Butterfield Bermuda Championship: Chris Kirk Tops the Key Metric, but the PGA Tour's Pick Is Davis Thompson

At the top of this week's Strokes Gained: Approach leaderboard sits Chris Kirk. Second is Ben Kohles. Davis Thompson is third, followed by Luke Clanton, Zac Blair and Ben James. A tidy, readable column of numbers. And if you accept the argument the PGA Tour's own pre-tournament analysis makes — that at Port Royal, approach play is “the first and most important place to look” — then the name worth discussing is on the first line of that table.

The article does not do that. It picks the third name, at +4000, under a headline telling readers to look for Davis Thompson and his hot iron play.

That is the detail that stopped me. Not because Thompson is an unreasonable pick — he has a case. But because the piece builds a hierarchy of analysis and then steps outside it without a word of explanation. In my trade, when an analysis does that, there are usually two possibilities: the author knows something the reader does not, or the author holds a conviction the data cannot carry. Both are worth examining, and this article does not supply enough material to rule out the second.

Based on my experience following pre-tournament coverage across many seasons, I read this kind of piece on two layers: the argument and the evidence. The first layer is solid. The second is far thinner than the tone suggests.

Port Royal: a course built to disarm the long hitter

Start with what the article does well, because it does it quite well.

Port Royal Golf Course is a 6,828-yard par 71 on the island, designed by Robert Trent Jones, with Bermuda greens averaging roughly 8,000 square feet and water on only four holes. Four holes. On a coastal course where wind is the governing variable, having water on just four holes means most of the penalty area is rough and the player's own head.

At that yardage, driving distance is largely neutralised. A player who hits driver 320 yards and one who hits it 285 will approach the greens from distances that barely differ, because both are clubbing down off the tee. When the distance edge is removed, scoring separates in two places: approach precision and green reading.

That is a universal principle of the modern tour, not an invention of this article. Strokes Gained: Approach is the category most tightly correlated with scoring. On a short course with large greens, that correlation tightens further.

The context is also correctly drawn. The Butterfield Bermuda Championship sits in the FedExCup Fall — the post-Playoff stretch where 125th is the survival line for tour membership and where Aon Swing 5 and Next 10 decide next season's Signature Event access. For a player grinding at the bottom, a top-10 here can change a career in the literal sense. For a star, it means nothing. That incentive structure explains why the field is thin.

And the thin field is the largest variable the article never names.

The evidence chain breaks exactly where it matters

The article's strongest asset is a single number: Thompson ranks 3rd in the field in SG: Approach over his last 36 rounds and 1st over his last 24, sourced to BetspertsGolf. That is specific, checkable, genuinely impressive data.

But the article supplies no other number.

No SG: Off the Tee. No greens-in-regulation percentage. No driving accuracy figure. No quantification of the putting deficit. No SG: Total — the one metric that would settle whether Thompson is actually playing well or merely striking it well while scoring badly.

And most importantly: no Strokes Gained data from previous editions at Port Royal. The article concedes this. This is where I want to linger, because it is the substance of the problem.

Every course-fit claim in the piece is geometric inference, not measured performance. From 6,828 yards, from 8,000-square-foot greens, from four water holes, the author infers that precision beats power. The inference is logically sound. But it has never been tested with data at this specific venue, and the article presents it as an established finding.

There is a technical hypothesis the piece never considers. Greens of 8,000 square feet inflate GIR% across the entire field. If Thompson is an accuracy-first player with modest distance, he will hit more greens from further out — but his proximity to the hole is likely only average. When the whole field is hitting greens, the approach advantage of a precise player compresses rather than amplifies. That is a hypothesis, and I am marking it as one.

The 24-to-36-round window carries its own problem. That is six to nine weeks of competition. Such a window is highly sensitive to two or three hot weeks and can flatter a player whose season-long baseline is materially lower. The article never gives Thompson's season-long SG: Approach. Only the short-window rank. The gap between those two numbers is the whole story.

The putter has no numbers attached

This is where the article digs its own hole.

Thompson is described as “remarkably bad” on the greens. The last time he gained strokes putting was at the Rocket Classic. Since then, no positive week. Not one.

A player who wants to win needs four consecutive neutral-or-better putting days. Thompson has not produced one in months.

The article handles the contradiction with a hypothesis: slower Bermuda greens might help him. The phrasing is that whether that applies to Thompson “remains to be seen, but the author is willing to consider it at this price”.

That is a conditional hope, not a projection.

Technically, the hypothesis conflates two things. Slow greens are a neutralising agent, not a corrective one. Slow surfaces reduce three-putt frequency across the entire field and shrink putting dispersion. They do not repair green reading, start-line control, or stroke mechanics. They compress variance. They do not restore skill.

A player losing strokes on the greens for months may be losing them for three reasons: technical, psychological, or physical. Slow greens touch the first only at the margins, and do not touch the other two at all.

The article also does not rule out the physical. There is not a line about Thompson's health. In golf, a prolonged unexplained putting decline is sometimes a downstream signature of wrist or lower-back trouble, because the kinetic chain transmits directly into the putting stroke. I flag this as an unknown, not a conclusion. But an analysis that spends 800 words on irons and four words on the putter is skipping the decisive variable.

When the stands are empty, the contest exposes what tactics conceal. Here, what is concealed is a putting stroke with no numbers attached.

The market priced it differently

There is a detail the article uses as a passing example that actually breaks the order it builds.

Denny McCarthy is quoted at +3300. Thompson is +4000. McCarthy is shorter.

The article cites McCarthy as an example of a non-long hitter who could contend. It does not explain why the market rates McCarthy above Thompson, when the article has just argued that distance is neutralised here and that approach play is the key metric.

The answer is fairly clear to anyone who reads pricing models: McCarthy is a complete player — elite putter, elite accuracy. The market is pricing the putter. The article is pricing the irons. Two parties looking at the same player and choosing different halves of the same profile.

That is a genuine model-versus-market divergence, and it deserves to be on the table. The article does not put it there.

Arithmetically, +4000 implies roughly a 2.4% probability. In a field of about 130, the random baseline is roughly 0.8%. The market has already given Thompson a premium of about three times an average entrant. The question is not whether Thompson is undervalued. The question is whether his true probability exceeds 2.4% — and book margin means the fair probability is lower still.

The article describes the author changing his mind: initially thinking an outright ticket was unlikely, then liking Thompson at 40-1 after breaking down Port Royal. That describes someone converging on the market, not diverging from it. Those are different claims, and the piece presents the second as though it were the first.

Rico Hoey is a different kind of bet altogether

The second selection is correctly typed but not the same class of argument.

Hoey is described as one of the best ball-strikers in the field, repeatedly undone by his own putter. His last four events: neutral at Bank of Utah, positive at Wyndham and the Rocket Classic, negative at Biltmore Asheville. A four-event sequence with no trend.

The author says he would bet Hoey to lead at any point during the tournament, not to win. That is the first-round leader market. Logically it is coherent: an elite ball-striker with a volatile putter is the textbook profile for an 18-hole bet, where putting variance compresses into a single sample.

But it is a different claim, with far higher variance. +4500 implies roughly 2.2%. First-round-leader markets also carry higher book margin than outright markets, meaning the real value of that ticket is likely worse than its face comparison to +4000 suggests.

What is more interesting is the analyst's behaviour. The author admits he has “lost count” of the times he backed Hoey on ball-striking conviction and watched the putter erase it. Backing the same player repeatedly, for the same reason, and getting the same negative outcome is the definition of anchoring bias. The shift to a one-round market is a discipline adjustment, but it is presented as a new insight.

The honest version of that argument would be: I have been wrong about this player repeatedly, so I am cutting my exposure to one round. That is a far more credible sentence than “his game is incredibly close to being back”.

A blank screen forces me to read the contest the way I read an unedited manuscript. Same here: the article breaks off mid-sentence at exactly the point where the counter-argument about Hoey would appear. The Hoey case cannot be fully assessed when its load-bearing section is missing.

Seven editions, four winners, and a sample that is too small

The article builds its “this course produces chaos” argument on one statistic: only one betting favourite has won in seven editions. Four winners are cited: Seamus Power at +2026, Adam Schenk around 50-1, Rafael Campos at triple digits, Camilo Villegas at triple digits.

Seven editions is a very small sample for venue-level conclusions. And four names come with no control group.

The structural point is this. A low-tier event with a weak field mechanically produces a wider outcome distribution. When the gap between the best player and the 120th-best in the field compresses, variance opens up on its own. A triple-digit winner is normal at FedExCup Fall events, whatever the course looks like.

Attributing that variance to Robert Trent Jones' design and Bermuda wind is a causal misattribution. To prove it, you would compare the long-odds winner rate at Port Royal against the same rate at peer fall events. The article has no such comparison group.

One detail also made me pause. The winner sequence in the piece — Villegas 2026, Campos 2026, Schenk “last season” — together with the fact that Thompson has one top-five finish in the 2026 season, is only consistent with a preview written in autumn 2026, when fall events count toward the following FedExCup season. If the piece in fact dates to autumn 2026, then “Schenk won last season” conflicts with the Campos-2026 attribution, and the entire winner list needs re-verification.

This does not affect the betting logic. It affects the reliability of the historical data the piece uses as its foundation.

The horses-for-courses paradox at an event where the field turns over almost entirely

This is where I think the analysis is conceptually wrong, not merely under-sourced.

“Horses for courses” is a legitimate framework. But it requires a condition: players must come back. You need a repeat group large enough to accumulate course knowledge and to generate a comparison sample.

At a low-tier fall event, that condition inverts. Most of the field is fighting for status and will not return next year. The ones who do return are the ones who failed to graduate to better events. Accumulated course knowledge at Port Royal correlates with not having progressed.

In other words, horses-for-courses here rewards a trait that is itself a marker of stagnation.

That is why I read “no Strokes Gained data from prior editions here” not as a small apology but as a statement about the limits of the entire method. A course-fit story needs one of two things: venue-level SG data, or a deep history of the same players returning. This event has neither. What remains is geometric inference dressed as an established pattern.

There is also a professional detail worth noting: “Horses for Courses” is a recurring column format. The frame is imposed before the analysis begins. Such a column has a structural bias toward finding a fit, because finding a fit is what it exists to do.

When the PGA Tour publishes its own betting picks

This is the most consequential fact in the whole story, and it has nothing to do with Thompson or Hoey.

The piece is published on PGATour.com. It quotes DraftKings and FanDuel odds by name. Both are official commercial partners of the PGA Tour.

A governing body that runs the tour, and that is responsible for competitive integrity and for its players' reputations, is operating a media channel that publishes named-player betting recommendations using its own partners' prices. This is now standard practice across US sports media, so it is not a violation. But it is a governance posture, not a neutral one.

Betting-content formats reward confident single-pick framing. In this piece the author himself concedes that “the odds of hitting an outright winner this week are lower than usual” — a sentence sitting directly beneath a headline naming a specific player to win. That tension is structural to the format. Readers should weight the caveat three or four times more heavily than the headline.

The ball rolls on the grass, but I am reading the money moving behind it.

The two-tier data gap

There is an industry-level meaning inside the sentence “no Strokes Gained data from previous editions here”.

ShotLink — the PGA Tour's shot-tracking system — is fully deployed at elite events and thinner at tier-two and fall events. Without ShotLink, the entire Strokes Gained analytical layer disappears, and the analyst falls back on raw indicators: GIR, fairways hit, putts per round.

That is a two-tier data economy, and it reinforces itself. The less data an event has, the less seriously it is analysed; the less seriously it is analysed, the less pressure there is to invest in measurement.

But betting content is expanding down-market, because that is where the cheapest audience sits. As betting-content demand reaches fall events, demand for granular data at exactly those events will rise. The gap the author is working around is a commercial opportunity for the data industry, and it is a signal rather than an apology.

I learned this from a transfer saga in 2026. When every outlet reported Frenkie de Jong's move to Barcelona at 75 million euros, I read the contract and found appearance-based variables that could push the total to 86 million. I did not publish. Three weeks later, Ajax confirmed that payment structure. The lesson was not that I was smarter. The lesson was that structure lives where nobody bothers to read.

At Port Royal, the structure lives where nobody bothers to measure.

The biggest risk is not the wind

The article spends a lot of space on wind, and wind is indeed the largest exogenous variable. When it blows, the course becomes a scrambling and trajectory-control test — a profile Thompson is not distinguished in. When it drops, the course is “very gettable” and the whole field's scoring distribution shifts sharply upward.

But the biggest risk is not the wind. It is the unquantified putter.

Every other risk in this analysis is secondary to that unexamined variable. A +4000 ticket is a low-probability event by construction. The analytical substrate beneath it is thin: no venue-level SG data, no season-long baselines, an unquantified putting deficit, a truncated argument, and a seven-edition historical sample.

High variance plus thin evidence is the definition of a high-risk analytical surface, however sound the course-fit logic may be.

A season is one sentence in a book that runs a decade. And one week at Port Royal is a comma inside that sentence.

Verification notes

Citable facts, with sources:

• Port Royal Golf Course: par 71, 6,828 yards, Bermuda greens of roughly 8,000 square feet, water on four holes, Robert Trent Jones design. Source: PGA Tour official media.
• Davis Thompson: 3rd in field SG: Approach over his last 36 rounds, 1st over his last 24. Source: BetspertsGolf, cited by the PGA Tour.
• Thompson's last positive putting week: the Rocket Classic. None since.
• Reference prices: Thompson +4000, Rico Hoey +4500 (first-round leader), Denny McCarthy +3300. Source: DraftKings, FanDuel.
• Cited winner history: Seamus Power +2026, Adam Schenk around 50-1, Rafael Campos and Camilo Villegas at triple digits. Only one favourite has won in seven editions.
• Thompson: one top-five finish in the 2026 season, no win for an extended stretch.

Unverified and requiring independent confirmation: the original publication date, the year-by-year winner list, the purse and OWGR points on offer, both players' OWGR positions, and their physical status.

What to watch after round one

If you want to test this analysis against reality rather than argument, there is one clean test: Thompson's SG: Putting after 18 holes. A neutral or positive day would validate the slow-greens mechanism. A negative day would suggest the problem is technical rather than surface-related.

A second test: compare the 24-round approach rank with the season-long figure as soon as it is available. A divergence greater than roughly 0.3 strokes per round would confirm the ranking is a window artefact rather than a form trend.

A third, less comfortable test: watch where Kirk, Kohles and McCarthy finish. If two of those three land in the top 10 while Thompson and Hoey fade, the problem is not the selection process. It is the framework.

Being cold is a long-term strategy, not a character flaw. I keep a three-layer rule for every number: verify the source, check the structure, and analyse the motive of whoever supplied it. Here, the supplier is simultaneously the tour, the publisher and the sportsbook's partner. Those three roles need not conflict. But readers should know who they are reading.

What I take from this piece is not a name to back. It is a question about how we read sport. When a short course, wide greens and sea wind turn every forecast into a probability, people still want a tidy story — one player, one reason, one number. Sport does not work that way. It works as a system in which the most important variable is usually the one nobody measures.

And if ShotLink arrives at Port Royal next year, we will know who was right. Until then, the question I keep is this: if Thompson's putter is not a surface problem but a wrist problem, how much does a beautiful approach ranking actually mean?