Elliott wave analysis from CSV: inspect the rows before interpreting waves
Use a small price-consistency exercise to catch the wrong dataset, a damaged row or an unexplained timestamp.
Published: (UTC+3)
Русский · English · Deutsch · Español · Français · Italiano · 繁體中文 · 日本語 · 한국어
A file that opens successfully is not automatically a sound research input. Check its meaning and a few basic relationships before asking what wave pattern it contains. These checks describe data quality; they do not wave hypothesis that the application silently repairs invalid rows.
Read the columns by meaning
Find the timestamp and the open, high, low and close prices. Keep a record of the timeframe and the source's timestamp convention. A familiar column name is helpful, but check a few values to confirm what the column actually represents.
Do not replace missing prices with a preferred number. If a transformation is necessary, preserve the original file and document the transformation in a separate working copy.
Work through two invented rows
Row A has open 100, high 104, low 98 and close 102. Its high is at least as large as the open and close, and its low is no larger than either. It passes these elementary price checks.
Row B has open 100, high 99, low 98 and close 102. Its reported high is below both the open and the close. Investigate the column mapping or source row before using it. Passing the checks in A would still not prove a wave pattern.
Inspect time without inventing history
Check for duplicate timestamps, unexpected ordering and a gap in the period you intended. A gap may come from market hours, a provider issue or an incomplete export; identify the cause before modifying the file.
Adding synthetic price bars to fill the gap changes the price series. Unless your research explicitly studies such a transformation, obtain the appropriate source data instead.
Write the data acceptance result
Keep a short record: correct dataset, expected coverage, timestamp convention, elementary price checks and unresolved issues. Then select the calculation range as a separate step.
If you cannot establish a crucial input field, leave the dataset unaccepted for that task. A clean-looking chart cannot recover missing source values or prove what a corrupted row originally contained.
Row B has high 99 and close 102. Can you fix it by replacing the high with 102?
That would be an invented repair. Check the source or column mapping, preserve the original and use a verified correction only when its origin is established.