Exit settings

The Exits tab sets "the ranges the optimizer samples exit genes within — you set the range, not the value." Every candidate gets its own exit parameters drawn from these ranges; Pin a range to fix that gene across every candidate (an optimized gene becomes a fixed setting).

Target and time stop#

• R-multiple target — the profit target as a multiple of the initial stop distance (reward:risk). Enable it and set Min/Max (default range 0.5–6, step 0.25). Check Pin — fix every candidate to the Min value to lock a single target.
• Max bars in trade — a time stop: force-exit after this many bars in the trade. Default range 5–480 bars, pinnable the same way.

Allowed exit genes#

Three checkboxes control which exit mechanisms the optimizer may use. Unchecking one forbids it on every candidate:

• Trailing stop — let the optimizer use a trailing stop.
• Breakeven move — let it move the stop to breakeven after a trigger.
• Partial scale-outs — let it scale out of part of the position.

Force trailing stop on every candidate goes the other way: every strategy must carry a trailing stop (seeded to the default ATR trail; the optimizer still tunes the trail distance). It overrides the Trailing stop toggle above.

Reversal exits#

Search exit-signal trees lets the optimizer evolve a reversal exit — an indicator condition the strategy checks while in a position, flattening when it fires (e.g. a long exits when momentum turns down) instead of only waiting for the stop or target. Off by default; turning it on widens the search space.

When enabled:

• Exit search complexity — the Min/Max tree depth the optimizer explores for exit conditions. Depth 1 is a single indicator condition; deeper trees combine more conditions with AND/OR. Blank uses the optimizer's global depth.
• Reverse into the opposite side (experimental) — when the exit signal fires, flip straight into the opposite direction instead of just flattening, but only when the opposite entry signal has flipped hard enough:
  • Reverse aggressiveness (flip strength 0–1) — the fraction of the opposite entry's conditions that must hold to reverse instead of flatten. Lower reverses more readily; higher only on a strong flip. Blank = 0.5.
  • Note: reversal is backtest/optimize only for now — the standalone NinjaTrader export flattens instead and warns about it on download.

Practical guidance#

• The default ranges are deliberately wide. If validated strategies keep landing at a range edge (say, every survivor uses the max R target), widen that side — the optimizer is telling you the optimum lies beyond the fence.
• Pinning is how you turn a discovered value into a design decision: run wide once, see where survivors cluster, pin to that neighborhood, and re-run for refinement.