Strokes Gained for Golf Betting: How Sharp UK Punters Read the Data
Available in US
Nine years of writing about golf betting has taught me one stubborn lesson — the punters who consistently beat the closing line aren’t reading more form pages than everyone else, they’re reading different ones. While the casual market still leans on world ranking and last week’s finishing position, the sharper end of the UK book has been working off Strokes Gained data for the best part of a decade. The metric was introduced in 2011 by Mark Broadie at Columbia University, built off the PGA Tour’s ShotLink shot-by-shot tracking system, and it has gradually reshaped how outrights, top-finish markets and matchups are priced.
The promise of Strokes Gained is simple enough — instead of measuring outputs (where a player finished), it measures inputs (how many strokes a player gained or lost on the field at each phase of the game). The application to betting is less simple, and that gap between knowing the metric exists and using it to find value is where most retail punters get stuck. This guide walks through the categories, the predictive weight of each, the venues where each matters most, and the workflow I use when I sit down on a Tuesday afternoon to build a card for the week’s tournament.
None of this is theoretical. The bookmakers themselves price off Strokes Gained models — every major UK operator runs one in the background, even if the front end still quotes fractional odds. The edge for the punter isn’t in having access to a metric the layer doesn’t have. It’s in interpreting that metric with course-fit, recency-windowing and field-strength adjustments the modelling teams sometimes get wrong.
Table of Contents
- What Strokes Gained actually measures
- The four Strokes Gained categories — and the two composites
- Strokes Gained versus OWGR — what each tells you
- The putting problem — why SG: Putting deserves the lightest weight
- Matching Strokes Gained categories to course archetypes
- The DataGolf model and how it differs from raw SG
- Building a value bet — from Strokes Gained to staked ticket
- How Strokes Gained moves during the event itself
- The blind spots — what Strokes Gained doesn’t see
- The toolkit — where to actually find the data
What Strokes Gained actually measures
I keep a notebook with the worst pre-bet decisions I’ve made, and the recurring theme on the early pages is that I was betting on finishing position when I should have been betting on shot quality. A player who finishes T15 having shot 65 in the final round looks very different from a player who finishes T15 having shot 73 — but on a results-only form line they look identical. Strokes Gained exists to make that difference visible.
The mechanic is straightforward. ShotLink, the PGA Tour’s tracking system, records the location of every shot at every tournament — about three million shots per season. For each starting location (fairway 150 yards out, rough 80 yards out, fringe 30 feet from the cup, whatever it is), the model computes a baseline expected number of strokes to hole out, based on the average tour pro starting from that spot. A player who takes one stroke from a position where the average pro takes 1.18 strokes has gained 0.18 strokes on the field. Add up every shot of a round, and you get that player’s Strokes Gained: Total for the round, relative to the field that week.
The framing is the breakthrough. Traditional stats like driving distance, greens in regulation and putts per round each measure one slice of a performance but don’t aggregate cleanly. A player who hits 13 greens but two-putts every one of them is statistically average. A player who hits 9 greens but holes a 30-footer for birdie and gets up-and-down from the bunker twice is statistically average on the same line. Strokes Gained collapses all of that into one currency — strokes versus the field — and lets you compare players across rounds, courses and tours.
The result is a measurement that strips out luck of the draw, weather wave and field strength and isolates the part of performance the player actually contributed. For a betting analyst, that’s gold. You can take the four most recent SG: Total numbers for a player and have a better predictor of next week’s performance than the OWGR or the last three finishing positions will give you. Whether that predictor is good enough to beat the bookmaker’s closing line is a separate question — and it depends entirely on what you do with the data next.
The four Strokes Gained categories — and the two composites
The first time I tried to use Strokes Gained for a bet I made a mistake that took me about six months to diagnose. I was looking at SG: Total, decided a player was in form, and backed him into a course where his actual deficiency had nothing to do with the holes that mattered. The lesson — SG: Total is the headline number, but the four sub-categories underneath it are where the actual betting edge lives.
The four categories partition the game by the location of the shot. SG: Off-the-Tee measures performance on tee shots at par-4 and par-5 holes — distance, accuracy and the trade-off between them. A player who bombs it 320 yards into the rough loses less Strokes Gained than the conventional wisdom would suggest, because the model values the proximity to the hole more than it punishes the missed fairway. SG: Approach measures everything from the fairway, the rough or the trees on shots intended to reach the green. This is where elite players make their living — Scottie Scheffler’s iron play is the single biggest reason he sits at world number one, not anything to do with his putter.
SG: Around-the-Green covers shots within 30 yards of the green that aren’t putts — chips, pitches, bunker shots, fringe putts that don’t qualify as putts. The category is small in volume but matters disproportionately at courses with tight green complexes or aggressive bunkering, where the average tour pro converts up-and-downs at roughly 60% but the gap between elite and average short-game players is wide. SG: Putting measures putts struck from the green surface.
Two composites sit on top of the four. SG: Tee-to-Green is the sum of OTT, APP and ARG — essentially everything except putting, which captures ball-striking quality. This is the metric I weight most heavily in my own modelling, because it has the highest year-on-year correlation with itself. A player who is strong tee-to-green this season was almost certainly strong tee-to-green last season. SG: Total is OTT + APP + ARG + Putting, and is the headline number you’ll see quoted in most coverage. It includes putting noise, which is the next problem to address.
Strokes Gained versus OWGR — what each tells you
The official world golf ranking and Strokes Gained are not enemies. They tell different stories about different parts of the field, and the punter who uses them in combination has a sharper read than the one who picks just one. The mistake I see most often in retail punting is treating OWGR as a complete picture — which it isn’t, and never was meant to be.
OWGR is a results-weighted ranking. It rewards finishing position over time, with stronger weighting for stronger fields and recent events. It captures durability and consistency, the ability to show up at events that matter and produce results. What it doesn’t capture is the quality of the shots that produced those results — and that gap matters for a punter trying to project next week’s performance.
Research published in 2023 by Laaksonen, looking at 19 major championships from 2018 to 2022, found OWGR was the strongest single correlator with outright odds at the top of the market, but Strokes Gained was the stronger predictor for lower-ranked players in the field. The reason is intuitive — for a top-10 player who’s been on tour for years, OWGR has accumulated enough sample size to be a reliable summary. For a player ranked 80th, with fewer big-field appearances, the ranking lags the player’s actual current ability, and Strokes Gained over the last 24 rounds is a better read.
The practical implication for outright betting at majors is that I use OWGR at the top of the market to set baseline probabilities for the favourites, then layer Strokes Gained over the mid-tier and longshot tickets where OWGR is least informative. At a regular Tour event with a weaker field, I lean more heavily on Strokes Gained throughout, because OWGR rewards major appearances and the field strength adjustment is doing most of the work in the ranking.
For matchup markets — the head-to-head 2-ball and 3-ball bets that run all week — Strokes Gained is the dominant input. OWGR contains too much accumulated history to be useful for a 4-round head-to-head call. SG over the last 12 to 24 rounds, ideally split into category-specific reads against the course profile, is the better tool. The bookmaker’s matchup line is built off the same data, so the punter’s edge depends on adjustments at the margin.
The putting problem — why SG: Putting deserves the lightest weight
The most uncomfortable finding I’ve absorbed in the last year came from an academic paper that crossed my desk in mid-2025. A Bayesian analysis of Strokes Gained Putting on the PGA Tour, published as arXiv preprint 2506.21822, concluded that putting skill is statistically nearly indistinguishable from noise once you control for sample size and field variance. Most of the observed week-to-week variation in SG: Putting is random, not skill.
The paper sent me back to my own records to check whether I’d been over-weighting putting in my model. I had. Players who’d had a hot putting week were getting a probability bump in my own outright projections that the data didn’t justify. The Bayesian finding doesn’t say putting doesn’t matter — it absolutely does, every Sunday afternoon. What it says is that the predictive value of last week’s SG: Putting for next week’s SG: Putting is much weaker than the equivalent relationship for SG: Approach or SG: Tee-to-Green.
The implication for betting is concrete. When I read a Tour pro’s recent form line and see a strong SG: Total driven mostly by SG: Putting over the last three events, I now discount that signal. The player has had three good putting weeks. The base rate says the next three weeks will regress towards their long-term mean. When the same player has a strong SG: Total driven by SG: Approach and SG: Tee-to-Green, I trust the signal more, because the underlying ball-striking quality is stickier.
This is the kind of insight that’s rarely surfaced in mainstream golf coverage but reshapes the way a sharp punter sizes positions. Backing a player whose recent form was carried by a hot putter is closer to a coin flip than the form line suggests. Backing a player whose recent form was carried by elite iron play is closer to a real edge. The market doesn’t always make this distinction, which is where the value sits.
Matching Strokes Gained categories to course archetypes
The most expensive mistake a new SG-driven punter makes is treating Strokes Gained as a context-free ranking. The categories matter differently at different venues, and the punter who learns the course-by-course pattern has an edge over the one who just stacks the strongest SG: Total players on every card.
Augusta National, host of the Masters, is the cleanest example. Strokes Gained: Approach correlates with Masters finishing position more strongly than any other category over a four-year sample, with SG: Around-the-Green a clear second. The reason is architectural — Augusta’s greens are large, severely contoured and demand precise iron play to leave makeable putts; missed greens trigger short-side recoveries off the slopes that punish weak short games. SG: Off-the-Tee matters less because the fairways are generous and driving distance, while useful, isn’t the primary differentiator. Backing a player whose recent SG: Approach is elite at Augusta has a structural tailwind the market doesn’t always price in.
The US Open lives at the opposite end of the spectrum. The USGA’s setup philosophy — narrow fairways, deep rough, firm greens, par-protected scoring — rewards SG: Off-the-Tee and the scrambling component of SG: Around-the-Green. Bombers who can’t keep it in play give up too many strokes off the tee; players who hit fairways and recover when they don’t are structurally favoured. SG: Putting variance is wide on US Open greens, which makes putting form a less reliable read than at most events.
The Open Championship and links golf generally reshape the categories again. Wind exposure means SG: Approach is volatile (a perfect iron shot from 175 yards can finish 40 feet from the hole if the wind shifts mid-flight) and SG: Putting on the firm, fast Open greens rewards patience and lag putting more than aggressive line-reading. Players with strong wind-game histories outperform their pre-tournament SG: Total at Open venues — Royal Birkdale in 2026 will reward this same archetype.
For regular Tour stops, I keep a simple typology. Bermuda-grass tracks with smaller greens (Sea Island, Sedgefield) reward SG: Approach. Bentgrass tracks with wide fairways and long carries (Memorial, Olympia Fields) reward SG: Off-the-Tee. Coastal courses with wind exposure reward SG: Around-the-Green and the patience that goes with it. Mountain courses at elevation, like the West Coast swing, reward power and reduce the impact of putting variance because the greens are typically softer.
The DataGolf model and how it differs from raw SG
Strokes Gained is the raw input. The most influential value-betting tool in golf, DataGolf, is a model layered on top of it. The site was founded by twin brothers Matt and Will Courchene, both with statistics PhDs, and over the last few years it has become the closest thing the sport has to a public-facing Bayesian projection engine. Every serious UK golf bettor I know either pays for the DataGolf subscription or reads it secondhand through the coverage that quotes it.
The model’s headline output is a projected probability for each player to win, finish top-5, top-10 and top-20 at a given event. Those projections are built from a time-weighted average of past Strokes Gained, adjusted for field strength, with around 60,000 simulations run for each prediction. The 2-sigma error band on a 50% probability projection is 0.4%, narrowing to 0.24% at a 10% probability — the model is genuinely precise within its own framework.
The DataGolf founders have been refreshingly direct about what makes the work hard. In a Q&A published in 2025 they noted that “the most challenging aspect to predicting golf is actually finding things that are meaningfully predictive! If you have a baseline model for projecting golfer performance that just uses a time-weighted average of their past relative scores — adjusted for field strength — this will be surprisingly hard to improve upon.” The implication is sobering — most of the apparent edges retail punters think they’ve found are noise.
What DataGolf does that a punter working from raw SG can’t easily replicate is the field-strength adjustment. A 1.5 SG: Total at the Hero World Challenge against an 18-player elite field is a much weaker signal than a 1.5 SG: Total at a regular full-field Tour stop, but the headline number looks identical. DataGolf’s model adjusts for opponent strength using a Bayesian framework that’s mathematically rigorous and computationally heavy.
The model isn’t infallible. It treats course fit relatively coarsely, applying a course-history adjustment but not the kind of architectural-fit reading a human analyst can layer on top. It also can’t predict the things the data has no view on — a player’s mental state, a new equipment switch, a course renovation. For the depth of how the model is built, where its strengths sit and where to apply human overrides, our deeper breakdown of the DataGolf model’s mechanics covers the methodology piece by piece.
Building a value bet — from Strokes Gained to staked ticket
The point of all this isn’t to admire the metric. It’s to find prices that are wrong. The workflow I use on a Tuesday morning when the markets open turns Strokes Gained data into a sized bet through three steps — implied probability from the odds, my own probability from the data, and a stake if my probability is high enough above the implied.
Step one is the easy maths. A player priced at 30/1 to win has an implied probability of 1 divided by 31 — about 3.2%. A player at 20/1 implies 4.8%. A 50/1 implies just under 2%. I write these numbers down for the top 30-40 players on the card. The implied probabilities will add up to more than 100% because of the bookmaker’s overround — the operator’s built-in margin — but the relative ranking of implied probabilities is still informative.
Step two is the harder bit. I take the DataGolf projected win probability for each player, then apply two layered adjustments — a course-fit adjustment based on the SG-category profile of the venue, and a recency adjustment if the player has had an unusual finishing position in their last two starts. For the course-fit adjustment, I’ll add a few tenths of a percent to a player whose SG: Approach is elite at an approach-heavy venue; I’ll subtract from a player whose recent SG: Total is carried by putting noise. The numbers are small — half a percentage point here, a third there — but they accumulate across a card.
Step three is the decision. I bet only when my projected probability exceeds the bookmaker’s implied probability by 2 percentage points or more — a hurdle high enough to absorb the overround and leave a genuine edge. A player I project at 5.5% who’s priced at 3.2% (30/1) clears the bar; a player I project at 4% at the same price clears it more marginally. Stake size scales with the edge — a 3-point edge gets twice the unit of a 1.5-point edge. Below 2 points of edge, I pass.
The discipline is in the passing. Most weeks, fewer than a quarter of the players I model clear the edge threshold. The temptation to bet a name I like, or a player I have a hunch on, is real, and the path back to losing money runs through ignoring the edge calculation in favour of intuition. The Strokes Gained workflow only works if you stake what the model says and skip what it says to skip.
How Strokes Gained moves during the event itself
Tournament-week Strokes Gained data updates round by round, and the punter who’s paying attention through Thursday and Friday gets a second cut of value the pre-tournament market doesn’t reflect. Live Strokes Gained for the round is published on PGA Tour stats during play and finalised at the end of each round.
The most useful application is Friday afternoon, when the cut line is settling. A player who’s two over par after 36 holes but has posted 1.5 SG: Tee-to-Green over both rounds is a better weekend bet than the leaderboard suggests — they’ve played well, the putter hasn’t cooperated, and the regression-to-mean has 36 holes to play out. Conversely, a player who’s six under par but only 0.3 SG: Tee-to-Green has been carried by the putter and is more likely to slip back over the weekend than the position implies.
The matchup markets for Saturday and Sunday rounds are where this insight cashes most cleanly. The bookmaker’s matchup price for a Saturday 2-ball pairing is set off pre-tournament numbers with some adjustment for the first two rounds’ finishing positions. If you can layer on the underlying SG quality, you’ll regularly find prices that are slightly off — not by a lot, but enough to build a card of half-a-unit bets that produce a positive expectation across the weekend.
Live outright re-pricing is less productive for retail punters. The book moves faster than any human can on outright markets once Saturday play is under way, and the margins available are thinner than the matchup side. I rarely chase live outrights unless something has happened — a withdrawal, a weather-driven leaderboard reshuffle — that the early Sunday line hasn’t fully digested.
The blind spots — what Strokes Gained doesn’t see
Every analytical tool has a frame of reference, and every frame of reference has things it leaves out. Strokes Gained is no different. The metric was built off PGA Tour shot-tracking data, and its predictive value degrades at the edges of that frame in ways the punter needs to understand before staking real money on the back of it.
The most important blind spot is weather wave. A morning-afternoon-morning-afternoon draw at the Open Championship can produce a four-stroke field-average gap between the two waves over 36 holes, and the Strokes Gained calculation that’s normalised against the field doesn’t fully correct for it. A player whose SG numbers look weak from the wrong side of the draw might be a better player than the data suggests; vice versa for the right side. The DataGolf model has a wave-adjustment but it’s coarser than the raw weather data a human can read.
Sample size is the next constraint. Strokes Gained over the last 24 rounds is reliable; over the last 6 rounds it’s noisy. A player coming off two events at a new sponsor, or returning from injury, has too small a recent sample to project from confidently. The temptation to over-weight a player’s last three rounds because they look exciting is real, and the discipline is in waiting for the sample to refill.
Tour transition is another. DataGolf publishes adjusted SG numbers for DP World Tour events that calibrate the field-strength gap with the PGA Tour, but the calibration isn’t perfect. A player who’s been crushing DP World Tour fields but hasn’t played a PGA Tour event in eighteen months carries more projection uncertainty than the model implies. Mental form, equipment changes, the death of a parent, a new coach, a contract dispute — all the human factors a results-blind data set can’t capture — sit outside the metric.
The takeaway isn’t that Strokes Gained is unreliable. It’s that the punter who treats it as the entire frame of analysis is mistaking the map for the terrain. Strokes Gained gets you 70% of the way to a defensible projection. The remaining 30% is course knowledge, weather reading, news flow and the small qualitative judgements no model captures.
The toolkit — where to actually find the data
None of the methodology above matters if you can’t access the underlying numbers. The data ecosystem for serious golf bettors has consolidated over the last few years around a handful of resources, most of them free or low-cost.
The PGA Tour’s own stats portal at pgatour.com publishes Strokes Gained data for every player, broken down by category and updated weekly. It’s the canonical raw source, and it’s free. The interface is dated and the historical filters are limited, but for last-12-events SG numbers it’s the first port of call.
DataGolf is the premium tool. The subscription is around the cost of one round at a decent UK course per month, and the analytics depth — model projections, course-fit overlays, live tournament tracking — is unmatched in the public market. For Tier 2 bettors getting serious about value modelling, it’s the closest thing to a must-have I can point to.
Fantasy National is a useful free supplement, with cleaner course-history data and salary-cap modelling that’s borrowed insights from the daily-fantasy ecosystem. OWGR.com remains the official source for world-ranking history. For the academic side of golf analytics, the arXiv archive hosts a growing body of peer-reviewed work, with the 2025 putting-noise paper among the most recent and most useful for the value-betting framework I’ve walked through above.
How many recent rounds of Strokes Gained data should I weight when sizing a golf bet?
The sweet spot for projecting next week’s performance is the last 24-36 rounds, which works out to roughly the last 6-9 events for a player on a full schedule. Below 12 rounds the sample is too small to be reliable; above 50 rounds the data starts to lag actual current ability. Weight the most recent 12 rounds at roughly double the older rounds in the window to capture form trends without over-fitting to a hot streak.
Can I bet using only DataGolf rankings without doing my own analysis?
You can, but you’ll forfeit most of the edge. DataGolf’s model is precise within its frame but applies course-fit relatively coarsely and doesn’t capture qualitative factors like equipment switches or weather wave. The punters who beat the closing line use DataGolf as the foundation and layer on their own course-fit overlay and recency adjustment on top. Pure model-following will roughly track the bookmaker’s own modelling, leaving the margin to the overround.
Which Strokes Gained category is most predictive at the four majors?
SG: Approach is the most consistently useful single category across all four majors, because elite iron play translates well across architectures. The Masters and Open Championship reward SG: Approach plus SG: Around-the-Green; the US Open shifts weight to SG: Off-the-Tee and the scrambling part of SG: Around-the-Green; the US PGA varies year to year with the venue. SG: Putting is the weakest single predictor at every major because of the noise documented in the 2025 Bayesian analysis.
Written by the editors at Golf bet.
