The OKC Thunder Strategy Kernel

Should an organization have a single Guiding Policy in their strategy? This is a question I’ve been asking myself after reading Richard Rumelt’s “Good Strategy Bad Strategy”. Rumelt’s framework is to diagnose the situation, craft a guiding policy, then define actions coherent with the policy. The Guiding Policy is singular; however, both the Wonderful pistacho case and the Crown Cork & Seal case appear to have multiple, separate policies. Rumelt even notes that the “coordination of policies” produces an advantage in the Crown example.

As I talked this through with a friend, I started using the Oklahoma City Thunder as an example to explore this question. Sam Presti, their general manager, pretty clearly operates more strategically than most other GMs. At the time, they were the favorites to win the 2025 NBA championship¹, making them a case study in excellence. More than just talking, I did what any slightly obsessive curious person would do: I created a Google Doc and started trying to write Rumelt-like strategies based on the Thunder’s actions.

The first policy I sketched was a long bulleted list of loosely related things. Many revisions later, I’d boiled the Thunder’s strategy down into a single Guiding Policy (that does have several facets). Here’s how I describe the Thunder’s strategy:

Diagnosis

The NBA championship is a zero sum game that is won by having a team that reaches a collective capability² greater than every other team it faces during the NBA Playoffs. Capability is dynamic. Players get injured. Their performance fluctuates based on a million factors. Results of an NBA season are probabilistic rather than deterministic.

In addition to randomness, the OKC Thunder face the following constraints:

• Exceeding the salary cap incurs penalties which heavily constrain team building options.
• Oklahoma City is a small market with smaller revenue relative to other teams, and they have a mandate to maintain profitable operations.
• Player salaries are time dependent: veterans who reach free agency command higher salaries than younger players on the rookie scale.
• Tactical advantages in the NBA are ephemeral. Teams quickly learn from each other’s success.

Since randomness is such a powerful factor, OKC believes that the surest path to achieving any success is to pursue a path of sustained success. In order to do that, they must increase both the number of and likelihood of chances that result in employing championship level capabilities in their players³.

Apart from those constraints, OKC has a structural advantage over most other teams: their owner is willing to be patient. He is committed to accepting the short term downsides of a strategy that evaluates over a multi-year time period.

Guiding Policy

The Thunder will field championship level collective capability through a culture of expert evaluation⁴ and creative experimentation⁵ which generates more and better options evaluated over longer time periods⁶ than the rest of the NBA is able or willing to endure. Decisions will be financially disciplined to avoid paying penalties they cannot afford but also increasing future optionality.

Actions

Actions	Coherence with Policy
Drafting Players: Acquired more first round picks than anyone else. Missed on many picks. Targeted a mix of “quiet” players and “high upside” bets.	Clearly the Thunder view the draft as the most efficient way to build and improve a team. They have drafted ~5 All-NBA level players, which speaks to their skill at player evaluation. Even then, they know how high the variance in a pick is, which led them to acquire as many “roulette wheel spins” as possible. Their drafted players share common traits, indicating strong evaluation preferences.
Trading Players: Traded young players for an established star at the end of the Russell Westbrook window. Blew up that team earlier than most teams would have and acquired many draft picks. Rented salary cap space for teams to dump their bad contracts in exchange for future draft picks. Regularly traded up in the draft to get the player they wanted. Required Shai Gilgeous Alexander in Paul George trade to LAC. Traded Josh Giddey for Alex Caruso without giving up any draft picks.	Their trade history follows the pattern of increasing the “championship capability” ceiling for multiple years. The actual action (trading for veterans, absorbing contracts, trading for picks, trading for role players) depends on the team’s current outlook. Notably, they did not cash in their draft capital to trade for another “star” in 2024/2025. This speaks to their patience, long time horizon, and confidence in their evaluations. The Giddey trade also falls right out of the policy: he was a good-but-not-great player with superfluous strengths and critical weaknesses (when combined with their “better” young core). Caruso plugged a defensive gap and wouldn’t jeopardize their future financial flexibility like Giddey would have. It’s incredible the Bulls weren’t able to get a draft pick included in the deal, but OKC clearly values the draft much more than the Bulls do.
Free Agents: Rarely sign free agents. Notably signed Isaiah Hartenstein to fill a need at center to a deal with a team option in the last year.	Free agents are risky, which to the Thunder means they are “chances” with a lower likelihood of success. There’s less firshand data to evaluate, and they are expensive (meaning the downside risk is high). So they avoid them as a team building tool.
Player development: Incorporated advanced analytics earlier than most (all?) other teams. Best 2 players were drafted outside the top 10. Shai developed beyond reasonable expectations.	Great evaluation + experiments increases the likelihood of each player reaching a championship level. Acquiring players via the draft reinforces this capability because it gives them more time to evaluate and develop players while they are not expensive.
Team Culture: Use the G-League and the regular season to intentionally experiment with unconventional tactics. In particular, playing Shair, Chris Paul, and Denis Schroeder together helped Shai develop and is not something any other team had done before. Created a play style that could “absorb a lot of different talent”.	Few teams use the G-League the way the Thunder do (the ones that do are also regularly competitive). Presti talks about a “discover/beginner’s mindset” and how even great scouts are mostly guessing at player evaluation. Details are scarce on how they believe their play style helps them absorb different types of players, but it reinforces the policy of maximizing the odds of each player raising their level towards a collective championship capability.

---

Additional Thoughts

After this exercise, I still do not have clarity on how to deal with multiple policies. While I believe the Thunder actually have a single overarching policy, the pistachio and Crown cases still seem to shoehorn independent things into one. Will Larson, whose work on strategy I greatly respect and whose blog first introduced me to “Good Strategy Bad Strategy”, freely adds multiple policies in his examples. My friend believes you need to move up an abstraction layer if the policy has multiple things bolted together. I’m undecided, even as my document has evolved from “multiple policies” to “single policy” in this OKC example.

I’m also not convinced I nailed it. Presti calls out “trading on time and playing the empirical odds” in his letter to fans from 2019², which reads as a Rumelt-like policy. My document speaks to this as part of the diagnosis instead. And I don’t love how wordy my policy ended up. Perhaps there’s some document in the Thunder intranet that’s clearer and more succinct, but I’ll never see it.

So I’m left with open questions …and yet, I’m satisfied with the document as written. It’s coherent, and I believe something like it would be useful for a team⁷. “Are we trying new things that will empirically help us?” and “How do we know that these things will work?” are questions that should emerge in an organization that has internalized the diagnosis and policy here. I certainly wish that my favorite team operated with those principles in mind⁸.

Lastly, I use a framing of “Problem Land” vs. “Solution Land” as I work through problems. My experience is that getting the Problem stated well is the hard and crucial part, and that once you do that good Solutions fall out of the sky. It rhymes with this document. The crucial insights are around 1) the proabilistic nature of the NBA making it an odds game 2) the constraints around spending and 3) the recognition of patience as an asset. One could imagine slightly different policies and actions, but the Diagnosis itself drastically narrows the problem space. My intuition is that there’s much room to improve what I’ve written here, but that there’s an 80/20 benefit to just writing anything like this at all.