We score public share tips the same way for everyone, with the rules fixed in advance. Hypothetical, paper-traded outcomes — entries are simulated at the first market bar after a call was detected; no real capital is at risk. Past performance is no guarantee of future results.
Every call is timestamped the moment it's detected, and that timestamp is immutable. A tip's simulated entry is the open of the first market bar strictly after detection — never the same bar, never a back-dated price. A call can't be credited for a move that had already happened when it was made.
Returns use adjusted closing prices, so splits, dividends and other corporate actions are reinvested into the figure (total return). The frozen, unadjusted entry is kept as evidence.
Each tip is compared against its market benchmark (S&P 500 for US, ASX 200 for AU) over its own entry-to-evaluation window — apples-to-apples, not a single index point-to-point number. The credibility metric is alpha (excess return vs that benchmark), not raw return: a buy "hits" if it beat the market; a sell "hits" if it underperformed it.
Sources are ranked by the lower bound of a 95% confidence interval on their hit rate, and established sources (≥20 settled tips) always rank above provisional ones. A small lucky streak can't top a long track record.
Each tip is bucketed by its stated horizon (short / swing / buy-and-hold) and scored primarily on the window nearest that horizon — 30, 90 or 365 days. Judging a multi-year thesis at 30 days, or a "this pops this week" call at a year, would be meaningless. The default leaderboard view is the 90-day window; the Primary horizon view scores every call on its own clock, and other windows are shown as secondary context.
The conviction-weighted view counts a source's high-conviction calls for more than its throwaway mentions. The confidence interval still uses the raw sample, so weighting reflects emphasis — it can't manufacture statistical certainty.
Delisted, acquired and failed tickers are kept in the record at their last value. Quietly dropping losers is exactly what inflates a published track record; we don't.
When a source states an explicit target level, we record whether and how fast the price first reached it. We publish the days-to-target, never the target level itself.
For each call we also report max drawdown (the worst peak-to-trough fall on its return curve), annualised volatility, and a Sharpe-style ratio of its excess return. A high hit rate with violent drawdowns is a different thing from a steady one, and we show both. These are derived from the same daily valuations and annualised treating each step as a trading day.
This is the current mark of $1,000 spread equally across every tracked call — the equal-weighted average of each call's return from its own entry, expressed as an index off a $1,000 base (not a dollar price). It is a cross-sectional average, not a single compounding portfolio, and it moves as new calls enter and existing ones are remarked. Closed and failed calls stay in the average, so it can't flatter itself by dropping losers.
Returns are currency-neutral ratios, so cross-market comparisons are fair. Any dollar-denominated figure states its currency.
We never publish raw market prices — only derived returns and alpha.