If, before this last race, his odds of beating Jake and Josh them were 98/100, that makes his odds of losing to Jake and Josh 2/100. Assuming he meant this 2/100 odds of losing were to the unit [Jake and Josh], that would (overgeneralizing) make his odds of losing to either of them (root(2)/10 (or about 14%) (assuming they're equally good and always race together).
If he meant that his odds of losing to Jake was 2% and his odds of losing to Josh was 2%, this generalizes to all cases. The odds of him losing to Jake and Josh the last times he raced them would be .04%.
Assuming that 98/100 are his true winning odds against just Josh, the odds of going 8-1 against Josh would be ~1.7%.
Assuming that 98/100 are his true winning odds against just Jake, the odds of going 6-1 against Jake would be ~1.7%.
Assuming that root2/10 are his true winning odds against Josh, the odds of going 8-1 would be about 4%.
Assuming that root2/10 are his true winning odds against Jake, the odds of going 6-1 would be about 5.6%.
These last 4 calculations are a bit misleading because I was too lazy to do proper statistical analyses, but you get the idea.