You know the moment. The scanner grades a setup A+. The structure is textbook — tight base, volume dried up, clean pivot. Then you pull up the chart and your gut says the same thing it always says: this thing has already run. I'm late.

So you pass. It feels responsible. Every piece of trading advice you have ever read agrees with you — don't chase, wait for the pullback, buy near support.

We did that for months. Then we built a tool that measures what actually happened on every trade we closed, ran it across the whole book, and found that the trades we felt worst about buying were the ones that worked — and the "sensible" entries near the average were close to a coin flip.

This post is the whole dataset, the finding, the reason we're not fully convinced by our own finding, and the entry checklist we now use.

Universal Advice, Personal Evidence

"Don't chase extended stocks" is one of the most repeated rules in retail trading. It is also almost never tested against the person repeating it.

The rule exists for a reason — a stock a long way above its recent average price has less cushion underneath it, and a small piece of bad news drops it further. That is true. What is not established is whether avoiding those setups actually improves your results, given your rubric, your holding period, and your exits.

That is an empirical question about your own book. Most traders never answer it, because answering it requires records that go beyond entry, exit, and profit or loss.

What We Measured On Every Trade

When a trade closes, our system automatically pulls the stock's price history around that trade and answers a fixed set of questions. No opinions, no memory, no story written after the fact:

Every closed trade carries that record. That is what makes the rest of this post possible.

The Book: 24 Closed Trades

Winners are roughly one and three-quarter times the size of losers, and there are more of them. Across all 24 trades at equal position size that is a net gain of about 80 percentage points. Those trades were logged in advance, published live, and none were removed after the fact.

Here is every one of them.

Sorted by return. ARW and LNTH each appear twice — the same stock traded on two separate occasions, once profitably and once not.

The Finding We Did Not Expect

We split the book by what was true at the moment of purchase. Two conditions stood out.

"Extended" means we bought the stock more than 15% above its own 50-day average price — the setups that look like you've missed it. "Already ran" means the stock had climbed more than 40% from its three-month low before we bought.

Eight trades were bought at prices that felt too high. Eight of them worked. Meanwhile the sixteen trades bought at reasonable, un-stretched prices produced a 44% win rate — worse than a coin flip.

The pattern shows up in the losses too. Several of the worst trades were bought close to the average, exactly where the advice says to buy: DCOM at 4.0% above its 50-day, SBLK at 4.5%, GVA at 6.8%. All three lost.

Why We Don't Fully Believe Our Own Result

Here is the part most track-record posts leave out.

Six of those eight extended winners closed in May. And May was our strongest month by a distance:

So there are two explanations for the same numbers, and 24 trades is nowhere near enough to separate them:

1. Buying strength works. Extended entries genuinely outperform in our system.
2. May was a strong market. Extended setups appear more often in a rising market, so the finding is really about the calendar.

Both stories fit. Eight trades is a small sample, and a single good stretch can produce a run of eight with no losses purely by chance.

What we can say is narrower, and still useful: across 24 trades there is no evidence at all that buying extended hurt us, and there is one un-argued counterexample against it in our own book — zero losses. That is enough to stop treating "it looks extended" as a reason to pass. It is not enough to treat it as a reason to buy.

We will publish this again at 50 trades and at 100. If the pattern breaks, we will say so in the same place.

What Losses Look Like — And What That Cannot Tell You

Our losses share a shape. Seven of nine were gradual declines, not dramatic collapses. Only one had unusually heavy selling volume. Only one happened on a broadly down day for the market — meaning eight of nine losses were specific to that company, not the tide going out.

Two were overnight gaps: the stock opened well below the previous close, straight past the exit price we had set, before anyone could act. No stop loss prevents that. Overnight risk is managed by position size, not by where you place the exit.

This is the trap in most trading post-mortems. Study only your losses and every trait of a loss starts to look like a warning sign. The traits have to be measured on the winners too, or you learn nothing.

Reading The Chart Before The Numbers Can Help You

If you are earlier than that — still learning to name what you are looking at — the Odd Lot chart pattern flash cards cover 21 chart patterns, 32 candlestick patterns and 12 trading strategies across 65 physical cards. Previously sold under the LFG Trading brand. Same product, new home.

Lessons Learned: The Entry Checklist

This is the part we actually use. Every line comes from the 24 trades above — nothing here is borrowed from a book.

Knowing The Patterns Was Never The Problem

Not one of the nine losses above happened because we failed to recognise a pattern. They happened on setups that graded well, in stocks that were genuinely in uptrends, bought at prices a rulebook would defend.

What separated the book was structure: an exit price decided before the buy, a position size that survived an overnight gap, and a record complete enough to check our beliefs against instead of arguing from memory. The reason we could write this post at all is that every trade was logged in advance and measured the same way afterwards.

That is the whole idea. You don't get signals. You get structure.

You can watch it happen in real time. Our live paper trades page shows every open and closed position with its original thesis and its post-mortem attached, and the free stock scanner shows the same grading that produced these entries. When you are ready to work out what you can actually risk on a trade, the position size calculator does that math.

Questions People Ask

These are documented paper trades used for education and process testing. Nothing here is a recommendation to buy or sell any security. Patterns and grades are probabilities, not promises, and past results do not predict future results. Position sizing and risk management are your responsibility.