An analogue outcome measures movement after the matched segment. Specify the starting price, subsequent horizon and result definition.
A higher final price does not imply uninterrupted ascent. Endpoint and path answer different questions.
Outcome
- Rise: 2/4; Fall: 2/4
- Mean: +1
- Hypothetical units
Fix the horizon
Ten hourly and ten daily bars are different intervals. Session bar counts need not equal calendar time. Do not end each case at its most convenient high: that selects favourable moments instead of comparable outcomes.
Two paths, one endpoint
Both hypothetical cases start at 100 and end at 102: +2%. One rises gradually; another first falls to 85. Their identical endpoint conceals different experience.
A minimum relative to the start and maximum drawdown from an intervening peak are also different measures.
Same endpoint, different path
- A
- B
- Outcome: +6%
- The endpoint does not reveal drawdown along the way.
Combine descriptive measures
Final change describes the endpoint. High and low relative to the start describe excursions. Mean and median summarize the centre; dispersion and extremes describe heterogeneity. Count and composition describe how much distinct evidence is present.
The mean is sensitive to large observations. Median is the middle ordered result, not the entire distribution. Do not select only whichever statistic looks stronger.
Positive frequency is insufficient
Four cases end at +1% and one at −8%. Positive frequency is 80%, yet mean change is (4−8)/5 = −0.8%.
These are hypothetical descriptive outcomes, not strategy returns: trade rules and costs remain unspecified.
Reliability depends on selection
There is no universal minimum analogue count. Dependence, selection and regime diversity matter. One case strongly affects a small set; many overlapping cases may still leave substantial uncertainty.
Different settings may compare different populations or horizons, not one unchanged quantity. The summary describes the past; see the distinction from predictive probability.
Connect the window, sample and hypothesis test
Before comparing outcomes, choose the reference window: its boundaries determine which events belong to the matched segment. Then consider the size and limitations of the analogue sample before interpreting the share of positive outcomes.
Moving from a description of the past to testing a trading hypothesis requires separate rules and independent data. The review of technical-analysis research explains what published studies tested and where their conclusions stop.