diff --git a/README.md b/README.md
index 1a3e5c5..9e5140b 100644
--- a/README.md
+++ b/README.md
@@ -2,4 +2,8 @@
 Rule engine supports only simple comparison operators. The trade-off of this approach is that you can't encode complex rules. To be able to support more robust rules we could embed a scripting languages like _lua_, possibly at the cost of performance bottlenecks.
 
 ### Performance
-for 200*4=800 sig and 3 rules this python code takes about 0.04 secs on a AMD6410
+for 200*4=800 sig and 3 rules this python code takes about 0.04 secs on a AMD6410. The complexity of the rule engine is O(1) on the rule list selection based on the signal, O(n) on the application of those rules.
+
+### Final Thoughts
+* Could have improved performance and reduced resouce usage by actually streaming the json, might introduce additionaly dependency
+* Add support for more features for the rules