I've always found it a bit silly how Retreat is amost entirely dependant on stats of who you are retreating from, rather than stats of who is retreating; Even if retreating model has a great initiative, you do not want to gamble on the fact that opponent may not get a full round of reactionary attacks.
So, what if we make it into a two-part initiative check for the retreating model?
1). Test your initiative. If the test is successful, you may move up to D6" away from any models you are engaged with, provided that retreating model didn't get seriously injured or OOA'd by reactionary attacks in step 2.
2). Regardless of result for previous initiative test, each engaged opponent gets a full round of reactionary attacks. However, you may cancel opponent's attacks by doing a successful initiative test for each attack.
This way retreating is still a risky endeavor, but at least fighters with Initiative 3+ or 2+ will feel a lot more slippery.
So, what if we make it into a two-part initiative check for the retreating model?
1). Test your initiative. If the test is successful, you may move up to D6" away from any models you are engaged with, provided that retreating model didn't get seriously injured or OOA'd by reactionary attacks in step 2.
2). Regardless of result for previous initiative test, each engaged opponent gets a full round of reactionary attacks. However, you may cancel opponent's attacks by doing a successful initiative test for each attack.
This way retreating is still a risky endeavor, but at least fighters with Initiative 3+ or 2+ will feel a lot more slippery.