his wins problem? Praying?
What is “fun” is when my players out perform your players and I win the matchup as a result. No outside influence, no arbitrary statistics…
What isn’t fun? When my players outperform your players but I lose because players that aren’t on either of our rosters changed the outcome.
You’d have a better argument if you wanted to ditch the “overall counter” statistic entirely. Solely judge pitchers based on rates and direct contributions.
There is simply no scenario where using “wins” is the appropriate choice.
You’re describing a different fantasy game entirely. Which is fine. It’s just not the point of 5×5 roto. The volatility and difficulty are there by design. It’s supposed to be frustrating. It’s supposed to force you to abandon your best laid plans for Plans B, C, and D.
It seems like a points league similar to ottoneu FGpts would better fit your preferences. Again, that’s fine. It’s important to learn your own preferences so you can play games you enjoy.
No, I think I am describing every fantasy league in existence that has ever tackled the thought of ditching wins for a better stat.
You are largely avoiding my main point here. Wins aren’t “fun” or strategic in ANY way, at all. This is directly due to their inconsistent application. Ditching QS to go *back* to wins is just an absurd premise that you probably should have spent more time thinking about before writing this article. It resolves exactly *none* of the problems that were avoided by leaving wins in the dust in the first place.
Volatility and difficulty still exist in the same exact manner with QS. You weren’t predicting anyone’s wins total via projections… not accurately… volatility is significantly lower in predicting QS totals. if anything you are making a more informed decision when you attempt to track/predict QS’s.
Strategy itself involves 2 parts, planning and execution. Using wins cheapens your planning and research at the expense of… I don’t even know what… “volatility”? You’re essentially just pining for an extra random factor to influence your outcomes. Why would anyone want that?
Can’t you make the same arguments about runs, RBI, and saves? These four categories are all inconsistent. Save opportunities depend on how many close games a team is involved in. RBI are dependent on other players being on base. Runs are dependent on other players driving in your player.
It seems like a lot of your argument is based around two concepts: 1) 5×5 roto is the gold standard by which all considerations should be made. 2) luck is what makes fantasy baseball enjoyable.
The latter puts fantasy more in the gambling category than the game category. Some may get a thrill out of lucking into victory, but some of us prefer to win by outsmarting our opponents, through strategy, research, and spotting trends. Given the nature of this site, I would imagine most of your readers being to the latter category.
Those same readers may not hold 5×5 to the same high regard as you. It is, after all, very one dimensional and often archaic in this advanced stat age. Just because it’s the default on most platforms, doesn’t suggest it’s superiority anymore than Lady Gaga’s record sales suggest she’s the best living musician.
5×5 is the standard because it’s the easiest for new players to understand. Those of us who frequent Fangraphs are probably more of the in-depth variety. What makes 5×5 roto any better than 6×6 roto?
Instead of parading a fluky, luck-based stat like Wins as the superior option, why not argue that fantasy baseball is better off with more categories? Then you can add in things like Holds to give middle relievers more value, add in things like Ks to devalue high strikeout batters, etc. The specifics are less important than the additional depth offered by taking off the 5×5 shackles.
Also, you can use DFS techniques to chase wins, especially if your league allows streaming. They’re effective but often come at the expense of ratios. Which is also something I consider a feature rather than a bug.
Quality Starts are not fair. Having a 4.5 ERA in six innings gives me a QS, but a 4.0 ERA in nine innings doesn’t? It’s a weird thing to root for a starter to get pulled as soon as they get a QS so that they don’t lose it. QS are the only stat that can get taken away as the game goes on. The world can be cruel sometimes.
Wins are not random dumb luck. People always tend to undervalue the teams that players play for in their evaluations. Mediocre pitchers on teams that win a lot have a different kind of value than pitchers that just pitch well.
I don’t like the idea of something non-pitcher related – like run support or managerial decisions – being able to affect the value of fantasy pitchers, but I can’t square that with how much team defense matters to a pitcher’s ratios. Unless we all move to purely isolated pitcher-stats, these things are all features not a bug.
League average is like 4.8 runs, but the most likely run outcomes are actually 3 runs or 4 runs (not 5, hardballtimes did a piece on this). Quality starts being limited to 3ER ensure that the pitcher gave his team a chance to win an average game.
How is that any weirder than getting a win for a 6.00 ERA in 6 IP? There is literally no limit to how poorly a pitcher can perform and still get a win, as long as the other pitcher gives up more runs.
A win basically eliminates all relation to a pitcher’s actual performance, which is the whole point of stat categories. If my player gives up 1 run in 9IP, but his team loses and your player gives up 4 runs in 7, how does that even remotely reflect the value and performance of the pitching in those games?
“It’s a weird thing to root for a starter to get pulled as soon as they get a QS so that they don’t lose it. ”
DING DING DING!!!
I will spend the rest of my life proselytizing about my awesome Wins replacement, GSv2>59. You get a “win” for every game started with a GSv2 over 59. That’s it. Game Score isn’t perfect but if you have a GS of 60, you probably pitched a pretty good game. That’s a clean cut-off that results in a pretty similar number of Wins, spread out more reasonably. DeGrom would’ve finished with 27. Corbin would’ve had 17. Godley would’ve had 10. Giolito would’ve had 5. Arguing for things like IP ignores the spirit of the W stat, a single binary yes or no number that’s supposed to sum up if a start was good or not. You can mess with other types of stats to get a different kind of game, but I enjoy the essence of the Win without the randomness.
Ron Shandler once advocated replacing Wins in a 5×5 roto with Wins plus Quality Starts (W+QS). I believe this is superior to either Wins or QS alone in generating a list of top starting pitchers.
I’m a fan of the concept, but finding platforms that host the stat can be challenging.
Fantrax
My league switched to W+QS back in 1993. 🙂
my league did this around 2010
After 25+ years of using wins only our league switched to W+QS. I think it worked out fairly well, at least better than using either category by itself. Since QS have been in a faster decline than Wins I’ll be altering my strategy a bit for 2019.
A pitcher really doesn’t have control on whether or not his team wins, but he does have control over whether his team loses. Wins, for pitchers, is the by-product of not-losing, not the other way around. The pitching stat to go with is Losses. Or maybe L+BS.
Also, your chart is interesting! I’d be really curious to see it with the time frame expanded. Both 2018 and 2017 seem like odd ducks, in their own way. My guess is that Bullpen Revolution has led to more and more managers pulling their starter when the game is tied/close than in the past. Is there some kind of pitching-change-per-leverage-index stat for managers?
Yeah, I was thinking a similar idea like “Crappy Start” being defined as IP < 5 and ER > 3. Combine Quality Start (+1) with Crappy Start (-1) for a net category.
we’ve been doing my league for around 12-13 years now and a few years ago we changed to the 5 pt split of wins and Qs’s. I don’t really know a better way to do things from a fantasy perspective but am all ears for any ideas. I try to make the league as fair as possible. My though process behind the QS was it was more indicative of how a pitcher actually pitched. The way baseball is going as stated in the article i just dont know how to treat pitchers in H2H fantasy anymore
If the complaint is that QS are a dying breed, and Ws offer attractive playability via strategy, including the ability to incorporate RPs in a manner representative of their current influence on the game…why the heck do you need a new stat? Just use the Ws.
Meanwhile, the rest of us who value non-random stats that ALSO indicate positive performance & meaningful contributions will be over here…using QS.
I value non-random stats that also indicate positive performance and meaningful contributions….but I already made the decision back in August that I’d be moving away from QS. I can’t stay with a stat that is going to plummet in occurrence over the next two years. The opener isn’t going to stop anytime soon. I’m not going to go to wins; my preference is IP, but I’m also open to W+QS. So, I totally hear what you’re saying, but to bury your head in the sand to the evidence that it won’t be a good roto stat moving forward isn’t a smart move.
Still waiting for someone to explain why the coming reduction in QS (which hasn’t been proven yet, but let’s just say for argument sake…) means that it has to be abandoned. There’s been reductions & fluctuations in a host of stats used for FBB over the last 3 decades, yet we still use all of the traditional stats. Forgive me for failing to jump at the siren of the “Millenium Bug.” But, well, I actually lived through it. Much ado about nothing.
I’m looking at the stats on ESPN and get 2,361 total QS for 2018. That’s just adding the totals for each team. I’m confident that any input error is minor as they list an average per team of 79.
If the 2018 total is 365 QS higher, does that change your thesis? Or does the potential trend convince you despite the slight uptick in 2018 from 2017?
Can you send me a link to the data source? As a matter of course, I trust Baseball Reference a lot more than ESPN.
http://www.espn.com/mlb/stats/team/_/stat/pitching/seasontype/2/sort/qualityStarts/order/true
Very weird. Here’s my data source.
https://www.baseball-reference.com/leagues/MLB/2018-starter-pitching.shtml
My best guess is BRef is using the 6 or more IP and 3 or fewer R definition and ESPN is using 6 or more IP of 4.50 ERA or lower.
I checked Arizona’s numbers. ESPN has then listed at 86 QS. B-Ref has them at 78.
ESPN’s individual stats have the following ARI pitchers with QS:
1. Greinke – 21
2. Corbin – 19
3. Godley- 14
4. Buchholz- 10
5. Ray – 7
6. Koch – 6
7. Walker – 1
Those seven pitchers combine for 78 QS.
ESPN says there were 240 more QS in 2018 than 2017 which does not pass a smell test. I think they’re actually double counting somewhere.