3. Time Series — Account Standardization and Standalone Quarterly Data

As seen in 2. Financial Statements, DART raw data has different account names across companies and quarterly data is cumulative. This tutorial covers DartLab’s time series engine that solves both problems.

• Why account standardization is needed
• Generating time series (timeseries property)
• Understanding the return structure (dict + period list)
• Cumulative → standalone quarterly conversion logic
• Querying specific accounts
• TTM (Trailing Twelve Months)
• Annual time series
• Comparing time series across companies

Setup

import dartlab

Why Account Standardization Is Needed

When fetching “revenue” from DART, each company uses a different name:

Samsung Electronics:  ifrs-full_Revenue      → 수익(매출액)
SK Hynix:            dart_OperatingRevenue  → 영업수익
Hyundai Motor:       ifrs-full_Revenue      → 매출액
Kakao:               dart_Revenue           → 수익(매출액)
KB Financial:        dart_OperatingRevenue  → 영업수익

Same “revenue” but different XBRL IDs and Korean names. Cross-company comparison is impossible in this state.

DartLab solves this with a 7-stage mapping pipeline:

Original account_id, account_nm
  ↓ prefix removal (ifrs-full_, dart_, ifrs_, etc.)
  ↓ ID synonym unification (English variant cleanup)
  ↓ name synonym unification (Korean variant cleanup)
  ↓ core account override (50 highest-priority mappings)
  ↓ learned mapping lookup (34,000+ synonyms)
  ↓ bracket removal + re-lookup (fallback)
  ↓ standardized snakeId returned

As a result, every company’s revenue becomes revenue, operating income becomes operating_income, and total assets becomes total_assets — all accessible with a single key.

Mapping rate across all 2,700 companies is 98.7%, and the balance sheet identity check (assets = liabilities + equity) passes at 99.7%.

Generating Time Series

Use the Company.timeseries property to get quarterly time series.

c = dartlab.Company("005930")
series, periods = c.timeseries

This single line internally:

1. Loads the raw financial parquet
2. Standardizes accounts using 34,000+ synonyms
3. Prioritizes consolidated (CFS), deduplicates row-by-row
4. IS/CF: converts cumulative → standalone quarterly
5. BS: keeps point-in-time balances as-is

Return Structure

series, periods = c.timeseries

series is a dict of dicts by financial statement:

series = {
    "BS": {
        "total_assets": [v1, v2, v3, ...],
        "current_assets": [v1, v2, v3, ...],
        "total_equity": [v1, v2, v3, ...],
        "total_liabilities": [v1, v2, v3, ...],
        ...
    },
    "IS": {
        "revenue": [v1, v2, v3, ...],
        "operating_income": [v1, v2, v3, ...],
        "net_income": [v1, v2, v3, ...],
        ...
    },
    "CF": {
        "operating_cashflow": [v1, v2, v3, ...],
        "investing_cashflow": [v1, v2, v3, ...],
        "financing_cashflow": [v1, v2, v3, ...],
        ...
    }
}

periods is a list of quarterly labels:

periods = ["2016_Q1", "2016_Q2", "2016_Q3", "2016_Q4",
           "2017_Q1", "2017_Q2", ...,
           "2024_Q3", "2024_Q4"]

The value lists in series correspond index-by-index with periods. series["IS"]["revenue"][0] is the revenue for periods[0] (2016_Q1).

Basic Access Pattern

# Samsung recent 4 quarters revenue
series, periods = c.timeseries

for i in range(-4, 0):
    period = periods[i]
    rev = series["IS"]["revenue"][i]
    if rev:
        print(f"{period}: {rev/1e12:,.1f}T KRW")

Checking Available snakeIds

# what accounts exist in the balance sheet
print(list(series["BS"].keys()))

# income statement
print(list(series["IS"].keys()))

# cash flow statement
print(list(series["CF"].keys()))

Key snakeIds:

Cumulative → Standalone Quarterly Conversion

DART quarterly reports are filed in a cumulative structure:

Q1 report:      Q1 revenue = 40B
Semi-annual:    Q1+Q2 revenue = 100B  (Q2 alone = 60B)
Q3 report:      Q1+Q2+Q3 revenue = 220B  (Q3 alone = 120B)
Annual report:  Q1+Q2+Q3+Q4 revenue = 300B  (Q4 alone = 80B)

The time series engine automatically converts cumulative values to standalone quarterly values:

Cumulative                  Standalone
Q1:         40B     →    Q1:   40B  (as-is)
Q1+Q2:     100B     →    Q2:   60B  (100B - 40B)
Q1+Q2+Q3:  220B     →    Q3:  120B  (220B - 100B)
Annual:    300B     →    Q4:   80B  (300B - 220B)

This conversion applies only to Income Statement (IS) and Cash Flow Statement (CF). Balance Sheet (BS) values are point-in-time balances and are not converted.

# already standalone quarterly values
series["IS"]["revenue"]  # [Q1 rev, Q2 rev, Q3 rev, Q4 rev, ...]

# BS is point-in-time balance as-is
series["BS"]["total_assets"]  # [Q1-end balance, Q2-end balance, ...]

Account Query Functions

Helper functions for easily extracting specific values from the time series:

from dartlab.engines.dart.finance import getTTM, getLatest

getLatest — Most Recent Value

series, periods = c.timeseries

# most recent quarter total assets
assets = getLatest(series, "BS", "total_assets")
print(f"Total Assets: {assets/1e12:,.1f}T KRW")

# most recent quarter cash
cash = getLatest(series, "BS", "cash_and_equivalents")

getTTM — Trailing Twelve Months Sum

TTM sums the most recent 4 quarters. It’s the most current approximation of annual performance, with seasonality effects removed.

# annual revenue (sum of last 4 quarters)
rev_ttm = getTTM(series, "IS", "revenue")
print(f"TTM Revenue: {rev_ttm/1e12:,.1f}T KRW")

# annual operating income (sum of last 4 quarters)
oi_ttm = getTTM(series, "IS", "operating_income")
print(f"TTM Operating Income: {oi_ttm/1e12:,.1f}T KRW")

# operating margin
if rev_ttm and oi_ttm:
    margin = oi_ttm / rev_ttm * 100
    print(f"Operating Margin: {margin:.1f}%")

TTM is only meaningful for IS and CF. For BS, use getLatest since values are point-in-time balances.

Annual Time Series

When you need yearly aggregated time series instead of quarterly:

from dartlab.engines.dart.finance import buildAnnual

annual_series, years = buildAnnual("005930")

# years = ["2016", "2017", ..., "2024"]
# annual_series structure is identical to quarterly time series

for i, year in enumerate(years[-5:], len(years)-5):
    rev = annual_series["IS"]["revenue"][i]
    if rev:
        print(f"{year}: Revenue {rev/1e12:,.1f}T KRW")

Comparing Time Series Across Companies

Since accounts are standardized, you can compare any company using the same key.

companies = [
    dartlab.Company("005930"),   # Samsung Electronics
    dartlab.Company("000660"),   # SK Hynix
    dartlab.Company("035420"),   # NAVER
]

for comp in companies:
    s, p = comp.timeseries
    rev = getTTM(s, "IS", "revenue")
    oi = getTTM(s, "IS", "operating_income")
    assets = getLatest(s, "BS", "total_assets")

    rev_str = f"{rev/1e12:,.1f}T" if rev else "N/A"
    oi_str = f"{oi/1e12:,.1f}T" if oi else "N/A"
    asset_str = f"{assets/1e12:,.1f}T" if assets else "N/A"

    print(f"{comp.corpName}: Revenue {rev_str} / OpIncome {oi_str} / Assets {asset_str}")

Side-by-Side Revenue Comparison

samsung = dartlab.Company("005930")
sk = dartlab.Company("000660")

s1, p1 = samsung.timeseries
s2, p2 = sk.timeseries

# compare last 8 quarters revenue
for i in range(-8, 0):
    period = p1[i]
    r1 = s1["IS"]["revenue"][i]
    r2 = s2["IS"]["revenue"][i] if abs(i) <= len(s2["IS"]["revenue"]) else None

    r1_str = f"{r1/1e12:,.1f}T" if r1 else "N/A"
    r2_str = f"{r2/1e12:,.1f}T" if r2 else "N/A"
    print(f"{period}: Samsung {r1_str} / SK Hynix {r2_str}")

Caveats

None Values

Quarters without data are None. Always check for None in calculations.

rev = series["IS"]["revenue"]
# [None, None, 40B, 60B, ...]  ← early quarters may lack data

Banks and Financial Companies

Banks, insurance, and securities firms have different financial statement structures. revenue, current_assets, etc. may not exist. Instead, financial-industry-specific accounts like interest_income, fee_income are present.

Data Start Date

Each company has a different data start date depending on when it was listed. Some start from 2016, others from 2020.

Next Steps

• 4. Ratios — Financial ratios auto-calculated from time series
• 5. Report Data — Dividends, employees, shareholders, audit, and more