The target price is the middle of the model’s forecast — the mean of the distribution of outcomes it simulates for that horizon. It is a centre of gravity, not a promise, not a deadline, and not a price the stock is expected to “hit.” The confidence band printed underneath it tells you more than the target itself does.
The model does not produce one price. It simulates a whole distribution of ways the stock could end up at the horizon, and the target price is the mean of that distribution — its centre of gravity. That is not the same as a midpoint: the outcomes are not spread symmetrically around it, so it is not true that half of them land above and half below. The spread around it is summarised by the 90% confidence band (q05 to q95) shown directly beneath it on the ticker page.
That makes it a very different animal from an analyst price target, which is a human analyst’s published opinion about where a stock should trade — usually over the next 12 months, often anchored to peers and revised slowly. Ours is the centre of a modelled range; theirs is a judgement call. Both are shown on the ticker page, and they routinely disagree.
This is the single most useful habit to build, so here it is as a worked example (illustrative). Stock A trades at $100 with a target of $110 and a 90% band of $105–$115. Stock B also trades at $100 with a target of $110, but its 90% band is $70–$150. The target is identical — $110, a 10% expected gain — yet these are not remotely the same trade. For Stock A the model is saying "I am fairly sure this drifts up modestly." For Stock B it is saying "I genuinely do not know; it could halve or it could rocket, and $110 is just the midpoint of my ignorance."
If you read only the target, both stocks look like the same 10% opportunity. The band is what tells you that Stock B could hand you a 30% loss well inside its own forecast. That width — not the target — is what should drive whether you take the trade at all, and how small the position should be.
These are two different claims, and conflating them is the classic mistake. On the committed TOP-20 backtest the 90% band covered 90.63% of realized prices at 3 months, 90.91% at 6 months and 90.20% at 1 year — all three within a point of the 90% they promise, so the band is honestly sized. That is a statement about the RANGE.
It says nothing whatsoever about the target price being reached. The clearest proof sits inside our own backtest: over that window AAPL’s 3-month 90% band covered 85.7% of realized prices — a reasonably sized band — while its terminal forecast mean of $318.65 landed against a realized $253.79, so the target itself was off by about 26%. A sanely-sized band and a badly-placed point, in the same name, in the same artifact. (Figures from the committed backtest window ending 2026-03-31, not a live reading.)
The aggregate also hides wide per-ticker dispersion: a structural-break name like AVGO had its 1-year band cover only 36.5% of realized prices, and MA and V covered just 55.6% and 57.1% at 3 months. Check the per-ticker forecast reliability label before leaning on any single target.
Live example: AAPL last traded around $333.74, and the model's current 3-month target price is $354.18 — sitting inside a 90% confidence band of $307.69 – $400.22. Read those two together, never the target alone: the band is the model telling you how much it actually knows. See the full AAPL forecast for the current figures.
It is not an exit plan. The take-profit ladder (TP1/TP2/TP3), shown separately, is the actual exit plan — it is built from your entry and stop as R-multiples, not from the target. Do not put a sell order at the target price because it looks like a goal.
It is not a deadline either. Nothing in the forecast says the stock reaches the target ON the horizon date; the horizon is simply where the distribution was measured.
And it is not a floor or a ceiling. Prices land outside the band too. The label says 90%, but the honest number is whatever the backtest measured — and for a name going through a structural break it can be far worse than the label suggests (AVGO’s realized price sat outside its 1-year band most of the time). The band is a range, not a fence.
No. The target is the mean of the distribution of outcomes the model simulates for that horizon — a centre of gravity for a range of outcomes, not a prediction that the stock trades at exactly that level. The realized price usually lands somewhere else inside the confidence band, and the forecast is not thereby "wrong."
No — the target is not an exit instruction. The R-multiple take-profit ladder (TP1/TP2/TP3), shown separately on the ticker page, is the actual exit plan; it is sized from your entry and stop, not from the target.
An analyst price target is a human analyst’s published opinion, usually a 12-month view, often anchored to peers and revised slowly. The target price here is the mean of the distribution of outcomes the model simulates for that horizon. They answer different questions and routinely disagree; both are shown on the ticker page.
Because the band tells you how much the model actually knows. Two stocks with the same $110 target are completely different trades if one’s 90% band is $105–$115 and the other’s is $70–$150 — the second could hand you a large loss well inside its own forecast. Band width should drive position size; the target alone should not.
AAPL analysis shows this metric in context, or browse all S&P 500 tickers.
Educational research only — not investment advice.