Full transparency on yesterday's R4 builds. 15 of our 150 MK lineups would have shipped a 200-max GPP. 48 hit top 1%. 124 cashed (82.7%). And the ONE single-entry build we sent at #1 by projection? Busted. Here's why — and the 3 model upgrades we're shipping next week because of it.
| Tier | Cutoff (DK pts) | Our 150 Lineups | % |
|---|---|---|---|
| 🏆 SHIPPED | 360+ | 15 | 10.0% |
| 🥇 Top 1% | 315+ | 48 | 32.0% |
| 🥈 Top 5% | 285+ | 97 | 64.7% |
| 💰 Cashed | 260+ | 124 | 82.7% |
That's an 82.7% cash rate from a 150-lineup MME — well above industry-standard ~50-60% for GPP MME. The process worked. But we need to talk about the one that lost.
Every SE lineup we sent yesterday, scored against actual R4 results:
| SE Recommendation | DK Score | vs Cash | Result |
|---|---|---|---|
| 🚨 MK #1 by projection — Bezuidenhout CPT | 225 | −35 | LOST |
| ✅ Cole CPT (leader anchor) | 346 | +86 | TOP 1% |
| ✅ Aberg CPT (chalk lock) | 280.5 | +20 | CASHED |
| ✅ Bhatia CPT (hot putter) | 313.5 | +54 | TOP 5% |
| ✅ Hughes CPT (max leverage) | 320.5 | +60 | TOP 1% |
| ✅ Gerard CPT (hot putter co-leader) | 336 | +76 | TOP 1% |
5 of 6 SE recommendations cashed. 3 of 6 hit Top 1%. ONE busted.
Our MK builder ranked Bezuidenhout #1 by projection. We sent that lineup as the single-entry "best build." He shot +4 R4. Lineup ended at 225 DK pts — 35 below the cash line.
The data, in hindsight: RRG's bounceback table HAD Bezuidenhout at #11 with positive ball-striking (+0.58) but negative putting (−0.92). Translation: he was hitting greens but couldn't make putts. Our projection model didn't separate putting from ball-striking — it aggregated everything into one number. That made him look like the slate's best play. He wasn't.
RRG's bounceback table predicted the top R4 bouncebackers from R3. Look at the top 5:
| # | Player | R4 Bounceback | Final Position |
|---|---|---|---|
| 1 | Russell Henley | +2.35 | 🏆 T1 (WON) |
| 2 | Zach Bauchou | +1.85 | T22 |
| 3 | Max McGreevy | +1.84 | T35 |
| 4 | Justin Thomas | +1.76 | T13 |
| 5 | Andrew Novak | +1.66 | T54 |
Henley was RRG's #1 bounceback target and WON the tournament tied with Cole. Our model didn't weight bounceback — that's a layer we're adding immediately.
Add RRG-style bounceback rank into the projection model. Top-25 bouncebackers get +5% weight bonus in R4-only builds. Catches Henley-type recovery plays the FTN model missed.
For putt-dependent courses (Colonial, Harbour Town), weight putting metric 2× normal. For ball-strike courses (Augusta, US Open setups), weight BS 2×. Stops the Bezuidenhout-type miss where good BS + bad putt averaged to a "great" projection.
Use RRG driver profile data (e.g., Scheffler 67.5% L-to-R, Aberg 45% L-to-R, Byeong Hun An 60.3% R-to-L) to match player shape to course dogleg direction. Right-to-left specialists get a +3% weight on courses with predominantly left-doglegs.
Bezuidenhout appeared in 55 of 150 (37%) of our MK lineups as CPT. That's over-concentration. New rule going forward:
The MK process is sound — 82.7% cash rate from a 150-lineup MME is elite. The single-entry process needs guardrails. We didn't have a "model conflict" check; tomorrow we do. We didn't separate putting from ball-strike in projections; this week we will. We didn't weight bounceback; we're adding it now.
Cole CPT, Hughes CPT, Gerard CPT, Bhatia CPT — all 4 cashed Top 5% or better. Echavarria-led builds averaged 342.9 DK. The methodology works. Tomorrow's it'll be tighter.
"The model that gets you 82.7% of the way still needs a circuit-breaker for the 17.3%. We're installing one." — Courtedge Research