A forecast band and a position size are two different numbers most retail tools never connect. This is the workflow bridge: how a calibrated confidence band width turns into a stop distance, a reward-to-risk ratio, and a bounded risk-per-trade sizing — a process for thinking about size, not a promise that following it makes money.
Plenty of forecasting tools stop at "here is a confidence band" and leave the trader to figure out, on their own, what that band should mean for how much capital to put at risk. That gap is where discipline usually breaks down: a wide band and a narrow band get sized the same way, or sizing is done by gut feel after the forecast is already decided. The fix isn't a better forecast — it's wiring the band width into the sizing math so a wide, uncertain forecast mechanically produces a smaller bet.
Every forecast run produces thousands of simulated price paths, not just a single number. Both legs of the plan are read off what those paths TOUCH, because that is the event an order fills on: the TP1 / TP2 / TP3 ladder sits at the 60th / 75th / 90th percentile of how HIGH each path gets, and the stop at the 25th percentile of how LOW each path gets — or a volatility-based floor (1.5σ scaled to the horizon), whichever sits closer to spot. That stop distance sets the downside used everywhere else in the plan. See the live figures on any ticker page.
From there, reward-to-risk is computed against TP1 (the conservative, "first rung that actually banks profit" reward) over that stop distance — this is the number gated at ≥2:1 per the house risk rule; a display-only "optimistic" R:R against TP2 is shown separately and is never used to gate. Position size then runs through a half-Kelly calculation using that R:R as the payoff ratio and, as the win-probability input, the measured share of simulated paths that reach TP1 before they reach the stop — not model confidence, which answers a different question and answers it far too high. The result is hard-capped so risk per trade never exceeds a fixed % of capital regardless of what Kelly alone would suggest.
Hypothetical numbers only — not a real signal or a specific ticker's current data. Say a forecast puts spot at $100, TP1 (Q60) at $108, and the stop (the closer of the Q25 quantile stop and the 1.5σ floor) at $94. The downside is $6/share, the TP1 upside is $8/share, so R:R against TP1 is 8/6 ≈ 1.33 — below the 2:1 gate, so the sizer refuses this hypothetical outright: it returns a zero size with the failed gate as the stated reason, exactly as the house risk rule requires ("DON'T BUY if R:R < 2:1"). No partial credit for a thin setup.
Now suppose instead the stop sits at $96 (a tighter, higher-confidence band) with the same $108 TP1: downside is $4, upside is $8, R:R is 2:1 — the gate passes. If model confidence for that ticker is, say, 0.65, the half-Kelly position size comes out smaller than a naive "just buy 2x R:R names the same size" approach would size it, because confidence scales the bet down further. A wider band (lower confidence, more distant stop) always produces an equal-or-smaller recommended size than a narrower one with the same reward target — that is the mechanical bridge, and it is the entire point of connecting sizing to the band instead of picking a size by feel.
The practical takeaway is procedural: before sizing anything, look at the stop distance the current band actually implies, compute R:R against the conservative TP1 (not the optimistic TP2), and only then decide a size — capped so a single trade's dollar risk never exceeds a small, fixed slice of capital. If the band is too wide to clear 2:1 even at a reasonable TP, the honest answer is to skip the trade, not to widen the target until the math works. A missed trade is cheap; a bad trade is expensive.
None of this is a claim that following the process is profitable. It is a discipline for turning an honestly-uncertain forecast into a bounded, repeatable decision instead of an ad hoc one — see the live calibration page for how wide these bands actually are in practice, including where they fail.
This is a sizing process, not a backtested money-making strategy: there is no published track record showing that trading this ladder and sizing rule turns a profit over time (see the track record page for what is and is not gated). The Kelly-derived size also inherits every weakness of its inputs — a low-confidence, wide-band forecast produces a small or zero recommended size, which is the point, but a narrow band is not a guarantee the price stays inside it either. And per-trade sizing does not bound portfolio-level risk: several individually-capped positions that share the same sector or factor exposure can still draw down together — correlation risk is yours to manage.
Educational research only — not investment advice.