In October 2024, Diamond Sports Group told a bankruptcy court it intended to end its relationship with 11 of the 12 MLB teams it televised in 2024, keeping only the Atlanta Braves.

The company had spent the prior year and a half in Chapter 11, and the announcement read as the terminal event for the regional-sports-network model that had funded local baseball broadcasting for three decades.

One company, carrying a dozen teams' games behind a cable paywall, collapsing under debt it could no longer service.

The actual outcome three months later was less clean.

When Diamond Sports emerged from bankruptcy in January 2025 as Main Street Sports Group, with nearly $9 billion in pre-petition debt reduced to $200 million, it had in fact renewed with eight of those teams: the Braves, Royals, Angels, Marlins, Brewers, Cardinals, Tigers and Rays.

The rebrand (the company's 16 regional networks now operate as FanDuel Sports Network) was framed as a stabilization, not a retreat.

It did not hold.

By February 2026, six of those eight renewed teams — the Royals, Marlins, Brewers, Cardinals, Tigers and Rays — had left FanDuel Sports Network anyway, their broadcasts folded into MLB Local Media, the league's own production and distribution arm.

The Braves left for a fully independent, team-built operation, BravesVision, launched in 2026 with no Main Street or MLB Local Media involvement at all.

The Angels terminated their FanDuel agreement in January 2026, one of nine teams the reporting describes as exiting largely to preserve optionality while Main Street's finances remained unstable, and the team subsequently took over its own regional network rather than staying inside anyone else's.

Of the eight teams that looked rescued in January 2025, all eight had left the FanDuel Sports Network structure within thirteen months.

The January 2025 renewals turned out to be a delay, not a resolution.

By that count, MLB Local Media produces broadcasts for 14 teams as of February 2026: the Diamondbacks, Reds, Guardians, Rockies, Tigers, Royals, Marlins, Brewers, Twins, Padres, Mariners, Cardinals, Rays and Nationals.

This is worth stating precisely, because the popular framing of "team-owned DTC channels" collapses a real distinction. Marlins.TV and Nationals.TV, both launched for 2026, are consumer-facing brand names for content that MLB itself produces and distributes. The league runs the cameras, the truck and the rights sale, and the team's name is on the storefront.

BravesVision and Rangers Sports Network are a different thing entirely: broadcast operations built, staffed, and owned by the team, with MLB Local Media nowhere in the production chain.

The distinction matters for everything that follows, because it separates two different bets on what replaces the RSN model: one in which the league becomes the default local broadcaster, and one in which the largest-market or most aggressive teams reclaim full control by exiting the shared system altogether.

Laid across all 30 teams, the local-rights map by 2026 sorts into three groups:

1. 14 teams under MLB Local Media.

2. 7 teams operating fully team-owned or team-controlled networks (the Dodgers' SportsNet LA, the Yankees' YES Network, the Mets' SNY, the Cubs' Marquee Sports Network, the Rangers' RSN, the Braves' BravesVision, and the Angels' recently reclaimed network).

3. 9 teams still inside other regional-sports-network or joint-venture structures shared with non-MLB properties (the Red Sox with NESN, the Phillies with NBC Sports Philadelphia, the Giants with NBC Sports Bay Area, the Orioles with MASN, the Astros and White Sox and Pirates and A's in ventures shared with NBA or NHL co-tenants, and the Blue Jays inside Rogers' cable systems).

Isolating the revenue effect from reshuffling noise

The question of whether this is actually a financial collapse, rather than a change in who signs the check, has one clean case study:

The Cardinals local media revenue is documented across three consecutive broadcast structures.

• In 2024, the Cardinals received $75 million from Bally Sports.

• In 2025, after Bally's restructuring into FanDuel Sports Network, that fell to $60 million.

• In 2026, having moved to MLB Local Media, the team's expected revenue is $20 million (a total decline of 73 percent in two years).

The final step is described as an "eat-what-you-kill" arrangement in which the team earns from advertising and carriage fees minus its own production costs, rather than collecting a negotiated rights fee at all.

The Cardinals' trajectory is the sharpest documented case, but the direction is consistent elsewhere.

The Texas Rangers were contracted to receive $111 million in 2024 under a 20-year, $3 billion deal signed with Fox Sports Southwest in 2010.

The actual 2024 payment, disputed by the team as overstated even at the lower figure reported, came in at "less than $100 million," a shortfall of at least $11 million against the original contract before the team left for its own network entirely.

The Pittsburgh Pirates' 2022 RSN fee, at $61 million against a $61.2 million payroll, illustrates how thin the margin already was for smaller-market teams even before the bankruptcy. Meaning a fee drawn from a broadcast reaching roughly 28,000 households, or about $2,179 in rights cost per household reached.

None of this is uniform loss, however, and treating it as such would overstate the case.

MLB and the players' union built a $75 million "media-disruption distribution" pool, funded from luxury-tax collections, to partially offset the transition for nine teams whose local rights fees fell between 2022 and 2024. Those teams being the Diamondbacks, Guardians, Rockies, Astros, Pirates, Padres, Mariners, Twins and Rangers, each eligible for up to $15 million.

Even at full disbursement to all nine, the pool would fall $60 million short of fully replacing the $135 million in maximum entitlements it was built to distribute, and commissioner Rob Manfred has said the program is not currently scheduled to repeat.

The pool is real, sourced from real luxury-tax money, and it is also explicitly a bridge rather than a replacement. MLB itself has been candid about that.

What's actually rebuilding the model

The RSN collapse did not leave a vacuum.

It coincided with, and helped justify, a restructuring of MLB's national television package that shifted meaningfully more revenue into the centrally-shared pool the league controls directly.

In November 2025, MLB signed three-year deals for 2026 through 2028 with ESPN, NBC, and Netflix.

ESPN retained its existing $550 million annual value for 30 nationally televised games per year, but lost Sunday Night Baseball to NBC and gained something structurally more significant: oversight of MLB.TV and out-of-market distribution rights, a business the league had run mostly on its own.

NBC returned to regular MLB coverage for the first time in 25 years at nearly $200 million annually, covering Sunday Night Baseball, the Wild Card round, and select Peacock-exclusive games.

Netflix, new to the sport, pays roughly $35 million a year for the Home Run Derby, the Field of Dreams game, and a standalone Yankees-Giants Opening Day broadcast.

The practical result is that ESPN, as of the 2026 season, sells the out-of-market MLB.TV subscription directly: $134.99 a season for existing ESPN Unlimited subscribers, $149.99 for new subscribers with a bundled free month.

That folds a previously separate league-run product into a national media company's subscription funnel.

That single move does more to centralize the sport's broadcast architecture than any single RSN's bankruptcy did on its own: it takes the one piece of MLB's media business that was inherently national already (any fan, anywhere, paying for out-of-market access) and hands its retail relationship to the same company paying the league $550 million a year for in-market national windows.

This is the mechanism worth naming plainly.

The RSN collapse forced roughly half the league's local broadcasts into a single, MLB-run production pipeline. The 2025 national-rights renegotiation simultaneously moved MLB.TV, the league's own direct-to-consumer product, under a national partner's commercial umbrella.

Two years ago, "national" and "local" were separate businesses with separate money. By 2026, MLB Local Media, MLB.TV, and the national broadcast package are converging into a single centrally-operated distribution stack.

That is not because the league announced a centralization plan, but because the RSN failures left it holding the local business by default at the same moment it renegotiated the national one.

The persistence of big-market advantage complicates the centralization story

If the RSN model were simply dying everywhere, the case for the league absorbing local broadcasting would be straightforward. It is not dying everywhere, and the exception is instructive.

The Los Angeles Dodgers' local television deal with Spectrum SportsCenter LA, signed in 2013 for 25 years, still averages $334 million a year and is contracted to exceed $500 million annually by its 2038 expiration.

That single team's annual local media revenue is more than four times the size of the entire $75 million disruption pool built to help nine other franchises.

The Dodgers deal also carries a structural advantage unrelated to its size.

Following the team's 2011 bankruptcy, MLB agreed to continue taxing the Dodgers' local revenue for luxury-tax and revenue-sharing purposes as though the team were still under bankruptcy-era valuation, even after the network contract was renegotiated in 2013 on far more lucrative terms.

The exact dollar value of that exemption isn't cleanly disclosed in public reporting. No year-by-year comparison between the Spectrum contract's actual value and the "fair market value" used for revenue-sharing purposes has been published.

Still, the existence of a bankruptcy-court-derived carveout, still in effect more than a decade later, means the league's largest local media contract is also its most sheltered from the revenue-sharing system that other teams' RSN losses are currently straining.

Compare that to the Miami Marlins, whose annual local broadcast revenue sits below $50 million even before their 2026 move to MLB Local Media's eat-what-you-kill structure. The Yankees, whose YES Network revenue, while substantial, runs at less than half the Dodgers' figure despite comparable market size and team value.

The RSN collapse has not touched the Dodgers, Yankees, Mets, or Cubs at all; all four still run their own team-controlled networks outside both MLB Local Media and the disruption-payment system built for teams that lost RSN money.

The teams absorbing the collapse and the teams shielded from it are not randomly distributed.

They track market size and, in the Dodgers' case, a specific piece of decade-old bankruptcy litigation that has nothing to do with the current RSN failures at all.

What the case establishes about the CBA negotiations

The RSN collapse and MLB's collective bargaining position are not two separate stories running in parallel.

They are the same argument made twice.

On May 28, 2026, MLB submitted a formal economic proposal to the players' union that would, for the first time, classify all local media revenue as centralized revenue shared equally among all 30 teams.

That would include not just the roughly half the league already routed through MLB Local Media by circumstance, but the Dodgers' $334 million and the Yankees' YES Network money as well.

The same proposal paired that centralization with a hard salary cap, calling for a 2027 payroll ceiling of $245.3 million and a floor of $171.2 million, and a 50-50 split of revenue growth between players and owners going forward.

Commissioner Rob Manfred has been explicit about why centralized revenue is the point, not a side effect.

He has argued publicly that NBA franchises sell for 10 to 15 times annual revenue while MLB teams sell for 4 to 11 times, and attributes part of that valuation gap to the NBA's heavier reliance on predictable, centrally-distributed national money over volatile local rights deals.

That argument lines up with a disparity reporting on the CBA fight has quantified directly: roughly 26 percent of MLB's total revenue currently comes from national sources, against about 66 percent for the NFL.

Under that framing, the RSN collapse is not a crisis MLB is managing so much as a preview of the outcome it is negotiating for: a league where the Dodgers' local check and the Marlins' local check both disappear into the same centrally-managed pool.

The players' union's counterproposal took a different route to a similar-sounding goal.

Rather than centralizing local revenue, it called for expanded revenue sharing that would guarantee every small-market team a minimum of $240 million in total revenue annually.

That is a redistribution mechanism layered on top of the existing team-by-team ownership structure, not a replacement for it.

The difference is not cosmetic.

Centralization changes who owns the local media asset; a revenue-sharing floor leaves ownership in place and taxes the proceeds.

One is the structure MLB Local Media has already imposed, by necessity, on the fourteen teams the RSN bankruptcies orphaned. The other is closer to the deal the Cardinals or Rangers might have preferred to the one they actually got.

What the last two seasons establish is narrower than either side's rhetoric suggests.

They establish that the RSN local-monopoly model is now structurally dead for a specific tier of the league, roughly the bottom half by market size, and that MLB's own production arm has become the default successor for that tier by circumstance rather than by a deliberate buildout.

They establish that the league's national broadcast renegotiation and its absorption of local production are converging into a single distribution stack faster than either was moving on its own.

They do not establish that full centralization would equalize outcomes.

The Dodgers' bankruptcy-era revenue-sharing exemption shows that even inside a system built around shared and centralized money, the largest markets have historically kept mechanisms to protect their advantage, and nothing in MLB's current proposal explains how a $334 million Dodgers local contract and a since-orphaned $20 million Cardinals contract would be reconciled into one equal share without a fight over whose number becomes the baseline.

Nor do the DTC and MLB Local Media experiments so far establish that the new model even replaces the old one's revenue.

The Cardinals' 73 percent decline and the Rangers' minimum $11 million shortfall are the clearest available data points, and both point toward the transition costing teams money, not neutrally reallocating it.

The RSN model's collapse is real and largely finished.

What replaces it, and who it will actually pay, remains the subject of a labor negotiation that has not yet been resolved.

OddMan exists for exactly this kind of work: a research partner for anyone studying the business and science of sports seriously, not just following headlines. Join us to get the same data and analysis layer used to build this piece, so the next question you have about a franchise’s finances, a rule change, or a roster decision doesn’t have to stop where the article does.

Get OddMan

With roughly three weeks remaining in the 2026 season (at the time of this writing), the Chicago White Sox hold sole possession of first place in the American League Central.

They lead by three games over a Cleveland Guardians team that is good enough to simultaneously lead the league's wild-card race.

They have reached this position while carrying a payroll of between $105 million and $117 million over the course of the season, depending on the measurement date, which places them 26th to 28th of 30 teams against a $244 million luxury-tax threshold.

That combination is useful for a specific analytical reason, not merely as a late-season curiosity.

Major League Baseball's ownership group is currently proposing a hard salary cap for the next collective bargaining agreement, and the argument for it rests on the claim that payroll disparity is the primary driver of competitive imbalance across the league.

A bottom-payroll division leader, in the same season that argument is being advanced publicly, functions as a natural test case for the claim.

This article treats it as one: it asks what the White Sox's performance data actually shows, what mechanism produced the result, and how far the case can reasonably be generalized.

The argument under examination

MLB ownership's core argument for a proposed hard salary cap in the next CBA is that payroll disparity is the primary driver of competitive imbalance across the league.

The proposal reported publicly pairs a hard cap near $245 million with a payroll floor in the $171 million to $222 million range, alongside a shift toward centralized local revenue sharing.

The MLB Players Association has rejected a cap in any form and countered with a different mechanism: a guaranteed local-revenue floor for smaller-market clubs, funded through expanded revenue sharing, with no ceiling imposed on the high end.

The two sides are not proposing variations on the same fix; they are proposing different diagnoses of what actually causes competitive imbalance, which is what makes a real-time counter-example like the White Sox analytically useful rather than incidental.

The 2026 White Sox are a useful test case for the owners' specific claim in something close to real time.

They are a franchise spending between $105 million and $117 million on payroll this season, depending on the point in the year measured. They are ranking 26th to 28th of 30 teams against a $244 million luxury-tax threshold and are presently the best team in their division.

It is worth being precise that this is a payroll position, not a valuation position: independent 2026 rankings from Forbes and CNBC place the White Sox 17th to 20th of 30 teams by enterprise value, well clear of the league's five lowest-valued franchises (Miami, Kansas City, Cincinnati, Pittsburgh, and Minnesota).

Chicago's low payroll is a front-office choice made during a rebuild by a mid-value franchise, not evidence that a structurally low-revenue team can compete on parity terms with the sport's largest markets, and the two claims support different conclusions about the underlying parity question.

Naturally, this leads us to three questions.

1. Is the White Sox's record supported by underlying performance, or is it a product of favorable variance likely to regress before the season ends?

2. What roster-construction mechanism explains the result, and is that mechanism specific to this team or observed elsewhere historically?

3. What does this case actually establish about the broader parity debate, and what does it not establish?

Isolating variance from underlying performance

The standard diagnostic for distinguishing a "deserved" record from a fortunate one is the gap between actual wins and Pythagorean expected wins, a function of run differential.

Chicago has scored 680 runs against 631 allowed this season, a +49 run differential that produces a Pythagorean expected record of 76-67.

Their actual record over the same span sits at 74-68, meaning the team is performing in line with, and if anything marginally below, what its underlying run production and prevention would predict.

A team benefiting from favorable sequencing — winning an outsized share of one-run games — typically shows the opposite pattern, with actual wins outpacing what run differential would predict.

That is not the pattern observed here.

The month-by-month data reinforces this rather than complicating it.

In August specifically, an overtaxed bullpen produced a losing Pythagorean split for that month alone (133 runs scored against 140 allowed), and the team's actual results tracked that decline rather than defying it.

A club riding sequencing luck would be expected to paper over exactly this kind of underlying dip; this one did not.

This matters because it forecloses the simplest dismissal of the case. If the White Sox were merely accumulating fortunate results in close games, the appropriate response would be to expect regression and set the case aside.

The data instead indicates a team whose run production and prevention are genuinely competitive across the full season and consistent at the monthly level, which is a materially different, and more consequential, finding for the question at hand.

What’s actually producing the result

The roster construction underlying this result is specific and identifiable, not diffuse.

A cluster of position players in their mid-20s or younger — Colson Montgomery, Kyle Teel, Chase Meidroth, Braden Montgomery, Sam Antonacci, and Miguel Vargas — account for the bulk of the team's value while earning pre-arbitration or first-arbitration-year salaries, a small fraction of open-market cost for comparable production.

The individual performance underlying that value is not marginal.

Colson Montgomery, in his first extended major-league look, slashed .239/.311/.529 with 21 home runs and an .840 OPS across 71 games.

Teel and Meidroth both posted numbers atop the team's rookie leaderboard the year before this core was handed the everyday roster: Teel hit .273 with a .786 OPS, and Meidroth hit .253, the two highest batting averages on the team alongside fellow rookie Edgar Quero.

This is production at or above league-average offensive output, generated by players whose salaries are set by service-time rules rather than by what that production would command on the open market.

Around this core, the organization made one notable financial commitment: a two-year, $36 million contract for Munetaka Murakami, whose $17 million 2026 salary makes him the highest-paid player on the roster by a wide margin over anyone in the pre-arbitration group.

Beyond that single addition, the front office otherwise avoided adding salary at the trade deadline, reportedly declining to include core young players, Montgomery specifically among them, in trades for short-term rental pitching help.

The roster is, in effect, one veteran bet layered on top of a cost-controlled core, rather than a broadly reshuffled 26-man group.

This is a known, previously observed mechanism rather than a novel one, and the historical cases share more than a payroll bracket.

Let’s examine a few of those cases.

The 2008 Tampa Bay Rays carried a $43.8 million payroll, second-lowest in the league, were described at the time as one of the youngest rosters in baseball, and won the American League pennant.

The 2020 Rays, at roughly $28.2 million and third-lowest in the sport, relied on a roster in which several regulars had not yet accrued enough service time to qualify for arbitration, and also reached the World Series.

The 2023 Baltimore Orioles, second-lowest payroll in MLB that year, were built around a young core fronted by catcher Adley Rutschman and finished 101-61 to win the AL East.

The 2023 Rays, 28th of 30 in payroll at $74.5 million with Zach Eflin as their highest-paid player at $11 million, opened that season 13-0, the best start in modern major-league history, before making the playoffs as a Wild Card.

Across every verified comparable case, the mechanism is the same: a talent-development and roster-timing advantage exploited during a specific team-control window, not a generally efficient approach to twenty-six-man roster construction.

This franchise has run this play before

The same organization provides its own caution against generalizing from this case. The White Sox's previous rebuild, built around Yoan Moncada, Eloy Jiménez, and Luis Robert Jr., followed a structurally similar template: a young, cost-controlled core given a runway to develop. That version of the strategy failed.

Development stalled, the organization extended several of those players to long-term contracts before they had established major-league performance, and the outcome was a three-season stretch that included the worst full single season in modern MLB history.

The mechanism observed in 2026 is not new to this front office; its prior application, under similar starting conditions, produced the opposite result.

That version of the strategy failed, and it failed specifically along the pre-established-performance dimension:

• Jiménez, whose rookie year had shown real power (31 home runs, a .513 slugging percentage), spent most of the rest of his Sox tenure battling recurring soft-tissue injuries. By 2024 he was hitting .231 in an injury-shortened season before Chicago traded him to Baltimore on July 30, 2024 for a single minor-league reliever, a return that reflected how little was left of the original investment.

• Moncada's decline followed a similar arc, and the White Sox declined his 2025 club option rather than pay a $25 million salary for a diminished everyday role.

• Robert Jr., the one player from that core still with the organization by 2025, was also the one whose bet looked the most justified in year one but whose production had become inconsistent enough by the time of the rebuild's end that he no longer resembled the foundational piece the extension had envisioned.

The outcome across the group, compounded by the broader roster's decline around them, was the 2023-2025 stretch of three consecutive 100-loss seasons, anchored by the 41-121 mark in 2024 that stands as the most losses by a team in a full modern-era season.

This is evidence that the strategy's success rate, even within a single organization working with players it evaluated and developed itself, is not high, and that execution and player-development and health variance are doing at least as much work as the strategic template itself.

It also suggests a live risk specific to the current core.

The 2026 roster has so far avoided the extend-before-you-know-what-you-have step that defined the Jiménez and Robert deals.

How the organization handles that decision with Montgomery, Teel, and the others as they approach arbitration will be a more direct test of whether the lesson from 2019-2020 was actually learned.

The limits of what this case proves

The 2026 White Sox constitute genuine, verifiable evidence against the strongest version of the owners' claim: that a bottom-quartile payroll forecloses contention.

The team's performance is supported by underlying run production, not variance, and the roster-construction mechanism producing it is well-documented and consistent with prior successful cases elsewhere in the league's recent history.

The case does not establish that payroll disparity is immaterial to competitive balance more broadly, for two reasons.

First, the mechanism observed is narrow and time-limited by construction: cost-controlled cores become expensive as players approach arbitration and free agency, and the same organization's history demonstrates that the transition from "developing core" to "sustained contender" is not reliable, even internally.

Second, Chicago's mid-tier valuation means this is not a test of whether the league's most revenue-constrained franchises specifically can compete without structural assistance. It is a test of whether a mid-value franchise choosing a lean payroll during a development window can compete, which is a narrower and less consequential claim.

The appropriate scope for this finding is accordingly limited.

It is a valid counter-example to categorical claims that spending is a precondition for contention, and it is not evidence bearing on the sustainability of low-payroll competitiveness across the broader set of franchises the collective bargaining negotiation is actually concerned with.

Whether the current White Sox core converts this window into sustained contention, or follows the trajectory of the organization's previous attempt at the same strategy, will not be determinable from the data available at this point in the season.

OddMan exists for exactly this kind of work: a research partner for anyone studying the business and science of sports seriously, not just following headlines. Join us to get the same data and analysis layer used to build this piece, so the next question you have about a franchise's finances, a rule change, or a roster decision doesn't have to stop where the article does.

Get OddMan

In our last article we walked through the data foundation underneath Ask OddMan: why we started with nflverse, how we mapped the ontology before trusting it, and how that work is becoming a real knowledge graph. This note is about what sits on top of that foundation: the agentic harness that actually does the thinking, and the guardrails we're building around it.

We’re building Ask OddMan as a multi-agent system, not a single model answering questions. The distinction sounds subtle but it changes almost everything about how the product is structured underneath.

We're currently working through a few core capabilities that make up the harness:

Agentic orchestration

This is the coordination layer, the part responsible for planning and routing tasks across our internal fleet of agents rather than asking one model to do everything at once.

A question about a player's usage trend and a question that requires pulling a chart, checking a narrative, and cross-referencing an injury report are different shapes of work. We also believe they deserve different handling rather than one model trying to be good at all of it simultaneously.

Data sources

This is the connective tissue between the harness and everything we described in Part 1: querying into our data warehouse, connecting out to third-party APIs, and running retrieval and vector search over both narrative content and memory.

This is also where the knowledge graph gets used in practice, not just stored.

Skill builder

We're composing workflows as structured files, markdown documents that define a skill, the tools it can call, and the connections it needs, rather than hardcoding every capability into the agent logic directly.

We like this approach because it means teaching Ask OddMan something new looks more like writing a document than shipping a new build.

The model

Given that we are a start up and that speed to market is what we are up against, we have decided to use an LLM for many reasons to begin with. Let’s breakdown the thinking behind that choice.

We’re building instrumentation into the harness itself to understand volume, cost, and token spend broken down by which skill is actually being executed, not just an aggregate bill at the end of the month.

We believe that's the only honest way to know what a system like this really costs to run at scale, skill by skill rather than as one number.

Once we see that volume cross a real cost threshold, we intend to use what we've learned from that instrumentation to start building and fine-tuning our own language models for the specific skills where it makes sense, rather than staying dependent on general-purpose models for everything indefinitely.

Right now we're still gathering the data that will tell us when we should be.

There's a version of Ask OddMan that only ever repackages numbers other people already published, and we don't think that's interesting enough on its own.

Alongside the harness work, we're currently building our own machine learning models to produce original stats and insights for the NFL, not just retrieve and narrate the ones that already exist across nflverse or anywhere else.

The part we care about most is what happens after a model produces something new.

We're not treating an original stat as a one-off output that lives in a chart and then disappears. We're feeding it back into the knowledge graph itself, as a real node connected to the players, teams, and games it describes, the same way everything else in that graph is connected.

That means a proprietary insight we generate isn't sitting off to the side as a separate feature. It becomes part of the same structure Ask OddMan is already traversing, available to inform the next question a user asks, not just the one that prompted it.

We're early in this, and we're being careful not to call something a "stat" until we trust the model behind it the same way we came to trust the ontology work in Part 1.

But this is the direction we're building toward: an Ask OddMan that doesn't just know what's already been said about a player, but is starting to say some genuinely new things of its own.

We think a lot about what "memory" should mean for a research partner, because the word gets used loosely.

For us it's a few distinct layers working together, not one big undifferentiated context blob:

• Thread sessions

  Thread sessions hold the immediate back and forth of a single conversation.

• Projects

  Carries context forward across an entire body of work, the way a real project should, so a session doesn't start from zero every time.

• Tools

  This lets the agent reach out and get fresh information rather than relying only on what it remembers.

• The way you view sports

  Learn how a given user tends to think and what they tend to care about, so the system gets more useful to that specific person over time rather than staying generic.

We believe this layered approach is closer to how an actual research relationship works. You don’t re-explain your point of view to a colleague every morning, and we don’t think you should have to re-explain it to Ask OddMan either.

We’re building this with governance in mind from the start, not bolting it on after the fact.

A system that can plan, route tasks across agents, query a live data warehouse, and remember things about a user is a system that needs real boundaries, not just capability.

We don't have a polished answer to give you here yet, because we're still building it. But we wanted to say plainly that we're thinking about it now, at this stage, rather than treating it as a problem for later.

That's consistent with how we think about everything else on OddMan: the tools should give Independent Creators real leverage, and the leverage has to come with protection built in, not around.

We're not finished. Most of what we described across these two articles is still being built, tested, and rebuilt as we learn where our own assumptions were wrong. That's the actual shape of this right now, and we'd rather tell you that than pretend otherwise.

We'll keep sharing this as we go.

We’re going to spend the next two articles walking through how we’re building Ask OddMan, our AI research partner grounded in live sports data. Not the finished-product version. The actual version, mid-build, with the decisions we made and why we made them.

We wanted to start with data, because everything else depends on getting this part right. An AI research partner is only as trustworthy as what it's standing on, and before we wrote a single line of agent logic, we spent real time just trying to understand the shape of the world we were about to model.

We’re currently building against nflverse as our primary NFL data source, and we picked it deliberately rather than defaulting to it.

The NFL doesn't run an open, self-serve developer program for its official statistical data. That data is licensed out through an exclusive partnership, currently held by Genius Sports, which covers real-time play-by-play and Next Gen Stats through the end of the 2027-28 season.

Gaining access as an outside company means entering into a formal licensing relationship, not just filling out a registration form.

We believe the right way to approach that kind of conversation is with a complete understanding of what we actually need, not a partial one.

Walking into a discussion with a league, a data partner, or a franchise before we know our own data requirements felt backwards to us. So we're building our foundation on nflverse first.

It's a community-maintained project, and the code itself is released under the MIT license. What mattered more to us was the data licensing underneath it: the majority of nflverse's datasets are published under CC-BY 4.0, meaning we can use them broadly as long as we attribute the source, with a narrower subset, the FTN charting data, carrying a CC-BY-SA 4.0 requirement instead.

That's a fundamentally different starting point than the NFL's own official data, which flows through an exclusive licensing partnership rather than an open, attributable license.

nflverse also happens to be the largest, most robust dataset available outside of the NFL's own systems, going back to 1999 for play-by-play.

We're using this phase to learn everything we can about the shape of football data, so that if and when we do sit down with a league, a data partner, or a team, we're bringing clarity instead of a wish list.

Twenty years of historical data is a lot to commit to a warehouse blind. Before we loaded any of it into our data warehouse, we pulled samples from every nflverse endpoint and started drafting the ontology underneath it.

Meaning, what a player object really is, how it relates to a team, a game, a play, or a season.

We used a free graph tool offered by Microsoft to actually visualize and draw those relationships before we committed to anything at scale.

Seeing the nodes and edges on the screen, rather than just imagining them from a schema doc, is what let us catch the relational gaps early.

A missing edge between two entities is easy to miss in a spec and very hard to miss when you're looking at the graph. We'd rather find that out with a sample than with twenty years of production data already sitting in the warehouse.

We're repeating this same process across the other major leagues we're bringing into OddMan. It's slower than just pulling everything in and sorting it out later, but we believe the upfront cost is worth it.

This is worth explaining, because it's easy to treat data cleanup as a chore that happens before the interesting work starts, and we don't see it that way.

An AI system like Ask OddMan doesn't reason about football in the abstract. It reasons about whatever structure we hand it, and if that structure is inconsistent the model doesn't know that. As an example, if a player object means something slightly different depending on which endpoint it came from, the model wouldn’t know. It just inherits the confusion and hands it back to the user with confidence.

Clean, well-related data isn't a nice-to-have ahead of the AI work. It's a direct input into whether Ask OddMan's answers are trustworthy at all.

We'd rather spend the time now finding the gaps than have Ask OddMan discover them later, in front of a user, as a wrong answer.

Here's where it stopped being a diagramming exercise and became infrastructure. The relationships we mapped, player to team, team to league, player to game, game to play, aren't just documentation now.

We're building them into our graph database, so that a league, a team, and a player exist as connected objects rather than rows that happen to share a foreign key.

That distinction matters more than it might sound like it does.

A relational table can tell you a player is on a roster. A graph can tell you how that player connects outward through every game, every transaction, and every teammate. It can also let something like Ask OddMan traverse those connections the way a person actually thinks about a storyline, not the way a database schema forces you to think about one.

Picture this. You’re asking Ask OddMan a question like "how has Patrick Mahomes' usage changed against teams that blitz heavily."

In a relational world, answering that well means writing a chain of joins across player tables, game tables, participation tables, and defensive scheme tables, and hoping the schema designer anticipated that exact question.

In the graph pictured above, Mahomes is a node, and the path outward is already there: from the player node to his participation records, from each participation record to the specific game it belongs to, from that game to the opposing team, and from that team to whatever defensive tendencies we've modeled as connected to it.

Ask OddMan doesn't need a query written in advance for that exact question. It walks the edges that already exist, the same way a scout would follow the thread from a player to a game to an opponent, and assembles the answer from real, connected structure instead of a query someone had to predict.

We're not saying every question resolves that cleanly today, as we're still early in this build.

But that's the shape of what the graph is for, and it's why we believe it's worth the extra investment over a purely relational approach.

We're still early in populating this graph fully, and we expect to keep finding gaps the same way the visualization tool helped us find them in the ontology phase. But the foundation is real, and it's the foundation everything else we're building on Ask OddMan sits on top of.

In the next article, we’ll get into the harness itself: how we’re coordinating multiple agents, how memory works across a session and across a project, and how we’re thinking about control as we hand more of this over to an agentic system.

We’ll keep sharing this as we go.

In the true spirit of sports science, we’ve built a lab to bring the latest AI technologies to sports storytelling.

We looked at what's available today, and we believe something better can be built. Something that helps the people who see the game differently get closer to the moments and stories they care about.

There's a scene in Moneyball that’s worth bringing up as a discussion point, because it's really a story about storytelling. Peter Brand, a young analyst nobody in the room takes seriously yet, is buried in spreadsheets, looking for players who get on base more than the rest of the league gives them credit for.

He isn't looking at swing mechanics or scouting reports or how a player looks in a uniform.

He's looking at one number, on-base percentage, and setting aside almost everything traditional scouting had trained itself to value instead.

The idea, once you see it, is almost embarrassingly simple. It doesn't matter how a player gets on base, only that he does. More runners on base means more runs. More runs means more wins.

That scene has stayed with us because of what it says about where a story actually starts.

Everyone else in that front office was still telling the old story, the one about tools and swagger and the eye test. Peter Brand was reading a different story entirely, hiding in the same numbers everyone else had access to.

The insight itself took a moment to have. Turning it into a season, a roster, a team the whole league eventually had to reckon with, was the hard part, and it's the part the rest of that story is actually about.

We believe today’s storytellers are living some version of that same story today, just at a different scale, and on a much shorter clock. Someone watching a game can spot the same kind of overlooked pattern Peter Brand found in a spreadsheet, a player quietly outperforming his reputation, a stat that contradicts the story everyone else is telling.

Having that insight still takes a moment. Turning it into something real, a graphic, a breakdown, a tool other people can actually see and use, means moving across a stats site, a screenshot, a design tool, a caption doc, and finally whatever platform they're posting to.

What took a front office and a season to prove out, today’s storytellers are expected to rebuild from scratch with every single post, on every single storyline, all season long.

We believe that gap, between spotting the story and being able to tell it well, is worth closing. That belief is the reason Oddman Labs exists.

Not a talent gap. A tooling gap.

The sharpest people working in sports content today are not, as a rule, engineers. They don't have a developer on call, and they don't have a design team sitting next to them.

What they have is the thing every good broadcaster and writer has always had: an intuition for the story inside the numbers, sharpened by watching more film, reading more box scores, and following more storylines than almost anyone else in the room.

As we've spent time with this problem, what's become clear is that turning that instinct into something publishable, professional, and repeatable currently means becoming a part-time user of five or six different products that were never built to talk to each other:

• A stats provider and research partner

• A listening tool

• A design tool

• A video editor

• A publishing workflow

None of them know the others exist, and none of them know they're all supposed to be telling the same story. Every piece of content becomes a manual assembly job, rebuilt from scratch, every time.

We don't see that as a talent problem. We see it as an infrastructure problem, and a quiet one, since it caps how far someone's storytelling instincts can carry them without ever announcing itself as the reason.

A word on what we’re not building

We know it would be easy to read all of this and assume we’re building another AI tool that writes sports takes on someone’s behalf. We’re not, and we think it’s worth saying plainly why.

We see what's available today, and none of the general-purpose AI tools out there are grounded in what actually happened in last night's game, or wired into a real box score, or able to catch a breaking storyline as it's still developing.

Even where they could get closer, we don't think that would be the point. A creator's voice is the storyteller's voice, and that's the product.

AI is wonderful, and when it's built correctly, it can transcend beyond a chatbot and a series of UI charts and dashboards. What we're hoping it can do inside this platform is remove the friction between spotting a story and telling it, not stand in for the person who spotted it in the first place.

We’re building Oddman as infrastructure and leverage to bring the fans closer to the sports they grew up loving and the moments that were created to bring us all together.

What we’re building, in one line

As we begin this path forward, the shape of it looks like this: ground yourself in the data, chatting directly with AI that's wired into live stats, film, news, and social feeds rather than general knowledge, the way a great researcher sits behind every great broadcast.

Build something from it, turning an insight into an app, a dashboard-style tool, a composite, or a single-page site, without needing to write a line of code.

Create the visual that carries the story, using a creative surface built specifically around sports context rather than generic prompts.

Publish it to a growing home built around your work.

And distribute it outward, protected, to the platforms you already use, so the story stays provably yours wherever it travels.

Data, build, create, publish, distribute, governed the whole way through. That’s the sentence-length version of what we’re hoping to build.

Governance is the part holding everything else up

Every tool a Creator uses today shares the same quiet failure mode. The moment a story leaves your hands, you lose it. Post a stat, a graphic, a video breakdown, and you immediately lose visibility into how it performs, who's using it, whether it's been reposted uncredited or repurposed commercially with nothing coming back to you.

There's no ledger. No way to prove authorship at scale. No way to revoke access if something's being misused. The byline disappears the moment the story leaves home.

We don't think of governance as a feature sitting alongside the others. We think of it as the wall the rest of the platform stands on.

Research, building, and creating are only worth investing in if the story stays provably, trackably yours once it's out in the world, and that protection is the piece we don't think anything else in a creator's current stack even attempts.

Who we’re building this for

We're building this for creators and media personalities already publishing takes, stats, and breakdowns on X, Substack, YouTube, TikTok, or their own site.

For the anchors on your local news station who need to have unique insights delivered to them as they deliver the recap of tonight’s game.

For those of you needing advice on trade deadlines or claiming a player off of waivers.

For people active in the collectibles communities who think about the game the way a professional does, even if it isn't their day job yet.

For the semi-professional and professional voices, podcast hosts and team-adjacent media among them, who need to tell repeatable stories rather than one-off posts, and who care about protecting what they make because it's their livelihood.

We're building it for the people who see the game differently, and just haven't had the ability to get that story out yet. We're also building it for the hours after that first spark of curiosity, the ones spent moving through the visuals Oddman generates, following one number into the next, watching a trend take shape, and following it wherever it leads.

That kind of wandering, the sort that starts with a question and ends somewhere you didn't expect, is part of the craft too, and we don't think it should feel like work to get there.

But more importantly, we are building Oddman so that we too can feel, once again, connected to the sports we grew up playing.

Where we are right now

We're early. In the true spirit of that word, we’re closer to the first inning than the ninth.

The waitlist is open, and the first version of the platform, research, build studio, creative canvas, and the governance layer underneath all of it, is in active development.

We'll continue sharing updates as this journey unfolds, and for a closer look at what we're building, you can read more here on Fumbles. Stay tuned and thank you for taking the time out of your day to allow us to introduce ourselves.