I think you're missevaluating the 8,9,12 rule here and I am clueless as to how you came up with 12.5! You're also overgeneralizing based upon a straight race since the contact from the 5 anchor creates some volatility and favors you somewhat.
I use my own methods for races with my own adjustments. I doubt I could teach this in a manner that would be easy for anyone to understand, but here's a bit of it assuming a similar pip count pure pip count oriented race that needs no adjustments.
For a race this long, 10%+2 as point of last take/first pass is superior to 12% which is too much. What I am looking for is the point of last take or first pass which at this pip count is 21.4% cubeless.
                    |
| ||||
| XGID=-----BBBCBBB-bbbbcbb------:0:0:1:D:0:0:3:0:10 | |||||
| White to play 65 | |||||
| Analyzed in Rollout | No double | Double/Take |
| Player Winning Chances: | 78.57% (G:0.05% B:0.00%) | 78.60% (G:0.00% B:0.00%) |
| Opponent Winning Chances: | 21.43% (G:0.00% B:0.00%) | 21.40% (G:0.00% B:0.00%) |
| Cubeless Equities | +0.572 | +1.144 |
| Cubeful Equities | ||
| No double: | +0.892 (-0.105) | ±0.004 (+0.887..+0.896) |
| Double/Take: | +0.996 | ±0.007 (+0.989..+1.004) |
| Double/Pass: | +1.000 (+0.004) | |
| Best Cube action: Double / Take | ||
| Rollout details | ||
| 1296 Games rolled with Variance Reduction. Dice Seed: 46746902 Moves: 3-ply, cube decisions: XG Roller | ||
| Double Decision confidence: | 100.0% | |
| Take Decision confidence: | 83.6% | |
| Duration: 14 minutes 35 seconds | ||
eXtreme Gammon Version: 2.00.158.alpha