+1 for actually having data and attempting to form a model. Your model is pretty rough though. Why are you assigning gains of SR to losses? That’s very odd.
You seem to be almost trying to a linear fit. If that is done, you get Placement SR = 230.3*wins+1305.2. It actually fits pretty well. Unfortunately, more data starts to break the fit.
I have these records (all new accounts):
0-0-0-1-1-1-1-0-0-0: 2210 My alt
1-1-0-0-1-1-1-1-0.5-0 2594 Kabaji alt (Rainbow)
1-1-1-1-1-0-0-1-1-0 3017 Kabaji alt (PlatChat)
0-1-1-1-1-0-0-0-1-1: 2834 Custa alt (Dildomum)
10-0: 3550 DongQuixote alt
7-1-2: 3267 DongQuixote alt
6-4: 3003 DongQuixote alt
7-3: 2780 DongQuixote alt
6-4: 2600 DongQuixote alt
6-4: 2300 DongQuixote alt
Note especially games there are four different, very different results, with 6 wins. I don’t think win streaks matter, as they no longer matter after placements.
However, game order does matter. Earlier games are worth more. Performance also likely matters, especially for top players. The first game is always around 2350 (quick play doesn’t matter). See Initial Competitive Skill Rating, Decrypted for more.