Three Prices (3price)

Any information generated by this software (its algorithms) does not constitute an individual investment recommendation. You make financial decisions on your own and at your own risk.

This software is provided "as is", without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose, and noninfringement. In no event shall the authors or copyright holders be liable for any claim, damages, or other liability arising from, out of, or in connection with this software or its use.

This code was created with the help of an LLM (Large Language Model) assistant.

A system for analyzing the prices of financial assets: for each instrument, the system determines which of three price ranges its current price falls into, and generates a PDF report containing a summary table and charts. The configuration file (Config) lists the tickers and descriptions of assets traded on the Moscow Exchange.

The code can be used to analyze any financial asset by changing the ticker and its description. The txt folder must contain the data downloaded for that asset.

🖼️ Report samples

       

The "Three Prices" logic is simple:

1. Closing-price history is taken for the analysis period — from January 1 of the year that began YEARS years ago to the present day.

2. Over this history the price minimum and maximum are computed; the [min … max] range is split into three equal parts by the levels min + ⅓·range and min + ⅔·range.

3. The current price (the latest available date) is classified:

   Condition	Label	Color in the table
   price ≤ ⅓ of the range	низкая ("low")	green
   price ≤ ⅔ of the range	средняя ("medium")	black
   price above ⅔ of the range	высокая ("high")	red

   If the data contains no valid price at all, a — label is set in gray.

The idea is a rough but vivid estimate of how "expensive" an instrument looks relative to its own multi-year history, with no reference to fundamentals.

---

Project layout

Три цены/
├── main.py           # entry point: orchestration only
├── config.py         # all configurable parameters and the asset list
├── log_setup.py      # logging setup and the startup banner
├── data_loader.py    # reads quotes from txt/ → {ticker: DataFrame}
├── analysis.py       # "three prices" levels and color labels (pure functions)
├── report.py         # PDF generation: title page, table, charts
├── txt/              # INPUT data — quote files (required before the first run)
├── logs/             # run logs (created automatically)
└── reports/          # ready PDF reports (created automatically)

The project folder itself is named Три цены ("Three Prices" in Russian).

---

Requirements and running

• Python 3.10+ (uses the str | None, dict[str, ...] syntax).

• Libraries: pandas, matplotlib:

  pip install pandas matplotlib

Run from the project folder:

python3 main.py       

What happens on start:

1. Run parameters are computed (dates, file names) — see config.py.
2. Logging is configured (file + console) and the startup banner is printed.
3. Quotes are read from the txt/ folder and filtered by the start date.
4. Levels and range labels are computed for each ticker.
5. The PDF is written to reports/; the logs/ and reports/ folders are created automatically.

Return codes: 0 — the report was built; 1 — no ticker survived the filtering (the log receives [ERROR] Нет данных для сохранения в PDF… — "No data to save to PDF").

Console output example (the program prints its log messages in Russian; they are shown here as-is):

[INFO] Успешно прочитан файл: SBER-d-5y.txt (354 строк начиная с 2019-01-01)
[ERROR] В файле T-d-5y.txt не найдены колонки даты или цены. Доступные: ['foo', 'bar']
[INFO] Проанализировано: 3 тикеров

================================================
[2026-10-18 14:33:36] 🚀 Запуск системы "Три цены" 3price.
================================================
[2026-10-18 14:33:36] ✅ Готово: 2026-10-18-3price-7.pdf — сохранён в папке reports

(Log lines go to both the file and the console; the order in the example may differ due to stdout buffering when the output is redirected.)

---

Modules

main.py — entry point

Sequence of steps: build_run_params() → setup_logging() → write_banner() → load_data() → build_price_maps() → early exit with code 1 if the data is empty → generate_pdf() → the final "Done" line written to the log and console. The module is safe to import — no code runs on import main.

config.py — configuration

All project constants and the ASSETS_CONFIG asset dictionary. The build_run_params() function assembles RunParams (a named set of run parameters: years, today_str, pdf_filename, log_filename, start_date) and has no other side effects. The period start date is computed as January 1 (current year − YEARS).

log_setup.py — logging

• setup_logging(logs_folder, log_filename) — creates the logs folder, sets up simultaneous output to a file (UTF-8) and the console in the [LEVEL] message format, and returns the path to the log file.
• write_banner(log_path) — the startup banner "🚀 Запуск системы "Три цены"" ("Three Prices system starting"), written to the log without the standard logging prefix.

data_loader.py — data loading

• load_data(txt_folder, assets_config, start_date) — the main function; returns {ticker: DataFrame} with the begin (date) and close (closing price) columns.
• _extract_ticker(file_name) — the ticker from the file name: everything before the -d-5y.txt or -all-d.txt suffix (e.g. SBER-d-5y.txt → SBER).
• _parse_price_file(file_path) — reads a single file:
  • the separator is detected automatically (sep=None, engine='python'), so TAB, comma, and semicolon all work;
  • column names are lowercased and angle brackets are stripped (<DATE> → date);
  • the date column is looked up among the candidates begin, date, timestamp, time; the price column — among close, last, price;
  • dates are parsed with pd.to_datetime, prices with pd.to_numeric; rows with NaN are dropped, the data is sorted by date, prices are rounded to 2 decimal places.

Rules for handling problem files (the system never crashes on a single file):

Situation	Reaction
The ticker is not in ASSETS_CONFIG	the file is silently skipped
Date/price columns not found	[ERROR] … не найдены колонки даты или цены. Доступные: [...] ("date or price columns not found. Available: [...]")
Other read/parse error	[ERROR] Ошибка обработки <файл>: <текст> ("error processing : ")
Empty after the date filter	[WARNING] Файл … не содержит данных после … ("file … contains no data after …")

If both files exist for the same ticker (*-d-5y.txt and *-all-d.txt), both will be read, and the last one read wins — i.e. the *-all-d.txt file (the patterns are iterated in that order). Keep only one file per ticker in txt/ so you don't have to guess which data made it into the report.

analysis.py — levels and labels

Pure functions without I/O, easy to cover with tests:

• compute_levels(df) → (min, min+⅓·range, min+⅔·range, max) — the only levels formula in the project; it is used both by the labels and by the dashed chart lines.
• classify_price(current_price, levels) → (label, color) — the rule from the table above; the ⅓ and ⅔ boundaries belong to the lower/middle range.
• build_price_maps(data_dict) → (price_level_map, price_color_map) — labels and colors for all tickers at once.

report.py — PDF report

• generate_pdf(pdf_path, ...) — the public entry point: a title page, then one page per ticker strictly in ASSETS_CONFIG order.
• _render_title_page(...) — a header with the date and period, the table Тикер / Наименование / Текущая цена / Диапазон цены ("Ticker / Name / Current price / Price range"); the text in the last column is colored green/black/red/gray according to the analysis results.
• _render_ticker_chart(...) — the price chart as a colored line; the min, ⅓, ⅔, max levels are shown as an orange dashed line (--), and the same values are highlighted in orange on the Y axis.
• plural_years(n) — Russian declension of "год/года/лет" ("year/years") with the 11–14 exceptions (one function for the whole project — used on the title page and in chart titles).
• assign_chart_colors(...) — line colors are handed to tickers round-robin from BASE_COLORS in ASSETS_CONFIG order.

---

Configurable parameters

All parameters are collected in config.py (plus two presentation constants in report.py). These are the only things you need to change.

config.py

Parameter	Default	What it controls
YEARS	5	Analysis depth in years: the period "from January 1 (current year − YEARS)". Affects data filtering, the title-page subtitle, chart titles, and the PDF name (…-3price-N.pdf).
TXT_FOLDER	'txt'	Folder with the input quote files.
LOGS_FOLDER	'logs'	Folder for logs (created automatically).
REPORTS_FOLDER	'reports'	Folder for PDF reports (created automatically).
ASSETS_CONFIG	19 tickers	The main asset list — see below.

ASSETS_CONFIG — the asset list

A dictionary of the form:

'SBER': {'name': 'Акции Сбербанка', 'currency': 'RUB'},
'NG':   {'name': 'Фьючерсы на газ', 'currency': 'USD'},

• The key is the ticker, which must exactly match the file-name prefix in txt/ (case matters: Si ≠ SI).
• name — a human-readable name for the table and chart titles.
• currency — the Y-axis label of the chart (Цена (RUB) / Цена (USD) — "Price (RUB/USD)"). Currency conversion is not performed: the NG, BR, GD, SV, PT futures are shown in dollars, as they appear in the data.

How to add an asset: add a line to ASSETS_CONFIG and put a TICKER-d-5y.txt or TICKER-all-d.txt file into txt/. To remove an asset from the report — delete its line from the dictionary (the file in txt/ is then simply ignored).

The order of the dictionary keys is the order of the table rows on the title page and the order of the chart pages.

report.py (presentation constants)

Parameter	What it controls
LEVEL_COLOR = '#ff7f0e'	The orange color of the dashed levels and of the level labels on the Y axis.
BASE_COLORS = [...]	A palette of 10 line colors; with more tickers the colors repeat cyclically.

Computed automatically (not configurable)

• start_date — January 1 of the year YEARS years ago;
• the PDF name — YYYY-MM-DD-3price-N.pdf, by the run date;
• the log name — YYYY-MM-3price.log, by the run month;
• the range-label colors (green/black/red/gray) — hardcoded in analysis.classify_price and _render_title_page.

---

Input data format

Files are copied into the txt/ folder with the names TICKER-d-5y.txt or TICKER-all-d.txt. The format is a text CSV with any separator; two columns are required — the date and the closing price (from the candidates listed in the data_loader.py description). A typical terminal export file:

<DATE> <CLOSE>
2020-01-06 100.00
2020-01-13 101.50
...

Content requirements:

• dates — in any format understood by pandas.to_datetime (2020-01-06, 06.01.2020, 20200106, etc.);
• prices — numbers; rows with empty/non-numeric values are dropped;
• enough data from start_date — everything earlier is filtered out, but if not a single row survives the filter, the ticker will not appear in the report (a [WARNING] is logged);
• data older than start_date in the file does no harm — it is cut off at load time.

Ticker data files can be downloaded using the download_moex program. Then copy them into the txt folder of this project.

Output files

reports/YYYY-MM-DD-3price-N.pdf

• Page 1 — the title: a header with the date, the subtitle "Analysis for N years (from …)", and the table of all tickers with the current price and the colored range label.
• Then one page per ticker (in ASSETS_CONFIG order): the price chart for the period with orange dashed min / ⅓ / ⅔ / max levels.

logs/YYYY-MM-3price.log

The full journal of runs for the month: startup banners, file-reading lines (how many rows were read and from which date), errors and warnings, and the final "✅ Готово" ("Done") line. The log is appended on every run.

---

Troubleshooting

Symptom	Cause and what to do
[ERROR] Нет данных для сохранения в PDF ("No data to save to PDF", return code 1)	There are no files in txt/ with tickers from ASSETS_CONFIG, or the folders are not being created — check that you run from the project folder and that the file names match the tickers.
…не найдены колонки даты или цены. Доступные: ['foo', 'bar'] ("date or price columns not found")	The file has no column with an expected name — rename the column (<DATE>/<CLOSE>, etc.) or add your own candidate to the _parse_price_file list in data_loader.py.
Файл … не содержит данных после ГГГГ-ММ-ДД ("file … contains no data after YYYY-MM-DD")	The file has no quotes for the analysis period — download a longer history.
The ticker is in txt/ but did not appear in the report	The ticker was not added to ASSETS_CONFIG, or the name/case does not match (Si ≠ SI), or the ticker has a competing *-all-d.txt file that overrode the data.
ModuleNotFoundError: pandas	Run with the wrong interpreter: the libraries are installed into a specific Python (on the author's machine — /usr/local/bin/python3.12).