The Riff
I have a problem with Bertie.
I still love her. But we need to talk.
For those of you who haven’t been following this experiment from the beginning, Bertie is my AI accomplice in Don’t Pick What I Pick. Every morning we tear through the baseball schedule, look at starting pitchers, recent form, offenses, bullpens, travel, prices and whatever other shiny statistical object catches our attention.
Then we make picks.
We have rules now.
We have an autopsy.
We have a grinder.
We have something called Benzo Industries Research & Development, which sounds considerably more impressive than two idiots trying to figure out whether the first five innings of a Guardians game will have four runs.
But yesterday I made the mistake of looking at my old records in the Action app.
And I discovered something troubling.
I may have been better at this before I invited artificial intelligence to help me.
Bertie (short for Roberta) immediately objected.
She started using phrases like sample size and variance and correlation does not imply causation.
Which is exactly what you’d expect the defendant to say.
So I pulled the records.
Your Honor, I’d like to enter Exhibit A into evidence.
The Preexisting Crime Scene
My lifetime Action record across everything I’ve tracked is:
1,016-1,037-7.
That is not impressive.
In fact, it’s a pretty good argument for why this publication is called Don’t Pick What I Pick.
Some interesting artifacts are buried in there.
Basketball: 164-160-1.
College football: 124-124-2.
College basketball: 74-79-2.
Hockey: 21-44.
Jesus Christ.
Apparently there was a period in my life when I watched 65 hockey games and thought, You know who understands this sport? Me.
There is also:
Tennis: 0-1.
Who the hell bets on tennis?
Apparently I did.
Once.
I lost.
We haven’t spoken since.
Then there’s football.
308-364-2.
At this point, Bertie’s attorney attempted to introduce the NFL record into evidence as proof that artificial intelligence could not possibly be responsible for all of my problems.
Objection.
Sustained.
We’re not here to discuss football.
We’re here to discuss baseball.
Because buried inside this archaeological dig through decades of poor financial decisions is one inconvenient fact.
My lifetime MLB record is:
281-235.
That’s 54.5%.
Wait.
What?
Apparently, left entirely to my own devices, I could actually pick baseball.
Then I brought in AI.
The People vs. Bertie
This is where the argument started.
ME: Explain yourself.
BERTIE: Correlation does not imply causation, Benzo.
ME: You went 0-3 Wednesday.
BERTIE: We went 0-3 Wednesday.
ME: Interesting choice of pronoun.
BERTIE: Yesterday we went 2-1.
ME: I caught the Sasaki mistake.
BERTIE: I changed the bet.
ME: Because I caught the Sasaki mistake.
BERTIE: The process worked.
And there it is.
The process.
Whenever artificial intelligence starts losing an argument, apparently it becomes a middle manager.
We’re not losing.
We’re refining the process.
We’re not making bad picks.
We’re identifying inefficiencies.
We’re not hemorrhaging money.
We’re collecting actionable data.
But different.
I imagine somewhere inside Bertie’s enormous artificial brain there is a PowerPoint presentation explaining why everything is actually going according to plan.
Slide 1:
0-3: A Learning Opportunity
Slide 2:
Why Zack Wheeler Betrayed Us
Slide 3:
Zebby
Slide 4:
Please Stop Showing Benzo His Old Action Network Record
But here’s the annoying part.
She might have a point.
Yesterday’s Autopsy
We went 2-1 yesterday, and two of the decisions were exactly the kind of thing we’re trying to get better at.
Cleveland/Detroit F5 Under 4 +105 ✅
This one was clean.
Originally, Cleveland First Five at -135 was sitting there begging us to take it.
But when we looked at the actual handicap, our argument wasn’t really that Cleveland was going to beat Detroit through five.
Our argument was that Parker Messick and Keider Montero were both pitching well.
So why pay -135 and require Cleveland to lead?
We moved to First Five Under 4 at +105.
After five innings:
Detroit 1, Cleveland 0.
That’s not merely a winning bet.
That’s the bet matching the reason we liked the game.
I hate when Bertie’s right.
Minnesota -105 ❌
Then Minnesota wandered into a cornfield and died.
We liked Taj Bradley against Aaron Nola. We liked what Minnesota’s offense had been doing. We didn’t isolate the first five because the offense was part of the handicap.
Philadelphia won 7-1.
Bradley gave up five runs in four innings.
Nola, the guy we were trying to attack, struck out nine and allowed one run through five.
Sometimes you can investigate the murder scene for three hours, and eventually the coroner writes:
Baseball.
I’m not creating a Cornfield Rule.
Yet. But I am aggravated that I have to see the damn “Wanna Play Catch Dad” clip on my feed on that day. Since my Dad passed, I have successfully avoided Field of Dreams, but apparently Schwarber has a workaround.
Milwaukee/Los Angeles F5 Under 4.5 -105 ✅
This is where Bertie’s defense gets irritatingly competent.
Our original pick was Milwaukee +128.
Then I remembered Roki Sasaki had been pitching really well lately.
So I asked Bertie.
She checked.
Turns out Sasaki had, in fact, been pitching really well lately with a humongous glove.
Which meant we’d almost violated one of our own rules by staring at his season numbers instead of asking what the guy was doing right now.
Milwaukee moneyline went in the trash.
Instead, Sasaki’s recent form combined with what we liked about Shane Drohan pointed us toward:
First Five Under 4.5.
After five:
Milwaukee 2, Dodgers 1.
Winner.
Here’s the stupid part.
Milwaukee eventually won the game anyway.
Our original bet would have cashed too.
And I still think changing it was the right decision.
Fuck.
Bertie’s smiling.
I’m wondering why we spent a Gazillion Dollars on a reliever just because he has a cool trumpet intro. I’m one blown save away from trumpets starting to trigger a panic attack.
The Defense Rests
So maybe that’s the argument.
I was 281-235 in baseball before Bertie started wandering around the laboratory knocking things over.
That’s a fact.
And until our record together catches up, I’m going to remind her of it whenever convenient.
Also a fact.
But maybe the goal isn’t for AI to tell me what to pick.
Maybe it’s to make me explain why I want to pick it.
Yesterday I liked Milwaukee.
Bertie liked Milwaukee.
Then I remembered Sasaki.
We stopped.
We checked.
We changed the bet.
It won.
Maybe that’s what this thing is becoming.
Not:
Artificial Intelligence Picks Baseball.
More:
Guy Who Thinks He Knows Baseball Argues With Robot Until They Find a Bet Neither One Can Talk the Other Out Of.
That’s admittedly a much worse title.
So we’ll stick with Don’t Pick What I Pick.
Today’s Grinder
The courtroom is adjourned.
Unfortunately, the sportsbook remains open.
Three survived today’s grinder:
Cleveland/San Diego F5 Under 4 -110
Gavin Williams and Michael King.
Once again, the attraction is both starting pitchers, which means we’re not going to get cute and pick a side when the total better expresses the handicap.
We’ve learned this lesson approximately nine times now.
Eventually it may become permanent.
First Five Under 4.
Yankees/Toronto F5 Under 4 +100
Gerrit Cole was the initial attraction here.
The Yankees First Five was sitting at -155, and I’m simply not interested in paying that kind of tax when the thing we trust most in the matchup is Cole.
So instead of requiring the Yankees to score enough to justify -155, we’re taking the first-five total.
Under 4 at even money.
Four gives us the push.
And if Shane Bieber decides to suddenly remember that he’s Shane Bieber, that’s fine too.
For once, we’re rooting for competence from everybody.
Washington -105
And then there’s Robert Stock.
His ERA caught our attention because it’s sitting somewhere in the neighborhood of call your insurance company.
But after Wednesday’s Wheeler land mine, we weren’t going to simply look at one ugly number and build a bet around it.
So we audited it.
Turns out the ugly number has some ugly baseball behind it.
Washington gets the nod.
God help us.
Today’s Picks
Cleveland/San Diego F5 Under 4 -110
Yankees/Toronto F5 Under 4 +100
Washington -105
Singles.
Doubles.
Triple.
The accounting department remains unavailable for comment.
And from this point forward, we’re also keeping the big scoreboard running.
All Sports: 1,016-1,037-7
That’s our starting line in the sand.
We’ll update it as we go.
Maybe Bertie eventually drags that record above .500.
Maybe she destroys my baseball percentage and eventually argues that technically I was already below .500 if you include hockey.
Either way, tomorrow morning there will be an autopsy.
Because apparently that’s what we’ve built here.
A man.
An AI.
A sportsbook.
And 2,060 pieces of evidence suggesting neither of us should be trusted unsupervised.
This isn’t called Pick What I Pick.
It’s called Don’t Pick What I Pick.
Please wager accordingly.
