diff --git a/kernel_tuner/strategies/simulated_annealing.py b/kernel_tuner/strategies/simulated_annealing.py
index 5055f302..962a1e34 100644
--- a/kernel_tuner/strategies/simulated_annealing.py
+++ b/kernel_tuner/strategies/simulated_annealing.py
@@ -59,7 +59,7 @@ def tune(searchspace: Searchspace, runner, tuning_options):
                     print(e)
                 return cost_func.results
 
-            ap = acceptance_prob(old_cost, new_cost, T, tuning_options)
+            ap = acceptance_prob(old_cost, new_cost, T)
             r = random.random()
 
             if ap > r:
@@ -90,7 +90,7 @@ def tune(searchspace: Searchspace, runner, tuning_options):
 
 tune.__doc__ = common.get_strategy_docstring("Simulated Annealing", _options)
 
-def acceptance_prob(old_cost, new_cost, T, tuning_options):
+def acceptance_prob(old_cost, new_cost, T):
     """Annealing equation, with modifications to work towards a lower value."""
     res = 0.0
     # if start pos is not valid, always move
@@ -104,10 +104,14 @@ def acceptance_prob(old_cost, new_cost, T, tuning_options):
         res = 1.0
     # maybe move if old cost is better than new cost depending on T and random value
     else:
-        if tuning_options.objective_higher_is_better:
-            res = np.exp(((new_cost-old_cost)/new_cost)/T)
-        else:
-            res = np.exp(((old_cost-new_cost)/old_cost)/T)
+        # we must have abs_diff <= 0, as new_cost >= old_cost
+        abs_diff = old_cost - new_cost
+
+        # relative to abs(old_cost), as the cost might be negative
+        rel_diff = abs_diff / np.abs(old_cost)
+
+        # exponential decay
+        res = np.exp(rel_diff / T)
     return res