I developed an algorithm to determine the support/resistance prices of stocks.But my question is What would be a good metric to measure the efficiency of the algorithm?simply considering the percentage of points outside the support/resistance lines might not be a good metric because then using the minimum/maximum of all points will give 100% accuracy.We would not need an algorithm at all.I am just curious to know what could be a good metric?
Asked
Active
Viewed 1,566 times
1 Answers
1
Given that the definition of support/resistance seems more descriptive than formalized, a good performance metric might not be easy to devise (If you devise the perfect metric, often you can derive the algorithm to optimizes that metric and it is a problem solved).
An alternative is to have experts determine the 'correct' (or at least, expected) value of the support/resistance for many stocks and see whether your algorithm provides values that are close or similar.
The ultimate test, though, is to take the end objective of what you are doing and use that as metric. Is your algorithm used in a trading strategy on specific stocks? Then you can use the money made or lost as a metric.
damienfrancois
- 52,978
- 9
- 96
- 110