v0.19 — clean layout + transparent criteria + public robot results

Green Investment Dashboard — Powered by AI

A cleaner decision-support dashboard for green investing, ETFs, AI-infrastructure ideas and long-horizon portfolio discipline. Use it to screen stocks/ETFs, inspect why each asset passed or failed, save watchlists and portfolios, compare performance versus S&P/global benchmarks, and monitor the daily paper robot. This is not an automatic buy signal; it is a transparent research and tracking tool.

Screen assets Selection criteria Compare portfolios Robot results Watchlists
1
Set rulesChoose country, horizon, ethical exclusions, currency and risk tolerance.
2
Update recommendationsRank stocks/ETFs, then inspect the pass/watch/fail selection criteria.
3
Save or compareAdd assets to a watchlist or save a portfolio for 3M/6M/1Y back-tests.
4
Track realityUse cloud/public robot records and benchmark comparison to see what worked.
Analysed
—
after hard filters
Top candidate
—
Run analysis
Median score
—
risk-adjusted
ETF share
—
within recommendations
Green average
—
green revenue / ESG blend
Excluded
—
ethics / country / size filters
Base currency
CHF
Checking API

How the recommendation engine works

Transparent sequence: hard exclusions → horizon-weighted score → ETF benchmark discipline → reality tracking.
explainable logic
1. Hard filters first.
Asset type, country/region, green-only, Israel exposure, defense and surveillance-sensitive exclusions are applied before scoring.
2. Score by horizon.
Shorter horizons emphasize momentum and catalysts; longer horizons emphasize quality, valuation discipline, balance sheet strength and risk control.
3. Track reality.
Every saved idea keeps entry date, entry price, score and horizon so performance can be checked later against a benchmark.
Selection criteria for stocks, ETFs and AI-upside basket
Stocks Must justify concentration risk through quality, growth/catalyst, valuation discipline, acceptable risk and a thesis that beats a broad ETF alternative.
ETFs Must have a clear portfolio role, diversification, acceptable cost/liquidity proxy, benchmark usefulness and no blocked hidden exposure.
AI upside hedge Tilts to chips, foundries, lithography, networking, cloud, data-center power/cooling, grid and nuclear/power equipment. Higher volatility is expected.
Reject / avoid Blocked exposure, weak score, high risk penalty, poor quality, weak valuation/growth mix, or no reason to prefer it over an ETF.

Investment preferences

Hard exclusions are applied before scoring. The Israel rule removes Israeli country/listing/domicile and material exposure based on the dataset, not name guessing.
filters first
Idle
Main action
Update
Refreshes ranks, prices/news and criteria.
Storage
Local + D1
Watchlists stay local; robot history can be cloud/public.
Benchmark
S&P / Global
Use benchmark comparison before trusting a stock.
Shareable settings link
Share the current filters, horizon, benchmark and selected ticker. Watchlists and saved portfolios are intentionally not embedded; export JSON if you want to move saved data.
The link updates after recommendations run or when you click Refresh link.

Recommended candidates

Click a row to inspect. The table shows a compact selection reason; the selected-asset panel shows the full pass/watch/fail criteria checklist. Tick several assets to add them quickly to the watchlist, save a basket, or back-calculate 3M/6M/1Y returns with stock-by-stock performance attribution.
waiting
0 selected
Select assets with the checkboxes, then click a return window.
Click “Update recommendations”.

Selected asset

No asset selected.
—
Select a stock or ETF after running the analysis.
Selection criteria: run the analysis and select a stock/ETF to see why it passed, what is only watch-level, and what failed.

Score breakdown

The risk component is shown as a negative penalty.

Price history

History range follows the selected horizon.

Recent news / catalyst

seed or live API
No asset selected.

Single-asset simulation

Scenario simulation uses score, asset type, horizon and volatility, then overlays the selected S&P/global benchmark as a reference line. It is not a forecast; it is a sanity-check tool.
asset vs benchmark
Run recommendations, select an asset, then click “Run simulation”.
What the simulation button does
It takes the selected asset's score, risk penalty, asset type, holding horizon, initial investment and monthly addition. It then builds bear/base/bull scenario paths and overlays the selected S&P/global benchmark. This is a stress-test, not a price forecast.

5) Auto portfolio comparison

Compares automatic groups from the current recommendations in the selected base currency.
Use ETFs as benchmark candidates: e.g. a global ETF should be the minimum hurdle for single-stock ideas.

Portfolio laboratory

Create named portfolios from current recommendations or the selected asset. Then test: “what would my result be if I bought this group 60 days ago?” The result shows the portfolio path, S&P benchmark, which stocks helped, which stocks hurt, and each asset contribution. Saved portfolios persist between browser sessions.
saved portfolios
Saved portfolios
0
Selected for test
0
Look-back
60d
Storage
local
Portfolio storage and benchmark logic
Storage: portfolios are saved in this browser under the active profile. Use Export JSON as a backup or to move portfolios to another computer. Benchmark: the selected benchmark is fetched over the same look-back window as the portfolios. SPY is the default S&P 500 proxy; CSPX.L is available as a UCITS alternative.
No saved portfolios yet.
Save at least one portfolio, select it, then click “Back-test selected”. The chart will show each selected portfolio's historical value path and the benchmark over the same look-back period.

Daily paper robot

The robot has two separate actions: first it records today's model-selected basket; later it recalculates the observed performance of every recorded basket against the selected S&P/global benchmark.
paper robot
Robot records
0
Latest basket
—
Latest return
—
Vs benchmark
—
True server robot status: use this panel after deploying green-ai-daily-robot. It checks the Worker itself, not just the browser cache. Default schedule in wrangler.robot.toml: one run per weekday at 22:30 UTC. Manual Run robot now can test the Worker immediately; it records or overwrites today's paper basket for the same strategy/profile.
True robot
Unknown
Cloud records
—
Last server run
—
Next expected run
—
Paste the Worker URL once, then click Check robot ON. A healthy robot should report ready=true, the current strategy, user/profile, cloud record count, and the latest recorded date.
Public robot results: visitors can load the shared robot track record in read-only mode. This shows the online robot's paper baskets and measured performance, but it does not expose or change personal watchlists.
Public robot feed not loaded yet.
Robot button logic
Record today's picks creates or overwrites one paper basket for today's date using today's ranking and entry prices. Recalculate performance does not create a new basket; it updates the observed return of recorded baskets using current prices and the benchmark.
Click “Record today's picks” after recommendations update. The robot stores the selected basket, entry prices, benchmark and horizon. Later use “Recalculate performance” to measure what happened.
No robot entries yet.
Local storage warning
Browser storage survives normal sessions but can be lost if browser data is cleared, if you change device, use private mode, or change browser profile. Use Backup JSON; use D1 cloud mode for durable robot history.
Cloud mode / login-ready: this package includes optional endpoints that can save the robot ledger to Cloudflare D1 under the authenticated user's email. It stays inactive until you configure Cloudflare Access and a D1 binding called DB.
Cloud sync not checked. Local mode is active.
True daily robot: the browser button only records picks when this page is open. For a robot that really runs by itself every trading day, deploy server_robot_worker.js with Cloudflare Cron and bind it to the same D1 database as DB. The Worker writes to the same cloud ledger, so this dashboard can later use Load robot log from cloud to pull the automated records.

AI upside hedge mode: a concentrated paper sleeve for AI infrastructure exposure: chips, foundries, lithography, data-center power/cooling, cloud and grid/nuclear power. It is not a safety guarantee and not an automatic buy instruction.

Reality tracker and user watchlists

Each profile has its own watchlist stored locally in this browser. Click “Update reality” later to compare recommendation vs real price movement.
local profile storage
Multi-user behaviour
Many people can use the same public dashboard link. By default, each person's watchlist is private to their own browser/device. Different profile names create separate local watchlists in the same browser. Cross-device watchlists require login + database storage.
Use the buttons above to test how the current watchlist would have performed over 3M, 6M, 1Y or 2Y. The table will show every stock/ETF, return %, contribution and whether it helped or hurt.
No watchlist entries yet.

Important guardrails

Keep the dashboard useful without pretending it predicts the future.
Not financial advice A high score means “review candidate under this horizon,” not a guaranteed return.
ETF hurdle For long-term money, broad ETFs are the default comparator. Single stocks must justify concentration risk.
Back-tests are retrospective The “60 days ago” tool shows historical behaviour, not future performance.
Storage discipline Watchlists are local unless exported; robot history is durable only when saved to D1/cloud.
Green Investment Dashboard — Powered by AI v0.19. Transparent screener, paper robot and portfolio-reality tracker.