Research overview

Analysis starts from trades you already have. Research is where trades come from: you describe an options strategy, and VEEMAN runs it on real historical option chains, minute by minute, to tell you what it would have done.

Research is built around three modules that form one loop:

The usual path is design → relive → validate: you sketch a structure in Define, open it in Rewind to watch how a single day would have played out, then send it to Backtest to see whether it holds over years. Every arrow also goes backwards: any trade in a backtest's log opens in Rewind, and a saved backtest becomes a strategy in Analysis with one click.

Next to the loop, Indicators holds your own indicators written in Pine Script: it shows exactly what runs in Veeman, previews them on the same data and checks them against your own chart, and a backtest can use them to decide its entries and exits. Read more.

What Research is, and what it is not

Research is a simulation and study tool. Like the rest of VEEMAN it never connects to a broker, never places orders and never sends alerts outside the app. Every number is computed on historical data, and every backtest is a statement about the past: what a precise set of rules would have produced on the days in the dataset.

The engine is deliberately conservative where backtests usually flatter themselves: fills are never at the pure mid unless you switch slippage off, every exit pays its slippage and commissions, no filter ever looks at data that wasn't available at the moment of the decision, and the parameter study treats every configuration you try as one more chance to fool yourself. Targets and stops are triggered by the market itself: with the default slippage model the legs are marked at their mid rounded to 0.05, and the slippage is then charged on the exit fill. See How the engine trades and Parameter study.

The universe: SPX, same-day expiration

Today Research works on one underlying and one expiration:

The underlying and expiration selectors appear in all three modules, so new underlyings and expirations will plug into the same screens. When you switch underlying, legs that use fixed strikes of the old underlying are removed (you are asked to confirm first); legs defined by a criterion (delta, % OTM, premium, offset) stay.

The dataset

Research reads a historical archive stored on the server that runs it. Its contents:

Conventions that hold everywhere in Research:

• Times are US Eastern (ET), the exchange's wall clock. The regular session runs 09:30–16:00; half-days close at 13:00, and the engine never enters after a half-day's close.
• Implied volatility is the IV of the mid (the average of bid and ask IVs), not the IV of the last trade, which goes stale on far strikes.
• IV and greeks are shown on the real time to expiry. The data provider computes them as if at least one hour were left, and always up to 16:00, half-days included. Define and Rewind convert them exactly to the real time to the settlement (16:00, or 13:00 on half-days): prices don't change, but in the last hour the IV, theta and vega read like the market's. At the settlement itself the option has expired and has no IV. The engine trades on bid and ask, so backtest results don't depend on this.
• mid exists only where there is an ask: a contract with no offer has no market.
• Greeks are stored in Black-Scholes units and converted for display: theta per day, vega per point of IV.

Each saved backtest records the dataset version it ran on, so a result stays reproducible and Rewind warns you when the data has changed since.

The economic event calendar

Next to the market data, the server keeps a calendar of the releases that move the market, used by the economic event rules of the backtest.

Release history starts in 2013. Every release records the date it actually came out: the releases moved by the government shutdowns of 2013, 2018–2019 and 2025 sit on their real dates, and those that were never published are absent.

The calendar stays up to date by itself. Every weekday morning, before the day's market data, the server reads the official calendars again: new dates (the next year's schedule as soon as it is published) and rescheduled releases enter the calendar. The past doesn't change silently: dates older than about six weeks stay as they are, and a source that doesn't answer, shows no dates or reads oddly (far fewer dates than were known, or more than a year can hold) leaves the previous dates in place for its events. An event whose source hasn't updated for more than a week says so in the rule's window. When a source comes back after a long outage, the releases it missed in the meantime are added; until then the calendar declares its dates only up to the last one it read, so those days are reported as not checked instead of passing for event-free.

From Research to Analysis

A backtest produces the same object Analysis is built on: a daily P&L series. That makes the bridge exact:

1. Save the backtest (it goes to the Research archive).
2. It appears at once in the Analysis library as a strategy with a Backtest badge, and it can be added to any session, combined with other strategies, stress-tested with Monte Carlo, sized, compared.
3. Save as strategy makes a permanent copy in the library, carrying the recipe and dataset version as provenance; from the library, Reopen in Research re-runs that recipe in the console.

The detailed metrics tab of a backtest is computed by the very engine of the Metrics module, on the same daily series: the numbers you read in Research are, to the cent, those you will read in Analysis.

Pages in this section

• Define: the structure studio.
• Composing a backtest: every field of the recipe.
• How the engine trades: the simulation, step by step.
• Results & archive: reading a run, comparing runs, saving.
• Parameter study: optimizing without fooling yourself.
• Rewind: the minute-by-minute replay.
• Tactics: mechanical adjustments.
• Indicators: your own Pine Script indicators.