Danelfin vs Kunkafa: AI Scores and Price Probabilities

Danelfin and Kunkafa both describe AI-based market analysis, but their published outputs answer different questions. If you are evaluating alternatives, first decide whether you want a stock-selection score or a view of possible price moves for a market.

Kunkafa publishes this comparison. It is based on product descriptions checked on 7 September 2026, not a controlled test of either product's investment performance. No winner or return advantage is claimed.

What Danelfin's AI Score describes

Danelfin describes an AI Score from 1 to 10 that rates stocks and ETFs according to their probability of beating the market over the next three months. Its documentation groups scores into buy, hold and sell ratings. The score is a ranking-scale output: a score of 8 should not be read as an 80% chance of profit. See Danelfin's explanation of the AI Score.

What Kunkafa's price-level probability describes

Kunkafa describes a forecast using a price move, a direction, a duration and a chance of reaching that level. It shows both directions. Its how Kunkafa calculates market probabilities explains the definitions, and its Kunkafa forecast performance results separates reaching a level from finishing in a direction.

A level-touch probability is not a stock ranking, an expected return or a direct instruction to buy. An upward level and a downward level can both be reached during the same period, so the chances need not add to 100%.

Why the percentages are not interchangeable

Consider an illustrative share priced at 100. It rises to 105, falls to 95, and finishes at 101. During the period it touched both an upward and a downward level. Whether it beat a benchmark depends on the benchmark's return over the same period. Whether an investor made money also depends on entry, exit, size and costs.

Those are separate events. Before comparing accuracy claims, write down the event and the observation window. A prediction about touching 105 cannot be evaluated using only the final closing price, and a ranking against a benchmark cannot be evaluated by asking whether any upward move occurred.

How to evaluate alternatives fairly

• Compare the same assets and dates, using outputs recorded before the outcomes.
• Ask how unresolved outcomes and missing data are handled.
• Distinguish a provider's historical analysis from an independently reproduced result.
• Do not treat a forecast's accuracy as a trading return; order execution and costs are separate.
• Choose the output that answers your research question, rather than choosing the largest-looking number.

For background, read AI stock prediction accuracy and what a 70% chance means. For Kunkafa's product views, Kunkafa market forecast demo. Neither a favourable score nor a high probability removes the possibility of loss.