Sports takedom is a firehose of malarkey. Truly just a deluge of hot air, hogwash, and horsefeathers. Hokum and hooey. You know, bullshit. I’ve been doing this semi-professionally for a year now, and while I try to be more rigorous than your replacement-level hucksters, I really don’t have a good sense of whether I have any idea what I’m talking about when, for instance, I say that I believe in Jarred Kelenic.
Philip Tetlock and Dan Gardner’s great pop-science book Superforecasting has some advice about this. They say that a lot of good predictionistas have these things in common:
- They make specific, falsifiable predictions. “Kelenic’s going to hit at least 20 home runs” is falsifiable. “The Mariners will trade for an impact bat” is not because what’s an “impact bat”? You can’t improve if you give yourself too much wiggle room to say that #actually, you were right.
- They think probabilistically. This means focusing less on whether a prediction is right, and focusing more on how likely something is to be right. If you give 10 events an 80% certainty, then even if all 10 things happen, you’re doing a bad job. Even though you were “right” about everything, you did a bad job of estimating how likely things are to happen. You should have been more certain. Projection systems work probabilistically, and the people who manage them (at least the good ones) are always tinkering to make sure they’re well calibrated. When the FanGraphs playoff odds gave the Guardians just a 7.5% chance to win the division to open the 2022 season, they weren’t wrong. They make a lot of projections. You want things with a 7.5% chance of happening to happen occasionally. 7.5% of the time in fact. People think probabalistically all the time, like when they play poker and consider how likely they’ll get a card they need or when they’re driving and consider the odds that they’ll got caught for speeding. (Don’t speed; it’s more dangerous than you think.) The difference is that people, in contrast to the machines that process the projection systems, aren’t usually in the habit of actually putting a number on likelihoods. But if you do it a lot, you’ll learn to get a lot more precise, a lot better at predicting the future. If this is new to you, here’s some further reading and watching.
- Most crucially, and I can’t stress this enough, they look back at how they did. This is the key to getting better at predictions. This is how to avoid bloviating. People who are good at predictions think critically about why they were wrong so they can get better. They also think about the types of things they were wrong about; if it turns out you just aren’t that good at predicting what will happen in the playoffs, maybe you should stop making confident assertions about the playoffs. And they look back to ensure that they’re calibrating well. Are they getting 60% of the things they predict with 60% certainty right? Then they’re doing a good job even though two out of five of those predictions are “wrong.”
So I’m going to try this out this year. I’ll still participate in LL’s staffwide predictions post with some of my gut takes. That’s a place for fun “I just have a feeling” type takes. There’s plenty of space for a bold take that Julio will hit 50 homers and steal 50 bases. This exercise, by contrast, is meant to discipline my actual thinking.
Below, I’ve made 10 predictions that I think are 60% likely to happen, and 10 more at each of 70%, 80%, and 90%. Some questions you might have about this:
- Why these 10% intervals? Normal humans aren’t capable of better precision than this. Maybe if you get really, really good, you can split the hairs at 5% intervals instead of 10%.
- Why no predictions at 50%? Because if you’re saying it’s a coin flip, you’re not really prediciting anything. Thinking something is as likely to happen as not is equivalent to having so much uncertainty that you shouldn’t be making prognostications about it—it’s the kind of thing I’m trying to stop doing.
- Why nothing at 40% or less? Because if you’re good with words, you can frame anything in the positive. “There’s a 40% chance that Luis Castillo leads the Mariners pitchers in fWAR” can be rephrased as “There’s a 60% chance that Luis Castillo does not lead the Mariners pitchers in fWAR.”
- Why 10 predictions at each of these intervals? So that the lookback is tidier. I’ve made 10 predictions with 70% confidence. If I’m doing a good job calibrating, seven of those predictions will come true. It’ll simply be easier to track my progress this way.
I’ll return at the end of the season to see how I did. To be clear, I expect to be really bad at this. The point is that I hope to get better.
You can play too! Here is a link to a survey that lets you vote on what you think the right probability is for each of these predictions. (It asks for email to prevent ballot-stuffing, but we are not collecting the email addresses.) Voting closes on March 17, and I’ll come back with a post the following week sharing the survey results. Then I’ll incorporate your predictions in the year-end lookback. Here’s the link again.
Without further ado, here are my predictions. All predictions are for the regular season unless otherwise stated. Here’s to keeping baloney on sandwiches and off my keyboard.
Mariners hitters performance
- Julio hits at least 22 home runs (90%)
- Julio steals fewer than 30 bases (70%)
I feel certain he’s capable of more, but they started being much more cautious with him after he jammed his hand, and I bet that trend continues.
- Julio leads the Mariners position players in fWAR/PA (90%)
- Ty France hits fewer than 30 home runs (80%)
He’s never done it before. One of the big reasons why is still a probable issue this year: his proclivity for getting hit by pitches. Even if it doesn’t take him off the field, we’ve seen two seasons in a row where an injury zapped his power for several weeks.
- Toro and Winker combine for more fWAR than Kolten Wong (80%)
Across the projection systems, Toro is projected to have among the biggest gains over last year. Relatedly, he was one of the biggest xwOBA underperformers last year. Winker gets a fresh start after one of the biggest dropoffs in recent memory. He got offseason surgery to boot. Plus, there’s two of them to the one Wong.
- Cal beats his Steamer-projected wOBA of .312 (80%)
- Evan White gets fewer than 400 MLB PAs (60%)
- Tommy La Stella gets fewer than 50 PAs as a Mariner (60%)
I drafted this before his injury. I might give this higher odds if I did this today.
- Mariners have at least five hitters with at least 250 PAs and a wRC+ over 100 (80%)
I’m thinking Julio, Teoscar, Ty, Eugenio, Cal, DMo, and Wong are all more likely than not to hit these marks, plus I bet someone else does too. But the odds that they all do can’t be that great, especially with possible injuries.
Mariners pitchers performance
- At least one Mariner receives at least one Cy Young vote (80%)
This is about how the odds get higher, the more good pitchers the Mariners have. There are five M’s who could plausibly end up getting votes (four starters plus Muñoz). That really increases the odds. I’m such a believer in this staff that I’m on the fence about giving this 90% odds.
- Luis Castillo does not lead the Mariners pitchers in fWAR (60%)
Castillo’s the best bet to lead the Mariners pitchers in fWAR. I’m giving him 40% odds—that’s really good. But between injury risk, the unknowability of pitching, and especially the possibility that Kirby goes supersonic, I can’t put Castillo at 50% or better.
- George Kirby improves his K/BB from last year (80%)
He got better, not worse, as the season went on. Plus I’m just personally very high on Kirby. Watch this space.
- Logan Gilbert finishes with at least 2 fWAR (80%)
Maybe this should be at 90% since he’s put up 3 in each of his first two seasons. But I’m trying to account for the risk that his contact problem becomes unmanageable and for the risk that he spends time on the IL.
- Marco makes at least 10 starts for Seattle (70%); Marco makes fewer than 20 starts for Seattle (70%)
- At least one Mariners reliever puts up at least 1.5 fWAR (80%)
- At least five Mariners record at least one save (60%)
Other Mariners Predictions
- Mariners win at least 80 games (90%)
- Mariners make playoffs (60%)
I’m nervous that this is homer-ism because I haven’t seen a single projection system give them odds this high.
- Mariners make at least one in-season trade (90%)
- One of Kelenic, Trammell, or Marlowe end the season in another organization (70%)
- Teoscar does not sign an extension before the final game of the regular season (70%)
He hasn’t yet is all I’m saying.
- Scott Servais ends the year as the Mariners’ manager (90%)
- Chris Clarke does not end the year with Seattle (90%)
Taking everything about Clarke as an individual aside, Rule 5 picks just almost never stick.
- Mariners have at least three All Stars (60%); Mariners have fewer than five All Stars (80%)
Julio, Muñoz, Castillo, Kirby, France, J.P., Cal, Gilbert, Ray, Brash, Sewald, Teoscar are all somewhere between “a lock if healthy” and “realistic possibility.” Plus they’ll get a bump for hosting the ASG. But five is just a ton of All Stars—even last year’s Astros only had five.
- Mariners have fewer than three players on the midseason update to the FanGraphs Top 100 Prospects (60%)
There are plenty of plausible candidates (Miller, who’s already on the list, plus Ford, Berroa, Young, Woo, and the possibility that anybody could pop). But there’s so much uncertainty with prospects.
Around the AL West
- Astros win the AL West (90%)
You’ve seen the Astros, right?
- A’s finish last in the AL West (90%)
You’ve seen the A’s, right?
- Rangers do not make the playoffs (70%)
Yes, their pitching got better, but the guys they brought in all have significant injury histories.
- Jacob deGrom leads the Rangers starters in fWAR/IP (90%)
- Astros have at least ten players with at least 10 PAs and a wRC+ over 100 (70%)
- A’s trade a reliever in season (90%)
Even leaving aside Trevor May specifically, with the position the A’s are in, if any reliever has any value, there’s every reason to trade him.
- Shohei finishes the year as an Angel (70%)
This is all but certain if either they are competitive at the deadline or if he’s hurt at the deadline. Plus there’s the extra possibility that they decide not to trade him anyway either as part of an effort to re-sign him or because he brings in so much revenue that two months of jersey sales outweighs whatever prospects they’d get. Maybe this should be 80%.
- Orioles do not make the playoffs (80%)
As great as Adley and Gunnar are, and as great as their farm system is, I think they’re still at least a year out, especially on the pitching side. Cole Irvin was an upgrade. Cole Irvin!
- Rays make the playoffs (60%)
- Dodgers do not win the NL West (60%)
I like the Padres to win the division, and there’s the outside chance the Giants pull it off.
- Mets finish the year with a better record than the Yankees (70%)
- Mitch Haniger plays fewer than 135 games (60%)
- The qualified player with the highest wRC+ is one of Judge, Yordan, Soto, Julio, Shohei, Vladito, Trout, Freeman, Tucker, or Mookie (70%)
On the one hand, I feel so sure it’ll be one of these players, but if baseball teaches us anything, it’s to never underestimate the field. Baseball is weird and there are just so, so many players.
One more time, here’s the link where you can cast your own vote.