SI/SO re-plan model

A short entered at a swing high. Drag the handles on the chart. Sz uses the app’s live sizer — calibration offset 0.02, λ 0.10, clamped 0.3–2%.

Second trade — the re-plan at TMP1

Scale in / scale out (% of the original position size)

Edge, original

Edge at TMP1 (PE has changed)

Curve & arithmetic notes (click to open / fold)

Every SL-and-size pair that meets the recommendation

Sz held = Sz₀ × size% × (SL₁p ÷ SL₀p)

Detailed Explanation

(click to open / fold)

These steps read the model live. Move anything and the numbers follow.

This is the original trade only. A short at the swing high. Two handles: SL orig and TMP1.
SL orig sets the risk distance — how many pips you give up if the trade is wrong. It is at now. TMP1 sets the reward distance, at .
Those two make RR (Risk : Reward) — currently .
RR fixes the base rate. BR = 1 ÷ (1 + RR) = . That is the win rate needed just to break even at this RR. Widen the target and BR falls — you need to win less often, because each win pays more. (But probability says you will lose more often trying to reach it!)
You estimate your edge from investigating the context. You add it to the base rate: BR + edge = Pstr — the estimated chance the trade reaches TMP1, now the edge is factored in. The red line at the SL carries the rest, — the estimated chance of hitting the SL instead. EP/BR = Pstr ÷ BR = : how many times better than the break-even chance your estimate is. Above 1.00 = positive expectancy.
Sz falls out of that — . Kelly on Pstr and RR, times λ, clamped to 0.3–2% (Bruce’s personal janky formula). It runs on EP/BR and the Pstr to TMP: better odds → place more, worse odds → place less. Sizing dynamically like this is supposed to earn a better equity curve than a fixed 1% every time — checked by simulating trades with this formula against a fixed 1% entry on the same trade body (the formula is being tuned in the sizing Lab).
Price is now at TMP1. The blue dashed vertical line is the present moment. Left of it has happened; right of it is plan. The readout shows Open P&L = — profit on the open position, not banked, and not the same thing as risk.
Now turn the second trade on — click here to tick it. That is the decision at TMP1.

What each lever moves

Lever	What you hold	What size is recommended
SL new	Yes	Yes, through RR
TMP2	No	Yes
SI / SO	Yes	No

SL new is the only lever that moves both sides. Move the stop and you have already changed your risk without trading.

The arithmetic

Base is equity, not the starting balance. Equity does not change when you scale — a partial close moves money from unrealised to realised and nothing else. Balance does change, which makes a balance-based target circular. (This page drops the tiny bal ÷ eq refinement — under 2% of any figure here — so that every displayed number adds up exactly.)

S₁ = Sz₁ × equity ÷ (distance from here to the new SL)

The entry price cancels out. Only the here-to-stop distance converts risk money into size. Anything anchored to the entry divides by zero at breakeven and flips sign beyond it.

“Close enough to pay the difference” is wrong everywhere except a stop sitting exactly at breakeven. Closing a partial funds nothing; it changes exposure, and exposure is priced at the distance to the new stop.

Pstr is base rate + edge, nothing else. BR = 1 ÷ (1 + RR). The red line shows the other side, 100 − Pstr. The sizer adds its calibration offset (0.02, from live_sizer.json) inside the Kelly step only — that correction is about sizing, not about the forecast, so it never shows in Pstr.