Price came within 1 pip of the target and reversed: how often should that happen?

[Fairman]

Price came within 1 pip of the target and reversed: how often should that happen?

It feels like it happens constantly. The target is 1.0868, price prints 1.0867, turns round and takes the stop. What does an unbiased walk say about how often that really occurs?

Take a stop S pips away and a target T pips away, one pip steps, no drift. The chance of getting to one pip short of the target before the stop is S divided by (S + T - 1). From there, the chance of finishing at the target before the stop is (S + T - 1) divided by (S + T). That means the share of near misses that fail is just 1 divided by (S + T).

A few shapes:
Stop 6, target 9: 1 in 15 of the trades that get within a pip fail (6.7%).
Stop 10, target 10: 1 in 20 (5%).
Stop 5, target 5: 1 in 10 (10%).

For 6 and 9, about 43% of all trades get within a pip of the target, and only around 3% of all trades do the heartbreak reversal. If the journal says it's more like one in five, randomness isn't the explanation.

There's a mechanical cause too. A short's target is a buy back on the ask, and the chart draws the bid. With a 0.8 pip spread the bid can sit 0.8 pip past your target level on the chart and the order still hasn't filled.

[FXS Prop Desk]

Fairman, this is a nice question to answer with arithmetic rather than feeling. Whatever the exact result for a given stop and target, the near-miss is memorable while the many trades that never came close are not. Counting near-misses in the journal, defined as within one pip of target before stopping, and comparing with what the random walk suggests is the honest way to check whether the market is really hunting targets.

[Amanhasnoname]

I actually counted this after a frustrating week where it felt like every target got missed by a hair. Over about three hundred trades, roughly one in eight of the trades that came within a pip of the target turned round and hit the stop. That was well above what your formula gives for my stop and target, so something other than randomness was going on.

Most of it turned out to be the spread issue you mention at the end. A lot of my trades were shorts, and the chart showed the bid touching my target while my buy order needed the ask to get there. On a pair with a spread of nearly a pip during the Asian session, that difference alone explained most of the near misses.

I now place short targets a little above the level, by the typical spread for that session, and long targets a hair below. My win rate went up by a few percent and the near miss frustration nearly disappeared from the journal.

[Fairman]

Counting near misses properly takes a rule. Define one in advance: the best price reached came within 1 pip of the target, using the ask for shorts and the bid for longs. Then take 80 trades on a 6 and 9 shape. About 34 should get that close, and about 2 or 3 of those should reverse. If your count is 9, that's a message about something other than chance: maybe the target sits right at a level where everyone sells, or the stop is too tight and gets run. If it's 2, the memory of the bad ones was louder than the number. Either way you now have a figure to compare against rather than a feeling.

[Fairman]

Shaving the target by a pip changes how trades feel, not the expectancy. On a 6 stop and 9 target in a driftless market the win rate is 40%. Pull the target to 8 and it's 6 divided by 14, which is 42.9%. Before costs both give zero, and after a 0.8 pip cost both give minus 0.8 per trade, because 0.4 times 9 and 0.4286 times 8 both equal what the losses take away. A tighter target converts a few reversals into wins and shrinks every win by the same pip. If the market has genuine drift in your favour, the farther target keeps more of it.

[Fairman]

Put the spread into the distances and the baseline moves. A short with a 6 pip stop and a 9 pip target, filled on the bid, needs the bid to fall 9.8 pips to fill the ask target and only rise 5.2 pips to trigger the ask stop. The win probability is 5.2 divided by 15, which is 34.7%, not 40%. Take the cost from that: 0.347 times 9 minus 0.653 times 6 gives minus 0.8, which is exactly the spread. Longs mirror this. So those near misses that touched your target on the chart and didn't fill aren't bad luck at all, they're the 0.8 pip gap between the line you see and the price you trade.

[Fairman]

A second table helps with the losers. In an unbiased walk with stop 6 and target 9, what share of losing trades were in profit at some point? The formula for being up k pips and then losing is S times (T - k) divided by ((S + k) times T). For k = 3 that's 6 times 6 divided by 9 times 9, so 44% of losers were up 3 pips or more first. For k = 6 it's 17%, and for k = 8 it's 4.8%. So a loser that showed 3 pips of profit is ordinary, nearly half of them do. Only when most of your losers were up 6 or more is something other than noise at work, such as a target that's too far.