Using Return Distributions to Set Intraday Targets and Stops
Summary
The note presents a simple statistical framework for setting a price target and stop level over a chosen time horizon. Estimate the mean and standard deviation of returns from historical data, select a confidence interval, and map its return bounds onto the current price. Under a Gaussian assumption, the example uses a 95 percent interval to illustrate how a target and downside threshold can be interpreted probabilistically.
The example shows that a target being reached often does not by itself demonstrate forecasting skill: with the chosen mean and dispersion, the target is crossed about half the time, while the lower threshold is framed as a less frequent tail event. A forecast should state its horizon, distributional assumptions, and expected return alongside the price levels. The answer explicitly cautions that a simple Gaussian model may be inadequate; the illustrative inputs are chosen for the example and are not evidence that the method predicts any particular stock’s intraday movement.
Key ideas
- Estimate returns over a chosen horizon before translating statistical bounds into price levels.
- A target and stop can be interpreted as quantiles of an assumed return distribution.
- A frequently reached target may reflect the chosen expected return rather than forecasting accuracy.
- The Gaussian illustration is simplistic and should not be treated as a validated model for intraday prices.
Tags
Full text
# How we decide the target price for stock # How we decide the target price for stock people giving intraday target price of particular share. Most of the times the target is achieved.I am still puzzled how the target price of stock for intraday can calculated. To elaborate my query let me take an example :- Person X says Current market price : 100 target : 105 stop loss : 95. Can someone enlighten us in this direction ## Answer by André Christoffer Andersen (score 1) https://quant.stackexchange.com/a/20921 This really depends on your methods. Earlier today in a different question I talked about confidence intervals using a very simplistic Gaussian model. I could reproduce that example to fit with your example: - Select some time lag for your data. - Calculate the rate of returns for each time step. - Calculate the standard deviation and mean of the rate of returns. Say the standard deviation is $5.1\%$ and the mean is $5.0\%$. - Choose a confidence interval, say, $95\%$. - Use the inverse of the distribution for the interval $2.5\%$ and $97.5\%$ (a width of $95$ percentage points). For the Gaussian distribution this gives z-values of $-1.96$ and $+1.96$. - Calculate the interval, which comes out to be $-5\% = (5\% - 1.96 \times 5.1\%)$ and $+15\% = (5\% + 1.96 \times 5.1\%)$. - If the price of the asset is $\$100$ then your confidence interval would be between $\$95 = \$100 \times (1-0.05)$ and $\$115 = \$100 \times (1+0.15)$ with a confidence of $95\%$, meaning that you typically wouldn't see the price go over or under this interval more often than once every $20$ days $(=1/0.05)$. Notice that this conforms to your statement that "Most of the times the target is achieved", in fact 50% of the time it would've reached the target, but only because I chose a standard deviation and mean which fit your example data. If the analyst did these calculations (with better assumptions I hope) then they could make statements that say something like: > The current market price is \$100. The expected/target return tomorrow is \$105, meaning that half of the time, in circumstances like this, we'd see the price go over this target. That said, one in twenty times we'd expect to see it fall below \$95, thus we set our stop loss there. Small print: Mean return 5%; standard deviation 5.1%; Gaussian model. Normally you'd hope the analyst used something better than a simple Gaussian model.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.