Just looking at the names in the quote above; is E Diaz more like Kimbrel/Miller/Rivera, or Speier/Pomeranz/Cecil/Linebrink?

How would the analysis look if you limited it to guys with, say, a career ERA+ of at least 125?

-was Drew Pomeranz a reliever? By what criteria?
-the Miller deal resulted in 9.1 WAR if I count correctly over the first 3 years. Nuthin wrong with that.

I guess the lesson to learn from history I would say is, don't give long contracts to relievers if they aren't elite.

Well free agency is underway with the official report of Keith Law hating a signing. It is almost impossible for a team to make a good free agent signing. The reality is that any FA is going to get a deal that will include some probable dead money. The Mets want to win in 2023 and Edwin Diaz on their roster makes that more likely than not. If I were a Met fan I'd be thrilled with this deal.

The Mets want to win in 2023 and Edwin Diaz on their roster makes that more likely than not. If I were a Met fan I'd be thrilled with this deal.

That's true if there's no budget. If there's a budget, we have to see what else they do, or don't do, b/c of this signing.

My preference would be to never, ever pay a reliever more than say $6M p.a. They're volatile as hell and you can find them anywhere. If there's one place on your team to go cheap, it's the pen.

snaoper, if 6M is your max for a reliever, what is it for a starter? Starters throw 2.5 to 3 times the IP, with less leverage; surely they can't be more than twice as valuable, so you max out at 12m? Starters are equally injury-prone, and over 200 IP may be just as volatile as relievers over that span.

snaoper, if 6M is your max for a reliever, what is it for a starter? Starters throw 2.5 to 3 times the IP, with less leverage; surely they can't be more than twice as valuable, so you max out at 12m? Starters are equally injury-prone, and over 200 IP may be just as volatile as relievers over that span.

Starters are almost certainly many times more valuable. There are very few guys who can throw 175 IP at league average quality or better. there are literally hundreds of guys who can throw 50 IP at league average or better. They appear and disappear constantly. If the best SP in the league gets $40M per, the best RP should get $10M. Would you really rather have two Edwin Diaz's than one elite SP?

Leverage is an illusion. The best RP on the team is always really good. Having an expensive guy locked into those innings, is just value shifting. Lots of time he's gets those high leverage IP even when he's the 4th or 5th best RP on the team/. Much better to have a flock of anonymous RPs.

Zach Britton’s age 28 season (2016) was better than Diaz this year. Would you have paid $105 million for Britton’s next five years?

Leverage isn't an illusion, but 150-200 IP at 1x leverage is more than twice as valuable as 50-70 IP at 2x leverage.

In 2019, the Mets had a terrible closer who went 2-7 with 7 blown saves and the Mets finished 3 games out of the Wild Card. Last year, they had a great closer who went 3-1 with only 2 blown saves and they won 101 games. The problem is both of those closers were Edwin Diaz. I feel reasonably confident that we'll get 2020-2022 Diaz rather than the 2019 version, but there's always that risk.

Leverage isn't an illusion, but 150-200 IP at 1x leverage is more than twice as valuable as 50-70 IP at 2x leverage.


Leverage is an illusion in that someone good is always taking those high leverage innings. Almost no teams have zero very good RPs. Nowadays, most have 3-4. Basically the only time you get a bad pitcher throwing lots of high leverage innings is when you have an expensive "proven closer" who sticks in the role when he should be demoted.

Deep and good bullpens are much more critical with starters throwing less and less. You basically need to sets of relievers to alternate each day. You need two middle guys who can throw 6-8 with one being a primary set up guy and a closer. And then you need another duplicative set. St. Louis deployed this arrangement in 2021. They had a a set of good relievers and a set of bad relievers but they pitched in a straight hierarchy.

I think high end relievers are the new market inefficiency.

I think high end relievers are the new market inefficiency.

High-end, maybe, but not high priced. The article talks about how almost all long/large contracts to RPs have been disasters.

I'm mostly with Snapper on this except the Mets should be a team with a really high budget that can splurge on now-WAR instead of having to mess around and cobble together a bullpen. The edge for the Meta should be that they have the money to throw 40 million or so at some relievers and still go out and find a bunch of young pitchers to pitch relief innings.

#2 When he first studied the matter (back in the 80s) Zimbalist found that only 11% of free agent contracts were good for the team from a revenue point of view and that as a group they were paid about 33% more than they were worth from a revenue point of view.

And now with marginal revenue a smaller piece of the pie I suspect that it's become slightly worse.

Having not studied it, the Mets should be trying to bring back DeGrom, and looking for innings eaters to keep Scherzer/DeGrom fresh and for when they likely miss time.

Also, Diaz is only 28, sure relievers can have variable results, but he's unlikely to collapse. The guys in the article didn't have down seasons until age 31 or 32. Even Diaz's "bad" year looks better by FIP and XFIP.

I copied this over from my post in the other thread. Various "big time" closers for ages 29-33 and Diaz for the last 5 years:

Mo 342 IP, 203 ERA+, 14 WAR
AC 227 IP, 155 ERA+, 6 WAR (reduced a bit by 2020)
CK 227 IP, 163 ERA+, 7 WAR
KJ 296 IP, 158 ERA+, 6.5 WAR (not sure why he's lower with all those extra IP)
TH 354 IP, 162 ERA+, 11 WAR
EG 114 IP, 104 ERA+, 1 WAR
FR 315 IP, 129 ERA+, 6 WAR
HS 205 IP, 130 ERA+, 4 WAR

Over the last 5 years for Diaz (roughly overlapping Chapman, Kimbrel, Jansen but 6 years younger) was 282 IP, 141 ERA+, 8 WAR -- 3 outstanding years, 1 bad year, 1 crap year.

Rightly or wrongly, either WAR is not a good measure for closers or teams have decided closer WAR is worth a lot more. As Diaz's numbers show, his track record already doesn't meet the standards that Law is applying (although arguably that's due to 2020). In fact, only Mo and Hoffman meet the 2 WAR per year line and we may never see that again because closers no longer throw 70-75 innings a year.

Obviously Diaz timed this well -- he wouldn't have gotten this contract after last year's 117 ERA+ -- so it's likely the Mets overpaid pretty badly here but the Mets will be happy if they get their 6-8 WAR out of him.

Shouldn't the comp be to similarly dominant short relievers and how they performed from ages 29 through 33, rather than other guys of various ages and various talents who got big contracts?

Quickly looking at ERA- and FIP-, Rivera, Papelbon, Britton, Robertson, Jansen, and Gagne look like a good cohort. That's a bit of a mixed bag. For what it's worth, ZIPS puts him at 9.2 WAR for the 5 years of the deal with a value of $69 mil.

I don't mind the overpay, but I do mind that it wasn't the issue so the money is really more needed elsewhere most likely. Honestly, fuck the Cubs for their unwillingness to move Contreras.

As a mets fan, I don't care about the money.
Cohen can spend whatever he wants as the cap tax is not a factor for him supposedly.
If we get 3 plus years for Diaz, who cares if they eat the money in years 4 and 5.
It ain't our money.

Exactly

#2 When he first studied the matter (back in the 80s) Zimbalist found that only 11% of free agent contracts were good for the team from a revenue point of view and that as a group they were paid about 33% more than they were worth from a revenue point of view.

And now with marginal revenue a smaller piece of the pie I suspect that it's become slightly worse.

Wow. I expected it to be bad, but not THAT bad. With only an 11% win rate and an expected loss of 33%, you've got to really expect a free agent to drive up your franchise's value for it to be worthwhile to sign one. It's a surprise that more teams don't go full Marlin.

My preference would be to never, ever pay a reliever more than say $6M p.a. They're volatile as hell and you can find them anywhere. If there's one place on your team to go cheap, it's the pen.

Or not. Let's say you sign one Diaz at $20 M, one top set-up guy at $10 (that's about what they get) and one "good 7th-inning" guy at $6. You've just spent $36 M for 175 innings of 140 ERA+. We know what that goes for on the SP market -- either 9/$320 (Cole) or 2/$85 (Scherzer) depending on age.

Of course in neither case are you guaranteed to get all those innings or that good of a performance -- the chance to get the innings is probably higher for the relievers while the chance of getting the quality is probably higher for the starters.

But also, none of this is a mystery. The Mets can sum a few lines of b-r just as easily as I can, they know Diaz had only 8 WAR over the last 5 years and they probably know they'll be lucky if he's equally as good over the next 5. They, and many other teams, have either decided WAR is a bad measure for relievers or that reliever WAR is worth more than regular WAR.

Starters are almost certainly many times more valuable. There are very few guys who can throw 175 IP at league average quality or better. there are literally hundreds of guys who can throw 50 IP at league average or better. They appear and disappear constantly. If the best SP in the league gets $40M per, the best RP should get $10M. Would you really rather have two Edwin Diaz's than one elite SP?

I think this is a misunderstanding of what "league average" really is for relievers. There are at least two tiers of relievers -- leveraged relievers and innings-eaters. These days, the innings-eaters are the AAA/waiver shuttle -- bring em in, have em eat 7-10 innings over 2 weeks (a 90-130 inning pace), ship em back out.

The standards for leveraged reliever are very high though. I suspect that if you look at the guys brought into higher-leverage situations, you find an average ERA+ (or FIP+ or whatever) around 120, maybe higher these days. (More on projections in a bit) Team up Diaz with 2 "average" leveraged relievers and you've got about 175 innings of maybe 130 ERA+ pitching -- a bit better than Yu Darvish who got 6/$126 back in 2018. Should you be paying Diaz about the same as Darvish? Probably not. Would it be OK if those three relievers cost you about $25 M per year with at least 2 of them not signed for more than 2-3 years -- that makes plenty of sense ... or at least as much sense as signing a 2023 Darvish for 6/$140-150. It's not so much that relievers are overpaid, it's that the distriution of that money among relievers may not be efficient.

I agree, Darvish's ability is more rare than those 3 relievers collectively. But "relievers" these days are pitching 41% of all innings ... and it must be close to 100% of all higher-leverage innings. Theoretically then you should be spending about 40% of your pitching budget on the bullpen. If we take leverage ot its extreme, maybe half of those reliever innings have essentially no value at all but those other half get valued at maybe 1.5 times as much -- if Darvish's 175 innings of 120 ERA+ are worth $20-25 M then 175 innings of 120 ERA+ hi-lev relieving might in theory be worth $30-35 M. This seems consistent with (some rich) teams' thinking.

Then there's WPA. I'm not a big fan and I suspect it's very much not good applied to closers but some of us nerds seem to like it. In 2018, 2019 and 2021 Darvish had essetially 0 WPA. (he had heaps in 2020 with that great performance and nearly 4 in 2022 for some reason ... it was just a standard Darvish season.) Still, Diaz had the same WPA as Darvish in 2022; he had nearly 5 WPA in 2018 ... and about 0 for the 3 years in between. Anyway, over the last 5 years, Darvish has averaged 2 WPA/162; Diaz has averaged 2 WPA/162. Are we sure they shouldn't be paid the same?

What I'd really like to do is sum the WPA across those 3 hypothetical relievers. But if Diaz is at 2 then presumably the other 2 guys are also positive ... although I'm not sure you can just sum WPA. The reason I don't like WPA for closers is because so much of the WPA is just the usage -- you are entering a game which your team has a 75% chance of winning and you get credit for the last 25% (a lot) because the manager brought you in for the 9th not the 8th and the 8th-innning guy pitched as well or better. But if we could look at that collective reliever, then it doesn't matter which of them individually gets the save. That's just back to the point that reliever salary may not be efficiently distributed.

Looking at the 2022 Mets, Diaz had 3.9 WPA, Ottavino 0.8 and the other relievers were all near zero. Diaz's 3.9 WPA is #1 on the team (Scherzer 3.4) and the gap is even larger in "championship" WPA. The pair account for 4.7 WPA. If WPA and WAR are priced about the same (I'm not sure WPA is priced at all nor am I sure it should be) then spending $30 M per year on them would seem reasonable. Alternatively, over the last 5 years, the pair have averaged about 3 WPA combined which might be worth $25 but you're the Mets, you can afford $30.

Back to the rarity argument -- sure, it's harder to find one Darvish than one Ottavino. But you need to find three Ottavinos -- or a Diaz, an Ottavino and a reliable guy. That evens the odds substantially. The latter is probably still easier but not outrageously so. Again, sure, Diaz probably isn't worth twice as much as Ottavino ... but Ottavino is quite possibly worth more than $10 per year.

Finally, though I'm sure not my last post of the thread -- it is pretty clearly easier to find a Scott Effross floating around for free than it is to find a Darvish. But you've also got to sorth through 3-4 guys to find an Effross, a period durig which those other guys are wiping out Effross's contributions and then some. Unfortunately, I consider projecting relievers to be something of a fool's errand -- we talk about needing 3 "full seasons" of a SP to make a good prediction (which is not really all that good), say at least 500 innings. That's 8 or 9 "full seasons" for a reliever which, even if they survived that long, you wouldn't want to combine stats from age 24 with age 32. The chance that Effross is a fluke is quite high; the chances that Diaz's 2022 was a fluke are extremely high. In other words, I wouldn't want put a lot of faith in a 2023 Taijuan Walker projection but at least he has 900 ML innings; Diaz just 400; Effross just 71 -- they're all about the same age.

Key summary: Diaz as an inividual is almost certainly overpaid but maybe not extremely so but bullpens as a whole are a place where investing substantial resources is fine. At the very least, a distinction needs to be made between hi-lev relievers -- the performance standards are quite high -- and the rest with quite low/non-existent performance standards. Team reliever spending makes more sense if evaluated on the amount paid for their hi-lev relievers and compare the collective quality of those relievers to an equivalent starter. Alas there is the very real question if we can reliably identify the consistently good relievers at all or is it all flukes as far as we can tell.

Starters are almost certainly many times more valuable. There are very few guys who can throw 175 IP at league average quality or better. there are literally hundreds of guys who can throw 50 IP at league average or better. They appear and disappear constantly. ....Leverage is an illusion. The best RP on the team is always really good.
--
No, actually, the notion that 100s of bullpen guys can toss a great 50 IP is the *real* illusion. In hindsight, sure, but not in foresight. It is an illusion of small sample size. Pick up stats near the end of May when SPs have 50 IP, and many of them will look awesome. You can't PREDICT bullpen guys to go 1.8 ERA. You can project them to be 1.2 WAR types, which when slightly leveraged (I agree, don't overdo it) might be 1.5. The best SPs will be 4.0 WAR guys. So there are 2-3 times as valuable. I would give Diaz 10 million, but an ace SP about 25-30 mil.

Over the last 5 years for Diaz (roughly overlapping Chapman, Kimbrel, Jansen but 6 years younger) was 282 IP, 141 ERA+, 8 WAR -- 3 outstanding years, 1 bad year, 1 crap year.

Rightly or wrongly, either WAR is not a good measure for closers or teams have decided closer WAR is worth a lot more. As Diaz's numbers show, his track record already doesn't meet the standards that Law is applying (although arguably that's due to 2020). In fact, only Mo and Hoffman meet the 2 WAR per year line and we may never see that again because closers no longer throw 70-75 innings a year.

One quibble: Why do we care what Diaz did 4-5 years ago? We should be looking at the most recent 2 to 3 years to project how he'll do in the future. Over the past 3 years, of relievers with at least 100 IP, Diaz leads in fWAR, FIP-, and is 5th in ERA-. His numbers stack up well against dominant closers like the ones mentioned above.

Wow. I expected it to be bad, but not THAT bad. With only an 11% win rate and an expected loss of 33%, you've got to really expect a free agent to drive up your franchise's value for it to be worthwhile to sign one.

The devil is always in the detail on those things ... and I'm not gonna try and find it nor quickly develop one of my own. A few things though:

1. The winner's curse in auction: a well-known, largely inescapable phenomenon. In any open/blind auction (for a rare/unique item), essentially by definition, the winner is the bidder who most values the object. Even leaving aside owner ego and team PR needs, that means the team that projects the most WAR or the most revenue returned will usually be the one that signs the player. If the Yanks think Judge will produce 25 WAR for the rest of his career and the Mets think he wokk produce 30 then the Mets rock up with an extra $40 M in their offer. The most optimistic team is almost certainly wrong. (When combined with the even better-known, even more inescapable phenomenon known as the Mets, this will always be the case.)

1A. In trades, the team that gets Juan Soto is the one that gives up the most (perceived) talent. But there at least there's a substantial lack of information on both sides -- services for money is quite a bit more straightforward from Judge's side of the FA auction.

2. So while 11% might still be very low, we should expect most FA deals to not break even.

3. The winner's curse mainly operates for the elite FAs. Team A might value average OF X more than Team B does but even Team A knows that average OF Y is pretty much the same player. Even more for bench players. So if it's just 11% of all FA deals, I'd be surprised if it's more than 5% of the big deals. On the other hand, no team thinks an average guy much less a bench player or a formerly anonymous reliever adds any revenue. Nor do they so maybe these are all those $3 M contracts are "bad."

4. The deferment angle. Everybody knows going in that the last 1-3 years of a deal (depending on length) will be "overpaid" and are really pretty much a crapshoot in terms of what will be produced. Our projections for 2023 aren't all that good, the ones for 2023-25 are quite dicey, 2026-28 may be waste of half-seconds of CPU time. As #17 suggests, what teams/fans really, really, really want is for the player to be the same good player they are right now for the first half of the deal. If they get that, the team will be content ... and the fans will turn on the player as soon as he declines and whine about what idiots the FO are for wasting money on that guy and they should have signed him for 3 years, not 6.

Cano is a classic example. Nobody expected Cano to be a good player at 40 (still a year to go!) ... or even 38. That wasn't a good contract but it was a contract priced as if he'd produce ... 35 WAR. (I don't remember what we thought the $/WAR figures were back then.) I think I finally talked myself around to thinking he'd end up around 30 WAR. Regardless of how much you expected, we all figured he'd produce about 2/3 of that within the first 5 years then (hopefully) decline steadily from there. Call it 20/10 with those last 5 years maybe going something like 3, 2, 2, 1, 1.

So the Ms got 24 WAR in those first 5 years -- 2 really big ones, 2 solid years, one potentially big year interrupted by a roid suspension. 24 WAR for $120 M is a massive return, even by the $/WAR standards of the day. The Ms were very happy I'm sure ... and also awake enough to know they wanted to get out from under the 2nd half of that contract. Even the Mets knew the last 5 years of Cano would not be worth $120 M so between eaten contracts and cash, the Mets made the Ms eat about half of that money. The Mets decided to pay about $60 M for what they hoped were the last 7-8 WAR Cano had, presumably hoping he'd put up 4-5 WAR in year 1 (he'd just put up 4 in a half-season). That didn't go well as we all know.

So from Zimbalist's perspective, I assume that comes out as a loser -- $240 M for 24 WAR at 2014 prices was definitely an overpay. But what would have been a fair price for 24 WAR over 5 years? Something in the $175-200 range? Now treat it as deferred payment of that $175-200 M, add in possible lux tax implications, possible payroll management advantages and the probability that you might escape some of that back-end money?? In the end, the Ms got 24 WAR for about $180 M spread over 10 years. The Mets got a terrible deal (but did get Edwin Diaz). Was that a "bad" contract? Not from the M's perspective.

an ace SP about 25-30 mil.

One quibble: Why do we care what Diaz did 4-5 years ago? We should be looking at the most recent 2 to 3 years to project how he'll do in the future. Over the past 3 years, of relievers with at least 100 IP, Diaz leads in fWAR, FIP-, and is 5th in ERA-. His numbers stack up well against dominant closers like the ones mentioned above.

Because 5 is a magic number. Also the question "what will Diaz do for ages 29-33?" leads to "what did other similar relievers do for ages 29-33" which leads to "for comparison, how has Diaz done for the last 5 years?" Turns out that for the last 5 years, he's done about as well as those other non-Mo guys so the Mets probably shouldn't expect more.

More seriously, for everybody, there is still useful information 4 years ago although year 4 isn't weighted very heavily. But two is certainly a bad limit. But for relievers, the samples are simply too small (esp if one is 2020). 2-3 years of Diaz is 110-170 innings. Three years for a reliever has a bad signal/noise ratio. It's certaily possible that adding 2 more years doesn't help but, in general, more data means less noise but, sure, it can "bias" the signal if there's been a significant change in the underlying signal over time.

The whole "we need three years of data for accuracy but more than 3 years adds little/no value" is actually based on "we need three full years of data for accuracy and, if we have that, then more than 3 years isn't of much value." Whether older data is more helpful when recent data is more limited is not a quesiton anybody's tried to answer as far as I know.

If I was head nerd for a team, I'd have my minions working on developing a model for relievers that focuses on components (velocity, movement, K, BB, etc.) to see if we can come up with something that will work better in small samples.

For Diaz in particular, the last 3 years is 150 innings (because 2020) which we quite correctly would consider not enough data for an accurate projection for a SP ... it doesn't become more accurate because we're only projecting his next 50-150 innings. That overall performance is excellent (176 ERA+, lots of Ks) but does include a year with a 117 ERA+ which is not a good "year" for a hi-lev reliever. For Diaz it's also a convenient endpoint -- includ his 2019 and that ERA+ falls all the way to 126!

That is not good from a data analysis perspective. If your original conclusion was sound then adding a "small" amount of new data should not shift your conclusion dramatically. If it does then you have to have less faith in your earlier conclusion unless you have a really good reason why that extra data definitely should not be included. Injury is a good reason but he wasn't hurt as far as we know. Somebody will always have an excuse (he changed his arm angle, he broke up with his girlfriend) and those should be ignored unless somebody has actually done an analysis. To hold any water at all in Diaz's case, I think somebody would have to build a good case that EV and HR rates are very flukey or "settle down" very quickly such that we can ignore those 2.3 HR/9 that year and be really confident that's irrelevant to his projection.

Now it turns out that adding another small bit of data (year 5) brings the ERA+ back up to 141, about 1/3 of what he lost adding year 4. That conveniently comes out very close to his career ERA+ ... but then the last 5 years is also about 2/3 of his career innings. We have the following pictures of Diaz (ERA+, K/9, HR/9, WAR/162)

career 139, 14.8, 1.0, 1.8 (400 IP -- still not really to 3 "full" years of data)
last 5 141, 15.3, 0.9, 1.9 (280 IP)
last 4 126, 15.4, 1.0, 1.6 (210 IP)
last 3 176, 15.4, 0.5, 2.5 (150 IP)
last 2 166, 14.9, 0.4, 2.3 (125 IP)

K-rate stable, HR rate shifted from 1 to 0.5, ERA+ bouncing around. He has two big WAR years, one atrocious one, 4 ranging from 0.8 to 1.4. One could argue to ignore that 74 ERA+ and then you have a 156 in his first 3 years and 176 in his last 3 years so maybe 166 overall. His HR/9 those first 3 though was still 0.9. It's not unreasonable to conclude that 2019 was just about the flukiest fluke ever and should be ignored; it's even more reasonable to point out that's a bit too convenient a conclusion and we'd better get more confidence it can be ignored.

So after all that you might start to feel comfortable and the main question you want answered is the HR/9 projection. So I'll get my minions working on that.

Anyway, feel free to use/not use whatever data you want. Your conclusion will be "Edwin Diaz has often pitched like a dominant reliever in the past but sometimes he has not." That period of dominance might fall short of Mo but it's very good. And we'd have said the same things at variouss times about Chapman, Kimbrel, Jansen, Gagne, K-Rod, Street, Papelbon, etc. and some of those guys continued to do that and some were off/on and some collapsed. A crude average of the non-Mo/Hoffman relievers in my above-table comes to about 250 innings, maybe a 140-145 ERA+ and about 6 WAR which, flukily enough, is almost exactly what Diaz has done over the last 5 years. Relievers with "amazing" 3-year runs are a dime a dozen (OK, 2.4 billion dimes a dozen but what's a little rounding error) -- the ones who have another 5 amazing years after that are not so common.

The article talks about how almost all long/large contracts to RPs have been disasters.

Before Law's dataset, but the Braves' final payment to Bruce Sutter (RIP) was this year. In total, they gave up 47 million over 38 years (no, I'm not calculating the 1984 present value of his contract).

His six-year deal quietly ended with 152 IP and a 4.55 ERA. Ouch.

I was wondering if maybe Law had missed a Hoffman or Rivera deal in his study/research, but the Yankees were constantly smart and never signed Rivera for more than three years, and Hoffman only signed a two year contract with the Padres as a free agent with a vesting option for the third. Three years would be about what I would be willing to offer as a GM, and maybe accept a four if it's necessary to get the deal done, but I've also been in the camp that relievers as a general rule don't have more than a 6 year shelf life, although a few pitchers over the past decade has made me less confident in that prediction.

...but the Yankees were constantly smart and never signed Rivera for more than three years

I know what you mean by this, but with hindsight we can see that the Yankees would have saved money if they signed him for a big long deal in 2000 rather than the actual shorter deals.

#19 Thing is that most signing aren't terrible. They're meh. And meh that addresses a specific need of a team in the general range of playoff contention is ... well perfectly fine. It's OK to pay a premium with eyes open.

If it matters, there's one specific type of signing that fared consistently poorly. The early 30s non-star. A significant number slip that little bit and are just useless.

Well that and pitchers who blow their arm out, but I think everybody knows that going in.

Do reliever studies from years ago really explain what is happening now? It seems the game is played much, much differently now (especially in the NL with the DH). You have shifts, the three man rule, the minor league shuttle, the 26 man roster, the larger bullpen staffs etc. reliever usage seems to be much more of a science now and I would therefore expect that the theory that relievers are fungible is no longer as true as it was and performance is a bit more steady and predictable. It seems to me that while there is a lot of turnover in the non high leverage arms, that the high leverage guys seems to be sustaining year over year performance.

Given that Diaz mixes merely good years in with great ones, and even had a total clunker of a year in 2019, we can safely say there is a lot of risk in this contract. They'd have been safer offering 5/200 to DeGrom in my opinion, because they could buy insurance to cover some of the "missed innings" risk in that contract and Diaz seems to stay on the field, though not great all the time.

Regardless, as a Phillies fan, I'm thrilled they're wasting money on Diaz. I'd have topped out a 3/66 for Diaz and told him to take a hike if he wanted the 4th and 5th year. I don't mind the $$$ figures abstractly with Cohen's money. But, 5 years is too long for ANY reliever imho, unless you get a guy to sign a 5-7 year deal cheap when they're youngish (7/20 for a guy just going into arbitration for example -- but, even then you're going to get burned sometimes).

For Diaz in particular, the last 3 years is 150 innings (because 2020) which we quite correctly would consider not enough data for an accurate projection for a SP ... it doesn't become more accurate because we're only projecting his next 50-150 innings. That overall performance is excellent (176 ERA+, lots of Ks)

sure, if you consider 257 K in 150 IP to be "lots."

:)

relievers from as little as a decade ago were more dependent on team defense.

Diaz in 2022 got 118 of his 186 outs via K. meaning, if his SS isn't good at turning DPs, it doesn't matter as much in this era. Diaz does more of the work himself. I think that's relevant.

Regardless, as a Phillies fan, I'm thrilled they're wasting money on Diaz.

not nearly as giddy as I was when the Phillies signed Ryan Howard to that longterm contract - that one was sort of a suicide pact.

Regardless, as a Phillies fan, I'm thrilled they're wasting money on Diaz.

Alvarado.

More seriously, for everybody, there is still useful information 4 years ago although year 4 isn't weighted very heavily. But two is certainly a bad limit.

Based on what? We may want more than 100-150 IP of a sample, but that doesn't make the older data any more relevant.

Whether older data is more helpful when recent data is more limited is not a quesiton anybody's tried to answer as far as I know.

Right, so we shouldn't assume it's useful.

For Diaz in particular, the last 3 years is 150 innings (because 2020) which we quite correctly would consider not enough data for an accurate projection for a SP ... it doesn't become more accurate because we're only projecting his next 50-150 innings. That overall performance is excellent (176 ERA+, lots of Ks) but does include a year with a 117 ERA+ which is not a good "year" for a hi-lev reliever. For Diaz it's also a convenient endpoint -- includ his 2019 and that ERA+ falls all the way to 126!

Yes, it's a convenient cutoff, and that would be a problem if we made the cutoff there because it made him look good. But we make the cutoff there because older data doesn't have much if any value. In fact, 4 years is the only cutoff in the past 5 years that makes him look anything but great. (Also, averaging in ERA+ from 4 years ago, and weighting it equally to more recent years, is especially problematic.)

If I was head nerd for a team, I'd have my minions working on developing a model for relievers that focuses on components (velocity, movement, K, BB, etc.) to see if we can come up with something that will work better in small samples.

I consider you the head nerd for team BBTF, so I think you should be able to assign this to any of your minions here.

If I was head nerd for a team, I'd have my minions working on developing a model for relievers that focuses on components (velocity, movement, K, BB, etc.) to see if we can come up with something that will work better in small samples.

You don't think teams are doing this stuff already? And more?

I have to assume, maybe incorrectly, that teams have far more comprehensive analysis of players individually then we have access to and also their own proprietary stats, etc that they prescribe to regarding all matters, large and small.

#34 Started happening a while ago. The guys who used to do this as a hobby (I think Keith Woolner was the first internet guy to be hired by a team) have been increasingly hired by a team -- and no public disclosure.

Used to be that places like rec.sport.baseball were the cutting edge of research but that hasn't been true for decades. You do get some very good stuff every now and then, but the teams now are way ahead of what's publicly available.