Recognition without negative examples is overfitting. Each pair on this page shows a setup that worked next to a setup that looked nearly identical and didn't — with the one feature that discriminated them called out. The point is not to memorize 4 patterns; the point is to make your eye go to the discriminator first.
How to use this page: for each pair, look at the two charts before reading the labels or the discriminator. Decide which would have worked. Then check. The first ~10 times you do this for any pair, you'll get it wrong on the harder discriminators (#3 and #4 below). That's the point — the calibration only happens through the misses.
Pair 1 · Origin trading
Two "polarized 1st-touch" entries — one was the conversion, one was the retest
Both charts show a break level that fails and a strong reversal candle. In one, longing the reversal candle worked. In the other, it was the trap from WT-5. The difference is one specific feature — find it.
Worked · +2.4R
The level was already an origin from a prior failed break (printed days earlier). This was the 2nd touch — the entry candle was a retest, not the conversion.
Trap · -1R (WT-5)
The big blue reversal candle itself printed the origin — the level was just being born. This was the conversion candle, not a retest. Longing it = trading the wrong side of the move.
→ Discriminating feature
Was the level an origin BEFORE this candle, or DURING this candle? The "1st touch" the method describes only exists if the origin was already polarized by a prior failed break. If this candle is the failure, you're inside the conversion event — wait for the retest.
Pair 2 · Cycle reversal
Two "trend break" reversal trades — one was array #5, one was array #3
Both charts show a topside trend break in a long uptrend. One was the cycle pivot. The other was WT-6's miscount.
Worked · +4.2R
Five completed arrays before the entry. The next continuation array failed to print — a continuation candle that should have extended the trend printed against it instead. Short on confirmed array failure.
Trap · -1R (WT-6)
Only three arrays counted before the short. The cycle continued to print arrays 4 and 5 — the actual reversal arrived later, at array 5's blow-off. The trader was short the parabolic acceleration phase.
→ Discriminating feature
Did a continuation candle FAIL to print after the array, or are you front-running the next array? A reversal entry requires a confirmed failure of the next would-be continuation. Without it, you're shorting parabolic continuation — the worst R:R in the cycle.
Pair 3 · Time & Levels
Two 1H "hard close" breakouts — one was real, one was a 4H wick
Both charts show a 1H accumulation candle body-clear of a level. One was a real breakout. The other was WT-7's trap.
Worked · +1.8R
The 1H signal was a real breakout — but the trader waited for the 4H candle to close before entering. The 4H body confirmed the break. Then long the retest.
Trap · -1R (WT-7)
The trader entered on the 1H confirmation before the 4H candle closed. When the 4H closed, body was back inside the range — the 1H "close above" was, on the 4H, a wick.
→ Discriminating feature
Did the level's home-TF candle close, or are you reading a sub-candle on a smaller TF? Time & Levels: a 4H level needs a 4H body close. A 1H body close inside a 4H candle is, by definition, a 4H wick.
Pair 4 · Averaging in
Two "averaged short" trades — one closed at +3.5R, one stopped at -3.8R
Both charts show a short with three averages-up. One was the WT-4 winner. The other was WT-8's blow-up. Same plan, different chart.
Worked · +3.5R (WT-4)
HTF stayed bearish through all three adds. The third add was at the range top — exactly where the rejection started. Each add improved the average entry; the rejection took the whole stack into profit.
Trap · -3.8R (WT-8)
HTF flipped bullish around add 1. Adds 2 and 3 were against the new HTF cycle — adding to a position whose thesis was already broken. By the time the stop hit, the stack was at -3.8R because the later adds had no room.
→ Discriminating feature
Was the original thesis still valid at the time of each add, or was the trader averaging into a winning trade against them? "At structure" is necessary but not sufficient. Each add must also pass: HTF still supports the thesis, AND total stack risk is still under cap.
Working drill: come back to this page after every losing trade and find the pair that matches your loss. If none of them match, write a 1-paragraph description of the discriminating feature you missed and add it as a personal note in the journal — that's how the calibration set grows from 4 pairs to your own bigger one. The pairs above are the canonical four. Yours will be more specific.