For fund managers and wealth desks

Numbers you can
put your name on.

An advisory desk on one rail, built so that no figure it shows you is better than the data behind it.

Talk to usSee pricing

Any platform can show you a clean number. The question worth asking a vendor is what happens when the underlying data is incomplete, stale, or simply has not arrived.

Here is what we do instead.

Measured is never blended with modelled

Cost drag comes from settled fills. Rows that carry no charge breakdown fall back to a model and are counted separately, so a part-measured number is never presented to you as measured.

Excluded rows are counted, not dropped

Benchmark comparison prices the counterfactual for every position: the same rupees, into the index, on the same purchase date. Positions that cannot be compared are shown as a count, never quietly discarded.

Stale marks are surfaced, not folded in

AUM is marked to market. Positions with no live price fall back to cost and the stale count sits next to the total, because an AUM that is largely stale is not really a live figure.

Metrics appear only once they are earned

Tracking error, information ratio and beta stay absent from your benchmark view until the daily NAV series is long enough to support them. An early number here would be a guess wearing a decimal point.

The daily run refuses bad data.

Everything your desk sees each morning comes off one chain, and the chain is gated twice before it will act on anything.

  1. 01

    Batch check

    Gate

    If today's market data has not landed yet, the run records that it is waiting and dispatches nothing, rather than acting on yesterday's prices.

  2. 02

    Quality gate

    Gate

    Row counts against the trailing average, and null-close checks. A failed gate quarantines the batch and nothing downstream runs on it.

  3. 03

    Point-in-time capture

    The tradeable universe and index membership are snapshotted daily, because the source tables are overwritten. A day not captured is lost permanently.

  4. 04

    Book NAV

    One day of NAV is appended per client book, which is what eventually earns the risk metrics on your benchmark view.

  5. 05

    Watch rules

    Every due Autopilot rule is evaluated against your own universe and filters, then delivered on the channels you chose.

The point-in-time capture matters more than it sounds. Because the source tables are overwritten every day, a platform that does not snapshot them cannot tell you what the index actually held last year. Every study we show you is run against what was really there at the time.

Watch the whole book. Deterministically.

Autopilot runs eight rule types across your clients every day: level bands, reversal patterns, LTCG windows, concentration, drawdown bands, inactivity, quota guards and saved-view watches. No model is guessing on your behalf. A rule either fired or it did not, and you can see exactly which condition met.

Four ways to show a client you are worth the fee.

Cost drag

What trading friction actually costs the book, in rupees and basis points, from settled trades.

Benchmark excess return

Every client's holdings against an index, position by position, with the uncomparable ones accounted for.

Track and prove

Hit-rate trend and days-to-target from resolved rows, alongside the live progress of the analyses you sent.

Alpha IC scores

Which backtested rule holds the strongest historical edge per stock, behind a liquidity floor and a minimum trade count.

The desk itself, end to end.

Five moves, one system of record, with revenue, billing and broker integrations underneath.

  1. 01

    Onboard the client

    One record holds the mandate, the investable capital and the risk brief. Everything downstream reads from it.

  2. 02

    Build the model basket

    Curate baskets on top of the shared base analytics. Only your clients see them. The curation stays your IP.

  3. 03

    Run the rebalance dry run

    Model against live holdings, per-holding drift, lot-aware sizing, and a full charge and tax estimate before anything is sent.

  4. 04

    Send the recommendation

    Reach many clients in one action. Your registered analyst authors the content; the platform is the delivery rail.

  5. 05

    Keep the record

    Every recommendation and client acknowledgment is captured as it happens, retained in-app for 90 days and archived after.

What this does not do.

It does not place orders. Broker connections are read-only, so holdings and positions flow in and order transmission stays with your existing dealing setup. It does not author advice either: MintX is a technology and analytics platform, and every recommendation that reaches a client is written by your registered analyst.

Bring us a number you do not trust.

Show us a figure your current stack reports and cannot explain, and we will walk you through how ours would arrive at it.

Talk to us