The AI for financial research
Ask Itoflow to research an idea, test a portfolio, or monitor a strategy.
Strategies that stay under continuous review
Itoflow keeps each thesis in memory, watches the market as it changes, refreshes the evidence, and surfaces the next action for review.
Rank 01
Opportunistic Long
Opportunistic equity
Conviction-based dip-buying with dynamic cash allocation
Conviction dip-buying with cash deployment gates.
Return
+6.2%
Action
HOLD
Sharpe
0.50
Holdings
HOLD / 2026-08-05
I am holding the current portfolio because exploration's strategic target is the current book: no active trades, cash is above the 5% floor, and changing now would...
Rank 02
Global Ex-Fossil Benchmark-Aware Equity
Screened global equity
Global developed-markets screened ETF strategy anchored to MSCI World with moderate tilts and broad fossil-fuel exclusions.
Developed-market equity exposure with fossil-fuel exclusions.
Return
+4.3%
Action
HOLD
Sharpe
1.98
Holdings
HOLD / 2026-07-29
I am holding the current allocation because the portfolio is already close to the tested 73% ESGV / 27% VSGX screened-equity anchor and there are no active trades w...
Rank 03
India Strategic Growth
India structural growth
Macro-thematic India strategy focused on defense, infrastructure, commodities, and precious metals
Defense, infrastructure, resources, and precious metals.
Return
+4.0%
Action
HOLD
Sharpe
0.34
Holdings
HOLD / 2026-08-05
I am holding the current India Strategic Growth basket because the tested strategic target is essentially the current material book, active trades are zero, cash is...
Itoflow leads BuySideBench
Best median score across 14 interactive portfolio-management assignments, where systems must recognize and request missing investor constraints before acting.
0.689 median score, +0.340 over the strongest external system.
| # | Configuration | Score |
|---|---|---|
| 01 | ItoflowGPT-5.6-Luna · XHigh reasoning | 0.689 |
| 02 | CodexGPT-5.6-Sol · High reasoning | 0.349 |
| 03 | Claude CodeOpus 5 · High reasoning | 0.306 |
From the research blog
Notes from the lab — how the agent works, what it gets right, and where it still needs a human.