Top picks today

How to read this

• Per day — the predicted target divided by the days comparable setups have typically taken to reach it, before anything is charged for the chance of it working.
• After odds — the same rate multiplied by how often that target has actually been reached. This is the order the list opens in, because it is what decides how much the same money earns over a year; sorting by Per day instead will reorder it, and both columns sort.
• Buy at or below — the level, set by this stock's own average and its own volatility. Paying more than it is a different trade from the one measured here.
• Predicted sell — the target size that has paid the most per day on this stock's own past setups. It is a prediction, and it is cut if the run loses its 50-day trend and stays below it; you are told the day that happens.
• Est. days — how long comparable setups took, not a deadline.
• Recovery in the evidence column — the target is not a channel level but the level comparable falls got back to, on a stock whose fall its peers shared and whose history says such falls resolve. It is bought at the market rather than at a dip level, because the fall is behind it; the stock's own page states what broke, when, and whether the record says it comes back.
• Dividend due — keeps only stocks where buying now at the last price is paid the next dividend (the buy settles one business day later, by the record date, so it has to trade before the ex-date), the ex-date is within 14 days, the dividend is at least the percentage entered of the last price, and the projected channel path expects the price above today's in 14 days.
• Evidence — how many comparable setups on this stock stand behind the target and how often they reached it. Where a stock has too few, the market-wide rule stands in and says so.

A caveat on the very fast rows. Time to target is measured against the highest price each day reached, so a stock whose target was cleared during the session counts as one day. Where the typical time is only a session or two, the rate is real but flattering: it assumes you sold into a spike you would have had to be watching for. Treat anything under about three days as the least reliable number on the page rather than the best buy on it.

Ranking is not a recommendation, and a high rank is not a high certainty: it is the order of expected rates of return given what has happened before, which is a different thing from what will happen next. Every call made is published, hit or miss, on the track record.