Home Methodology Pick Tracker Subscribe →
How it works

The
Model.

A systematic, data-driven framework for finding edge in PGA Tour betting markets — built on the same principles used in quantitative finance.

5+
Years of course
history modeled
3
Primary filters
per tournament
SG
Strokes gained
core framework
0
Gut feels.
Ever.
01 — Philosophy

Built like a quant model.

Golf betting markets are inefficient. The public prices players on name recognition, recent narrative, and TV airtime — not on data. That inefficiency creates opportunity. Our job is to find it systematically, every single week.

Core principle

"Every course has a fingerprint. Every player has a stat profile. When those two things match — and the market hasn't priced it in — that's the edge."

— StrokesEdge framework
02 — Process

Three steps to every pick.

The same process runs every single week before a PGA Tour event. No shortcuts. No exceptions.

01
Course profiling — identify what actually matters here

Before looking at a single player, we analyze the course. Using 5+ years of historical data, we identify which Strokes Gained categories most strongly correlate with finishing position at this specific venue. Not what the commentators say matters — what the data says matters.

Every course is different. TPC Sawgrass rewards SG: Approach and penalizes approach misses. Augusta rewards SG: Putting and ball-striking. Bay Hill rewards driving accuracy. The model weights each stat category based on its actual predictive power at that course — not conventional wisdom.

SG: Off-the-Tee SG: Approach SG: Around-the-Green SG: Putting Scrambling % Driving Distance Driving Accuracy
02
Player screening — match stat profiles to course demands

With the course fingerprint established, we run the field through the model. Each player gets a course-fit score based on their rolling performance in the stats that matter most at this venue. We use multiple time windows — last 8 events, last 24 events, last 50 events — to separate in-form players from structural fits.

A player who's gained 0.8 strokes per round on approach over their last 24 events will score well at an approach-dominant course regardless of whether they won last week. The model cuts through recent noise to find structural advantages.

Last 8 events Last 24 events Last 50 events Course history Form weighting
03
Value screening — find where the market is wrong

Course-fit score alone isn't enough. A player can be a perfect course fit but still be a bad bet if the market has already priced it in. The final step compares each player's model-implied probability against their market odds to find pricing gaps.

When a player scores highly on course fit but is priced long by the market — that's the edge. This is where picks like Akshay Bhatia at +6600 at Arnold Palmer, or Jacob Bridgeman at +9000 at Genesis, come from. The model found the gap. The market hadn't caught up.

Implied probability Market odds Value gap E/W value Place market
03 — Glossary

What the terms mean.

Golf betting uses a lot of jargon. Here's exactly what every term in our analysis means — no assumptions.

Strokes Gained (SG)
A stat that measures how many strokes a player gains or loses relative to the field average from a given situation. Breaks down into four categories: Off-the-Tee, Approach, Around-the-Green, and Putting.
Example: SG: Approach of +0.8 means a player gains 0.8 strokes per round on approach shots compared to the average tour player.
Course Fit Score
Our internal model score for how well a player's stat profile matches the demands of the course that week. Weighted by the stats that historically predict results at that specific venue.
Example: A high course fit score at TPC San Antonio means the player excels in SG: Approach — the #1 stat at that venue.
Each Way (E/W)
A two-part bet. The "win" part pays if the player wins. The "place" part pays (at reduced odds) if the player finishes in the top 5 or 6 depending on the sportsbook. Doubles your stake but gives you two ways to cash.
Example: Bhatia E/W at +6600. He placed T4 at Arnold Palmer — the win bet lost, but the place bet paid out at 1/5 of +6600.
Market Implied Probability
The probability the sportsbook's odds are implying a player has to win. Calculated by converting the American odds into a percentage. This is what we compare our model probability against to find value.
Example: +5000 odds implies ~2% win probability. If our model says 4%, that player has significant value.
Fade
A player we explicitly recommend betting against — or avoiding entirely. Typically a popular name whose game doesn't fit the course profile, or who is overpriced by the market relative to their course-fit score.
Example: A big hitter faded at a course where driving distance has low predictive power and approach play is dominant.
Top-20 Lock
Our highest-confidence finish market picks. Players whose course-fit profile suggests a strong probability of a top-20 finish, priced at a value relative to that implied probability.
Example: Tommy Fleetwood as a Top-20 Lock at courses where SG: Approach dominates — he's elite in that category and consistently finishes in the top 20 at those venues.
Value Play
The pick with the largest gap between our model-implied probability and the market odds. Usually a mid-range player (not the headline names) who scores exceptionally well on course fit but hasn't been priced accordingly.
Example: Ryo Hisatsune at +5000 at Valero — elite approach numbers, but priced like a longshot because he's not a household name yet.
Longshot
High-odds speculative plays where the model identifies a real statistical case for contention. Small unit size, big upside. Think of these as lottery tickets backed by data rather than hope.
Example: Jacob Bridgeman at +9000 at Genesis — scrambling numbers fit the course perfectly. He won. $5 returned $455.
04 — Pick Tiers

How picks are structured.

Every weekly analysis uses the same tiered structure. Each tier has a different risk/reward profile and a suggested unit size.

LOCK
Top-20 Locks
Highest-confidence finish market plays. Strong course fit + fair or better market pricing. These are the foundation of every weekly analysis. Suggested: 2-3 units.
VALUE
Value Play of the Week
The single biggest pricing gap the model identifies each week. One player whose course-fit score is significantly higher than their market odds imply. Suggested: 1-2 units.
LONG
Longshot
High-odds speculative play backed by a real statistical case. Not a random flyer — a player the model genuinely likes at a price the market has undervalued. Suggested: 0.5 units.
FADE
Fade Alert
Players to actively avoid or bet against. Typically overpriced names whose game profile doesn't match the course demands. Useful for DFS and head-to-head markets.
PARL
Parlays
Multi-leg bets combining top-20 and make-cut picks from the same week. Lower probability, higher payout. Only built when the individual legs are already strong standalone bets.
E/W
Each Way
Two-part bets used on longshot and value plays where the place market offers significant value relative to the win odds. Allows exposure to big payouts with partial downside protection.
05 — Principles

What we never do.

As important as the process is what we deliberately exclude from it.

No narrative betting
"He's due for a win" is not a reason to back a player.
Player narratives don't predict finishing positions. SG data does. We never back a player because of a storyline.
No recency bias
Last week's result is one data point, not a trend.
A player who won last week isn't automatically worth backing this week. The model looks at rolling averages across multiple time windows, not just recent form.
No name recognition premium
Rory McIlroy isn't always worth backing just because he's Rory McIlroy.
The market overprices famous names and underprices unfamiliar ones. That inefficiency is a core source of our edge.
No gut feel picks
If it's not in the data, it's not in the analysis.
Every pick has a data-backed rationale. If we can't point to specific stats that support a play, we don't make the play.
No chasing losses
Bad weeks happen. The model doesn't panic.
We don't increase unit sizes after losing weeks or abandon the process when picks miss. Long-term edge comes from process consistency, not reaction.
No selective transparency
Every bet — win or loss — is on the pick tracker.
The pick tracker shows everything. Wins, losses, open bets, P&L. No cherry-picking. No hiding misses. Full accountability, every week.

See the model in action.

Every week's analysis — course profile, picks, value plays, and fades — delivered to your inbox before the first tee shot.