Change screening cutoffs, condition weightage, and entry model parameters. Saved values are written into config and will be used in the next source sync, shortlist, and stage-3 diamond run.
Screening Cutoffs
These are hard gates. If a stock misses one of these cutoffs, the stage-3 model will reject or skip it even if the other numbers look good.
Minimum closeness to the 52-week high. Higher means only stronger momentum names stay in the pipeline.
Maximum PE allowed in stage 3. Lowering this makes valuation stricter.
Upper size limit in crore. Lower values keep the system focused on smaller companies.
Minimum year-on-year sales growth required before the stock can qualify.
Minimum company-news score. Raise this if you only want candidates with stronger company-specific news flow.
Condition Weightage
These weights do not act like pass or fail rules. They control how much each condition affects the weighted quality score and final entry model.
Extra weight given to stocks trading very close to their 52-week high.
Influence of valuation quality. Increase this if PE discipline should matter more in ranking.
Influence of company size. Increase this if smaller market cap names should get more weight.
Influence of sales growth strength on the weighted score.
Influence of company-specific news quality. Increase this if business updates should move the entry model more.
Entry Model
These fields shape the modelled entry value after a stock clears the screen. Change them when you want the system to become more conservative or more aggressive on future runs.
Bonus added for each company-specific positive news item. Higher values reward strong business updates faster.
Penalty applied for each company-specific negative news item. Higher values make risk events hurt the score more.
Final discount applied to the model value to calculate the preferred entry. Lower means a more conservative entry price.
Weight of revenue growth inside the growth engine.
Weight of PAT growth inside the growth engine.
Weight of sales momentum inside the growth engine when revenue and PAT are not telling the full story.
AI Enrichment
Optional. When enabled with a Claude API key, the system re-checks company-news relevance and sentiment before scoring. If the AI request fails, the system falls back to deterministic rules automatically.
Turn this on only after adding your Claude API key.
Provider used for AI enrichment. Right now only Claude is implemented.
Model used for headline analysis. Example: claude-sonnet-4-6.
Claude API key stored in config so the server can call the model during sync and stage 3.
Use Test Key to confirm the current key and model are accepted by Claude.
Maximum number of headlines sent in one Claude request.
Timeout for the Claude API request. If it times out, the pipeline continues without AI.
Idle
{}