We Rebuilt Our October Pitching Model — Then Priced the DraftKings Postseason Board

October baseball creates a different pitching problem.

Rotations get shorter. Managers have quicker hooks. Elite starters can be asked for more. Traditional starters suddenly become bulk relievers. And the pitching plan you expect on Friday can look completely different by Saturday.

So before running our Division Series simulations, we rebuilt one of the important inputs in the Basewinner model:

sxIP — expected starting-pitcher innings.

Rather than simply carrying regular-season workloads into October, we're adjusting expected innings based on the individual pitcher's Basewinner bwERA+ rating and postseason circumstances.

Elite starters on rested teams can receive additional expected innings.

Pitchers with a bwERA+ above 100 receive a substantially shorter expected leash.

And every inning removed from the starter has to go somewhere.

It goes to the bullpen at that team's Basewinner bullpen rating.

The process becomes:

bwERA+ → postseason sxIP → bullpen exposure → staff run rating → projected RA/G

Our Division Series Pitching Numbers

That process produced these current Basewinner runs-allowed projections:

Los Angeles — 3.03
Milwaukee — 3.54
Cleveland — 3.63
New York — 3.75
San Diego — 3.76
Atlanta — 3.89
Tampa Bay — 4.24
Chicago — 4.72

From there, we ran the entire postseason 100,000 times.

Then we took those probabilities and priced them against the current DraftKings postseason menu.

Seven Model vs. DraftKings Discrepancies

Cleveland over New York in the ALCS

Basewinner probability: 17.42%
Basewinner fair price: +474
DraftKings: +700
DK break-even: 12.50%

This was the largest percentage-value discrepancy we found.

Our simulation produced this exact result 17,423 times in 100,000 postseason simulations.

Cleveland to Win the American League

Basewinner probability: 32.19%
Basewinner fair price: +211
DraftKings: +290
DK break-even: 25.64%

This is another expression of the model's higher assessment of Cleveland, but it doesn't require identifying its ALCS opponent.

Cleveland OR Los Angeles to Win the World Series

Basewinner probability: 49.28%
Basewinner fair price: +103
DraftKings: +145
DK break-even: 40.82%

The model has Los Angeles winning the World Series 39.36% of the time and Cleveland 9.92%.

Because those outcomes are mutually exclusive, the combined probability is 49.28%.

Los Angeles to Win the World Series

Basewinner probability: 39.36%
Basewinner fair price: +154
DraftKings: +200
DK break-even: 33.33%

This is one of the cleaner disagreements.

Our simulation has Los Angeles winning the World Series nearly four times in ten.

The DraftKings +200 price requires one in three.

Los Angeles OR San Diego to Win the World Series

Basewinner probability: 46.93%
Basewinner fair price: +113
DraftKings: +150
DK break-even: 40.00%

This is another way the strength of our Los Angeles projection shows up in a derivative market.

Atlanta OR Los Angeles to Win the World Series

Basewinner probability: 47.38%
Basewinner fair price: +111
DraftKings: +135
DK break-even: 42.55%

Los Angeles accounts for most of this probability, but Atlanta's 8.02% championship probability pushes the combination to 47.38%.

Los Angeles over San Diego in the NLCS

Basewinner probability: 22.49%
Basewinner fair price: +345
DraftKings: +390
DK break-even: 20.41%

This is where expanding the simulation became especially useful.

Originally, our postseason output tracked how often each team advanced.

We modified it to record the actual LCS matchup, winner and loser in every simulation.

Los Angeles defeated San Diego in the NLCS in 22,491 of our 100,000 simulations.

That allows us to price a market like this directly rather than trying to approximate it from two separate advancement probabilities.

And Now We Can Price the Division Series

The same simulation gives us probabilities for each Division Series.

Two current DraftKings series prices stood out:

Cleveland over Chicago

Basewinner probability: 64.97%
Basewinner fair price: -185
DraftKings: -140
DK break-even: 58.33%

New York over Tampa Bay

Basewinner probability: 58.71%
Basewinner fair price: -142
DraftKings: -120
DK break-even: 54.55%

Los Angeles is also priced slightly cheaper at DraftKings than our model's fair series price.

Milwaukee is the opposite.

Our simulation makes Milwaukee only 57.93% to beat San Diego, approximately -138 fair, while DraftKings is asking -180.

That's useful information too.

Finding where we disagree with a market doesn't always mean finding something to bet. Sometimes the model tells us that the sportsbook is charging considerably more than we're willing to pay.

What I Find Most Interesting

There are two themes in the output.

Cleveland produces the largest individual model-versus-market discrepancies.

But Los Angeles appears repeatedly across multiple markets.

Our postseason pitching model currently projects the Dodgers at only 3.03 runs allowed per game, the best number among the eight remaining teams.

That strength propagates through the bracket.

But the larger story isn't Cleveland or Los Angeles.

It's what we're now able to price.

We're starting with individual pitcher performance.

Then:

bwERA+
↓
postseason sxIP
↓
bullpen exposure
↓
team runs allowed
↓
game probabilities
↓
series probabilities
↓
100,000 postseason simulations
↓
World Series, league, series and exact-result prices

The final probabilities look clean.

Getting to them isn't.

Next Up: The Messy Part of Modeling October

The next Basewinner post is going under the hood.

Building the postseason Monte Carlo is complicated enough. But before we can run 100,000 simulations, we have to solve another difficult problem:

Who is actually going to pitch — and for how long?

Jesús Luzardo provided a perfect example when he gave Philadelphia five innings despite considerable uncertainty surrounding his expected workload.

Erick Fedde provided another. A pitcher who appeared to fit naturally into Chicago's bulk-inning plans wasn't used in one game, then recorded the final six outs the next.

Those aren't minor details.

If we project a starter for 3.0 innings and he actually gives his team 5.0, we've shifted two innings away from the bullpen. Depending on the quality of the starter and bullpen, that can materially change the team's expected runs allowed.

That's why we rebuilt sxIP for October.

In the next post, we'll look at how difficult constructing postseason rotations really is, why expected innings can be just as important as the pitcher's rating, and how we turn all of that uncertainty into inputs our postseason simulation can actually use.

The simulation is only as good as the assumptions we feed into it — and October makes those assumptions especially difficult to build.

DraftKings prices referenced in this analysis were captured October 2, 2026, prior to publication and are subject to change.