Independent · paper-first

Systematic trading,
built honestly.

nasable is an independent workshop for rules-based research and automation on Indian equities. The systems are tested without flattering themselves, proved on paper before real money, and left to run every day with no one watching.

Explore the projects How the work is done
Nifty 500
the universe the systems trade
Paper-first
no real money until it is proved
Every weekday
runs itself after the close
1,200+ tests
guarding every change
Projects

Two systems, one way of working

Each one is a complete platform, covering data, signals, risk, execution and reporting, and each runs on its own schedule on an always-on server.

NiftyPing

Momentum with a trend overlay

Holds the strongest stocks of the Nifty 500 and steps aside into cash when the market loses its trend. It screens, rebalances, records and reports on its own every trading day.

  • Ranks the universe by medium-term momentum and weights each name by how calm it is
  • A market-wide trend switch, plus a drawdown breaker that flattens the book
  • Daily email and Telegram digests, and a bot that answers status questions
  • A staged path from paper to dry run to live, with a kill switch and a daily loss cap
PythonPostgreSQLKite Connect🔒 private dashboard

AlphaSieve

A point-in-time execution platform

A research strategy turned into something that can run every day without fooling itself. It filters out the edge that was never really there and executes only what survives.

  • Rebuilds the investable universe exactly as it stood on each date, with no lookahead
  • Turns target weights into orders and reconciles every fill and cost to the rupee
  • Executes a fixed, pre-registered strategy, so a parameter change is a code change
  • Reports against the same benchmark and accounting rules the research used
PythonPostgreSQLAlembic🔒 private dashboard
Principles

Rules that come before any result

Most backtests are too kind to themselves. These rules exist to make that hard.

01

No lookahead

A decision uses only what was known at the close, and the fill happens at the next open, never at a price seen afterwards.

02

Point-in-time universe

Companies that later fell out of the index stay in the test. Leaving out the losers is the easiest way to invent returns.

03

Pre-registered holdouts

Settings are chosen without looking at the most recent period, which is then used once, to see if they hold up.

04

Paper before money

Every system builds a forward record on paper and must pass explicit gates before it is allowed near a real account.

05

Fingerprinted results

Published figures are tied to a hash of the settings and of the engine code, so silent drift fails the build.

06

Watched from outside

Every job reports to an independent monitor. A late, failed or missing run raises an alarm without anyone having to look.

How it runs

A trading day, without hand-holding

Everything below happens on a schedule on an always-on server. All times are IST.

08:45
Broker check

The broker session is validated and renewed only if it has expired.

09:25
Protection audit

Every live holding is checked for its safeguards, twice each morning.

15:30
Market closes

Daily bars settle, and the systems wait for final prices.

16:05 – 16:45
The daily runs

Screen, apply the risk rules, rebalance, park orders for the next open, then record everything.

Evening
Reports

The digests go out, the dashboards refresh and the database is backed up.

Notes

Engineering notes

Short write-ups on what building these systems taught, mostly about the mistakes.

read

The returns that were never there

What happened to a backtest once delisted and demoted companies were put back in.

 read

The breaker that could never reset

A drawdown switch whose way back depended on a recovery it had made impossible.

 read

A check that never pinged never alerts

Lessons from moving scheduled jobs off a laptop and onto a server.

 read

Making configuration prove itself

Auditing settings that look like controls but can never actually fire.