te_fo_index.py — TE-FO Index
Added 2026-10-03. backend/nseeod/te_fo_index.py publishes an index of all NSE F&O stocks as a regular
te_asset / te_price series (code TE-FO, name TE-FO Index), so it can be charted next to every other
instrument. Unlike the research series in ci_* (custom_index.py), it is meant to be used
day to day: the Bhav Downloader extends it after every download.
Method
Base 1000 at the first stock-futures expiry (2001-11-29). At each monthly expiry R(k) (expiry-day close):
| Step | Rule |
|---|---|
| Members | Every stock whose future expiring R(k+1) trades on R(k): the F&O list as it was then (custom_index.cycle_data, same rule as the leverage study). New F&O stocks join at the next expiry; stocks NSE drops leave when their last contract expires. |
| Quantity | q_i = (NAV(R(k)) / N) / F_i(R(k)): the capital is split equally across the N stocks |
| Daily bar | open / high / low / close = Σ q_i × the future's open / high / low / close; NAV = close |
| Roll | The cycle ends at the expiring contracts' close on R(k+1); the next one starts at the new contracts' close with the NAV reached |
| Splits / bonuses | Running futures' earlier prices are scaled by the scr_corp_action factor when the contract's own jump confirms it (custom_index._ca_scaled) |
- No costs and no interest on the margin money. It's a pure price index of the futures portfolio. The leverage study's
FO-EW-FUT-1Xis the variant with 5 bps per side and NIFTY-implied collateral interest. - Open and close are exact. High and low are bounds, the sum of the members' highs and lows, because the stocks do not all top out at the same moment. EOD Chart therefore opens it as a line chart.
- A day a contract does not trade (or has a zero open) takes its last close, so that leg's bar is flat.
Results (local, 2026-10-03)
| Period | 2001-11-29 → 2026-09-25, 6,169 days, 298 expiry cycles |
| NAV | 5,666.91 |
| CAGR | 7.2% |
| Max drawdown | −73.4% |
| Members per cycle | 29 (Nov 2001) → 264; 207 now |
| Splits / bonuses inside cycles | 284 scaled, 0 unconfirmed; 0 cycles skipped |
| OHLC consistency | 0 rows with high < max(open, close) or low > min(open, close) |
The gap to FO-EW-FUT-1X (11.9%) is the collateral interest (≈ 5.9% a year at the NIFTY-implied rate) less that
series' 1.2% a year of costs.
Survivorship-free membership
Membership is kept in the index constituent tracker tables under TE-FO INDEX, so it is
queried exactly like NIFTY 50:
SELECT * FROM idx_members_on('TE-FO Index', '2008-06-30'); -- 227 stocks, incl. AIRDECCAN, ADLABSFILM ...
| Table | TE-FO content |
|---|---|
idx_constituent | Source of truth. One snapshot per expiry: symbol, futures entry price (close_price), weight (weightage_pct = 100 / N), source = 'TE-FO' |
idx_membership | Intervals [from_date, to_date), rebuilt from the snapshots. from_date = the day after the expiry it joined (its returns count from the next session); to_date = the day after the last expiry it was held; NULL = in the running cycle |
idx_change | IN / OUT per expiry (the Nov 2001 launch batch has no IN) |
idx_index | stock_count = members now, history_from = exact_from = 2001-11-30 |
index_members.py build --apply reloads only the NSE indices; it skips everything in index_members.EXTERNAL
(TE-FO INDEX), which used to be wiped by its TRUNCATE.
Daily update
BhavProcessor._update_te_fo_index() (nseeod/downloadbhav.py) runs after the corporate-action sync in both
run_update() and run_update_for_date(). It calls update():
- find the last stored expiry on or before the last stored day;
- recompute from there, starting from the NAV stored on that expiry: the running cycle, plus any cycle that ended since;
- upsert
te_price, replace that expiry's snapshots onward, rebuild intervals and changes; - delete any stored day after the last computed one (a rolled-back bhav day).
The result is identical to a full build: tested by deleting the last 42 days and re-running, which matched to
0.0001 (4-dp rounding), with identical membership. Nothing stored yet → update() runs a full build. Log tag
TE-FO. A force-update of an old date does not rewrite history: run build --apply after back-filling
F&O data.
Usage
Run from backend/:
python -m nseeod.te_fo_index build [--apply] # full recompute (~2 min); --apply replaces every TE-FO row
python -m nseeod.te_fo_index update # incremental (what the Bhav Downloader runs)
python -m nseeod.te_fo_index members --date 2015-06-30
API (app/api/te_index.py, generic over te_asset rows with sub_type = 'custom_index'):
GET /api/te-index/symbols, /chart?code=TE-FO&from_date=&to_date=&timeframe=1D|1W|1M|1Q|1Y,
/symbol-range?code=, /members?code=TE-FO&date=.
EOD Chart: NSE → FUT shows a TE-FO Index chip next to "Available Stocks". Picking it skips the expiry step and fills the date range with the full history.
te_asset row: asset_class = 'equity', sub_type = 'custom_index', value_kind = 'level', unit = 'points'.