vonT

There have only been five occurrences of more than 100 casualties. This late in the question, it's not likely to occur.

Since the armistice in 1953, there have been four major US troop reductions in ROK. The first was the immediate draw down while the subsequent three were during the Vietnam and Iraq Wars. Also, removing a BCT sized element from ROK (about 20% of total US forces on pen) would be significant geopolitical event. It may embolden PRC, DPRK, and RUS to try more aggressive actions in the region. Equally it would probably have a chilling effect on ROK and JPN, undermining their confidence in the US. Threatening in private to reduce troops may be a negotiating tactic of the administration but saying it publicly--even if they're just "considering" the action--is probably less effective.

9 Cases in 12 day is reasonable given monthly rates.

There's not sign of incidents slowing down over the past week. The average daily rate for this month would put us into the 1200's. Even a low rate would cross 1000 total. To break 1250, the average daily number of incidents would need to be 42.2 since 22 Aug. In August so far there have only been 5 days greater than that. Roughly 25% chance of a given day being over 42 and every day that isn't the bar gets higher. 

Average daily incidents for the most recent week were about 18, below the 23 incidents a day needed. In the final week, there'd have to be about 26 incidents a day on average to cross 690. It's possible, but I'm lowering the chances substantially. Playing the averages puts us even closer to staying below 650 than before. Still think it more likely than not they cross the 660 mark, but much less likely they go over 690.

Count is nearly 500, with about 16 incidents a day since last adjustment. We're still lagging the yearly average incidents per day, but slightly higher than the August average of 14. If the trend holds we stay under 550. 

Coming late to this and following the crowd and the trend...and taking a bit of a swing for score purposes. A 15 day average of 21 and a 30 day average of 23 events a day gives a range of 655-689 for August. I tend to bias towards the more pessimistic outcome which tracks with the crowds' reasoning as well. Going north of 690 is more likely but still requires an increase in daily average over the past month. 

With just over a month to go we have over 1400 cases. The average number of cases per week for the year is 42. Over the last three months it is about 18. With five weeks to go the weekly average would put us a hair under 1499. Anything close to the full year's weekly average and it will cross the 1500 mark. Effectively we are looking at one moderate outbreak to put us over the 1500. With schools starting back up, it is reasonable to expect some increase in cases that would push the weekly average above 20.