Research tool · not investment advice. Full disclaimer →
New here? How to read
Every number on this page names its source — start with “How to read”.
Known limits
what this site does not do — the unflattering half, firstBottom lineThe limits of the conclusions, stated in writing — where our own research came out negative, where a number is thinner than it looks, and where something describes rather than predicts. Every item below is quoted from our own record, with its source named. The coverage page states the limits of the data; this states the limits of what we claim from it.
📊 Coverage & settlement ledger → — what data we hold, since when, and how fresh. 📖 Glossary → — every metric, defined.
Most strategies here do not survive real trading costs
The strategy pages report backtested behaviour. The recorded champion blend's headline bar is a flat-cost number that has never been re-run under realistic participation-rate costs, so its fundable status is unproven — and the whole leaderboard's ranking rests on that bar. One family has survived a full real-cost test. Treat the rest as descriptions of the past, not as books you could have run.
Recorded in: our own cost post-mortem, and the strategy ledger it corrected
The one factor that survived our own study is beta, not skill
A factor study across this estate's data left momentum as the only survivor. Corrected for a point-in-time data leak, its residual alpha is +6.69% at t=1.80 — which fails the t≥3 bar this estate set for calling something selection skill. Fixing the leak made that conclusion clearer, not softer. Momentum here is exposure, not edge.
Recorded in: our factor study, re-run after we found and fixed a data leak in it
What this site calls a risk-adjusted ratio is not a Sharpe
The engine computes
mean/sd × √periods and no risk-free rate is subtracted. That makes it a return-over-volatility ratio and not a Sharpe. Comparing two of our own numbers to each other is still valid. Comparing one of them to a published Sharpe is not — ours reads high against it, by construction.Recorded in: our own recorded decision about how this ratio may be labelled
Some columns are far thinner than the screen implies
Coverage is not uniform across metrics.
ROE %, for example, resolves for 99 of 2,073 symbols (5%) — it is a financing-template line item — so a filter combining it with two other conditions can return zero rows for a reason that has nothing to do with the market. Where a query returns nothing, check the coverage page before concluding that nothing qualified.Recorded in: an unfixed defect in our own register — see the coverage ledger
Several surfaces describe; they do not rank, score or recommend
Accumulation/distribution, the Wolfe structure work, the credibility index and the rotation phases are descriptive by ruling — they passed as descriptions and failed as rankers, so they are not allowed to order a list or imply an action. Where a page says descriptive, that is a recorded verdict about the evidence, not modesty.
Recorded in: three recorded verdicts on our own methods, each a failure to rank
One fundamentals dependency is not a primary source, and is being removed
This estate's rule is primary sources only — exchange and regulator filings. One inherited exception remains: part of the fundamentals history originates from a third-party aggregator rather than from XBRL filings. It is disclosed where it is used, it is not being extended, and it is being migrated to BSE/NSE filings.
Recorded in: our primary-sources rule, and the one exception it is tracking
The universe has edges, and they are hard edges
NSE-listed Indian equities on the EQ/BE/BZ series. SME is excluded. Nothing here covers unlisted companies, bonds, commodities or non-Indian markets, and no page silently widens its universe to make a count look better.
Recorded in: the coverage ledger, where every boundary is stated
An event edge being real does not make it tradeable
The post-earnings drift study replicates as a genuine effect on this data. Every tradeable wrapper built around it has come out net-negative after costs. It is published as a description of the tape and explicitly not as a strategy.
Recorded in: our event-study ledger, including the wrappers that failed
Nothing on Patearn is investment advice, and Patearn is not a SEBI-registered research analyst or investment adviser. Everything here is a description of publicly filed and publicly traded data. Any decision you take is yours.
This page is maintained as findings change. An item is removed only when the finding that produced it is overturned on the record — never because it reads badly.