Investment performance and attribution engine
Performance numbers you can prove.
Tickmark calculates portfolio returns, contribution, and attribution inside your own cloud data platform. Your data stays in your account, every figure traces back to its inputs and method, and the same inputs always give the same answer, to the last decimal.
Performance and attribution
Checking Sealed
| Portfolio return | 1.85% | checked |
|---|---|---|
| Benchmark return | 1.70% | checked |
| Attribution against the benchmark | ||
| Allocation | +35 bp | checked |
| Selection | −40 bp | checked |
| Interaction | +20 bp | checked |
| Excess return | +15 bp | checked |
Show the math
- Excess
- 1.85% − 1.70% = 0.15 percentage points = 15 bp
- Effects
- +35 − 40 + 20 = +15 bp
The three effects add up to the excess return, so the attribution explains all of it: allocation and interaction added 55 bp, and selection cost 40 bp. A basis point is a hundredth of a percentage point.
- Calculates
- Returns, contribution, attribution
- Runs in
- Your Snowflake account Databricks coming soon
- Data sent to Tickmark
- None
- Every figure
- Traced to its inputs and method
Solutions
Returns, contribution, and attribution, with the math shown.
Each method is documented and versioned, so every published figure names exactly how it was calculated. A change to a method is a new version, and earlier figures keep theirs. Tickmark covers cash, listed equities, and funds and ETFs, held long or short, in any currency.
-
Performance
Time-weighted returns, calculated from your positions and transactions.
- Daily returns, gross and net of fees
- Month, quarter, and year to date, and since inception
- Contribution by holding and by segment, cash included
- Active and relative return against your benchmark
- Holdings in any currency, measured in each portfolio’s base currency
Coming soon
- Annualized and rolling returns
- Cost, fee, and flow analytics
- Composite returns and dispersion
- Directly held bonds
-
Attribution
Where your active return came from, effect by effect.
- Brinson-Fachler allocation, selection, and interaction
- Long and short positions attributed separately
- Effects linked across periods, with residuals shown
Coming soon
- Currency effect
- Fixed-income attribution
-
Exposures Coming soon
How each portfolio is positioned, from the same validated holdings.
- Weights by any classification
- Long, short, gross, and net exposure
- Active weights against your benchmark
- Concentration
- Duration, yield, and other bond characteristics
-
Risk Coming soon
Statistics from your published return history.
- Volatility and drawdown
- Sharpe and Sortino ratios
- Tracking error and information ratio
- Beta, correlation, and capture ratios
Built into every solution
- Reconciliation with the figures you report today
- Sealed results, controlled publication, and restatement
- An audit trail of every run, check, and publication
- SQL views, a CSV export, and a reference report
Runs on
- Snowflake
- Databricks Coming soon
- Other data platforms On request
Need something else? We’ll scope it, price it, and build it into the product.
Where did the extra 15 bp come from?
Single-period Brinson-Fachler attribution, September 2026. Illustrative figures.
- Total return
- Adds to excess return
- Takes from excess return
Show the math
| Segment | Portfolio weight | Benchmark weight | Portfolio return | Benchmark return | Allocation (bp) | Selection (bp) | Interaction (bp) |
|---|---|---|---|---|---|---|---|
| Technology | 40% | 30% | 5.00% | 4.00% | +23.0 | +30.0 | +10.0 |
| Financials | 35% | 40% | 1.00% | 2.00% | −1.5 | −40.0 | +5.0 |
| Energy | 25% | 30% | −2.00% | −1.00% | +13.5 | −30.0 | +5.0 |
| Total | 100% | 100% | 1.85% | 1.70% | +35.0 | −40.0 | +20.0 |
Scroll the table sideways to see every column.
Allocation = (wp − wb) × (rb − Rb) · Selection = wb × (rp − rb) · Interaction = (wp − wb) × (rp − rb), with Rb = 1.70%. Effects in basis points, hundredths of a percentage point; they sum to +15 bp, the gap between 1.85% and 1.70%.
How it works
From your tables to a published figure, inside your account.
Five recorded steps take a run from your tables to a published figure, so any figure can be traced back through all of them.
-
MAPPED
Step 1: Map
Reviewed configuration maps your source tables to one standard data model. No custom code, and nobody edits the generated SQL.
-
VALIDATED
Step 2: Validate
Checks confirm the delivery is complete and every input is valid before anything is calculated. A failed check holds back only the affected portfolios and dates, and names the rule.
-
COMPUTED
Step 3: Calculate
One SQL engine computes every figure in fixed-point decimals, and every rounding step is declared.
-
SEALED
Step 4: Seal
Results are sealed: recorded with row counts and checksums, and never changed afterward.
-
PUBLISHED
Step 5: Publish
You decide when sealed results become official, and your reporting tools read only official figures. A correction publishes as a restatement, and the earlier figures stay on record.
From Tickmark
- Signed engine bundle
- Signed adapter pack
Verified in your account before any SQL runs
Your Snowflake account
- Source tables read-only
- Mapping adapter pack
- Checks validation
- Engine calculation
- Results sealed
- Published views BI and exports
Audit trail Every run, check, and publication is recorded: who ran it, when, and on which inputs and versions.
Who does what
- Your data team
- Grants read-only access to the source tables, schedules runs, and operates the platform.
- Your performance team
- Owns the methodology, reviews results, and approves what is published.
- Tickmark
- Builds your source mapping, ships each signed release, and supports you through access you grant.
Proof
Same inputs, same answer. To the last decimal.
Replay a run from its recorded source state, with the same mapping and methodology, and every result matches, down to its checksum.
-
Exact arithmetic
Every calculation runs in fixed-point decimals, never floating point, so the same inputs give the same digits in every run and every release.
-
Declared rounding
Every division and multiplication rounds half away from zero at a fixed scale: eight decimal places for amounts, twelve for returns and weights.
-
No hidden inputs
No clock, random value, or session setting reaches a calculation. Every input to a run is recorded with it.
-
Evidence with every release
Each signed release comes with its test results and tested limits, so you see what was proven before anything is installed.
-
Tested against independent answers
Expected test values come from hand calculations and a separate reference implementation, never from the engine under test.
-
Tests that are themselves tested
We break the code on purpose, with a flipped sign or a moved rounding step, and a test must fail every time.
| Run | Started | Result checksum |
|---|---|---|
| Original | 1 Oct 2026, 06:00 | 9f2c41d0…a71b |
| Replay | 3 Nov 2026, 14:22 | 9f2c41d0…a71b |
Identical across all 1,248 portfolio-days
The replay reads the original run’s recorded source state with the same adapter pack and methodology, so it must reproduce every value. Illustrative figures.
Reconciliation
Compare with whatever you report today.
Load the returns you publish now, from your current system or your custodian, and Tickmark compares them with its own, figure by figure. Differences beyond your tolerance are flagged as breaks, your team records the cause of each, and every comparison stays on record in your account. Use it to check an existing system, to run in parallel before you switch, or as a standing control after you go live.
How a reconciliation works
-
Load the figures you report today
They come in through the same mapping and checks as any other input. Tickmark never uses them in its own calculations.
-
Compare like with like
Tickmark compares each reported figure with its own wherever both are defined the same way: same portfolio, date, period, and return basis. A difference larger than your tolerance is flagged as a break.
-
Record the cause
Your performance team classifies each break: different input data, a timing difference, a different calculation method, or a reported figure that doesn’t follow its own stated method.
-
Control publishing
If you choose, no figure is published until its breaks are resolved or signed off by a second person, never the one publishing.
When corrected figures arrive, Tickmark compares them again. Earlier comparisons stay on record as evidence.
- Compared
- 1,248
- Matched
- 1,241
- Breaks
- 7
| Portfolio | Date | Tickmark | Reported | Difference (pp) | Result |
|---|---|---|---|---|---|
| GROWTH-01 | 30 Sep | 1.0000% | 1.0000% | 0.0000 | Match |
| INCOME-02 | 30 Sep | 0.4213% | 0.4213% | 0.0000 | Match |
| GLOBAL-03 | 29 Sep | −0.2871% | −0.2650% | 0.0221 | Timing break |
| VALUE-04 | 30 Sep | 0.7402% | 0.7402% | 0.0000 | Match |
| ASIA-05 | 30 Sep | 1.1290% | 1.1356% | 0.0066 | Data break |
Illustrative figures. Differences are absolute, in percentage points (pp). Scroll the table sideways to see every column.
Security and control
It runs in your account. Nothing comes back to Tickmark.
Tickmark installs as signed SQL in a database you own. It calls no outside service, needs no credentials from you, and sends nothing back.
-
Your account, your data
Inputs, calculations, and results stay in your own account. Tickmark sees nothing unless you grant a time-limited support role.
-
Nothing phones home
No telemetry and no license checks at runtime. License usage is counted from the run records in your own account.
-
Signed releases
Engine bundles and adapter packs are signed and checksummed, and every engine release lists its components. Installation verifies them before any SQL runs.
-
Least privilege
Nine separate roles, from installer to reader. Read-only grants on exactly the source tables your mapping reads.
-
Results that can’t be rewritten
No Tickmark role can change a sealed result or delete a published one. A correction is a new version beside the original.
-
Support on your terms
Support access is a role you grant: time-limited, audited, and read-only.
Everything stays in your database.
Tickmark keeps nothing on its side. Each run keeps the validated inputs it calculated from, its sealed results, and its audit trail in your database, so any figure can be traced and replayed. Maintenance removes only what nothing depends on, once it passes a cutoff you set.
- CORE Validated input snapshots While a retained run uses them
- RESULT Sealed results Always, once published
- RECON Comparisons and breaks Always
- CTRL Audit trail and controls Always
- API Published views Views only
Pilot
Shadow first. Go live when the numbers agree.
A paid pilot runs in your account, beside the figures you report today. It ends with a signed reconciliation report and a mapping you keep, so nothing is rebuilt for production.
-
Assess
We review your data with you: positions, transactions, and benchmark data, the history the pilot covers, and anything we should source or load for you.
-
Map
We configure the mapping of your source tables. You keep it: the same mapping carries into production.
-
Shadow
Tickmark runs for three month-ends beside the figures you report today. Every return is compared, and every break is traced to its cause.
-
Go live
Once you sign off the reconciliation and a rehearsed correction, you publish from Tickmark, with runbooks for scheduling, restatement, rollback, and upgrades.
What you get
- A signed reconciliation report against the figures you report today
- A mapping of your source tables that carries into production
- A methodology document for every method version
- A data dictionary for every field Tickmark reads
- Runbooks for scheduling, restatement, rollback, and upgrades
- Signed releases, each with its test evidence
What we work with
- Your positions and transactions, in Snowflake. Held elsewhere? We can load them into Snowflake for you.
- Daily benchmark returns, and for attribution, segment weights and returns, under your index license. Not held daily? We can source them from your index provider.
- A data owner who can grant read-only access to your source tables.
- A performance owner who can confirm your methodology and approve the reconciliation.
Sourcing or loading data for you is quoted in your proposal, with third-party costs passed through to you.
Pricing
Priced by solution and portfolio.
An annual license for the solutions you use, and a one-time implementation for your source systems. There are no seat or query fees, and Tickmark runs on your own Snowflake compute, billed by Snowflake as usual.
-
Solutions and portfolios
Your annual license is priced on the solutions you use and the number of portfolios they cover.
-
Analytics, not asset classes
Specialized analytics, such as fixed-income attribution, are priced as solutions of their own. What your portfolios hold never changes the price.
-
Implementation by source system
Mapping work depends on how many source systems you have and how they are structured, so implementation is priced on that, once.
-
Your pilot carries forward
The paid pilot is your first implementation. Its mapping goes into production, so you never pay for it twice.
-
Counted in your account
Usage comes from the run records in your own account. Nothing is metered, enforced, or reported to Tickmark at runtime.
-
Third-party costs passed through
Snowflake bills your compute and storage directly. The cost of data we source for you, such as daily benchmark returns, is passed through to you.
FAQ
What teams ask before a pilot.
Something else? Ask us directly.
Does Tickmark receive our portfolio data?
No. The engine sends no portfolio data, logs, or telemetry to Tickmark, and Tickmark operates no service that could receive them. If you want Tickmark’s help with an incident, you grant a support role that is time-limited, audited, and read-only, and revoke it when you’re done.
Do we need an existing performance system?
No. Tickmark calculates from your positions, transactions, and benchmarks, and never needs another system’s figures. If you do have one, or your custodian reports returns, the pilot compares against it, and you can keep that comparison as a standing control for as long as you like.
Which holdings and strategies does Tickmark cover?
Cash, listed equities, and daily-valued funds and ETFs, including bond and money-market funds, in any currency. Portfolios can be long-only or long-biased long/short, such as 130/30, with physical short positions. Directly held bonds are coming soon.
What does Tickmark include, and what comes next?
Daily time-weighted returns, gross and net of fees, with month, quarter, and year to date and since inception; contribution by holding and segment; active and relative return against your benchmark; and Brinson-Fachler attribution, with long and short positions attributed separately. Reconciliation, publication control, and the audit trail are built into every solution. Coming soon: annualized and rolling returns, return statistics, exposures, cost, fee, and flow analytics, composites, the currency effect, and fixed income, from directly held bonds to fixed-income attribution.
Which platforms do you support?
Snowflake. Databricks is coming soon, and other data platforms are available on request: the calculations are shared, and each platform gets its own tested implementation, held to the same results to the last digit. If your data lives in another system, we can also load it into Snowflake for you.
What does Tickmark work from?
Your positions, transactions, and exchange rates from your book of record, and your benchmark data, all in Snowflake, with a data owner and a performance owner on your side. If something isn’t in place, such as daily benchmark data or a load into Snowflake, we can source or build it for you; it is quoted in your proposal, with third-party costs passed through to you.
Can you build something we need that isn’t listed?
Yes. Tell us what you need, and we’ll scope and price it. It is built into the product for every customer, never as a custom version, so it stays in every future release.
How do you match our current methodology?
We document it with your performance team and set up a methodology version from the conventions Tickmark supports, such as flow timing, which fees each return basis excludes, your reporting periods, and your classification schemes. Then we reconcile, and every remaining difference is explained as data, timing, methodology, or a reported figure that doesn’t follow its own stated method.
What happens when source data is corrected?
Published figures that depend on the changed input are flagged as stale. A restatement recalculates the affected portfolios and dates, with the later periods that build on them, and publishes them as a new version. The original figures stay on record, and a rollback restores them.
Do we need to run dbt or Python?
No. We build and test Tickmark with dbt and Python, but neither ships. You install signed SQL and call stored procedures from the scheduler you already use.
What if Tickmark stops operating?
Your installed engine keeps running. It has no dependency on any Tickmark service, repository, or network, and the SQL lives in your account. Production contracts include continuity terms.
Get started
Request a demo.
Tell us a little about your setup. We’ll walk you through Tickmark on a sample portfolio, answer your questions, and quote for the solutions and portfolios you need.
- A walkthrough of Tickmark on a sample portfolio
- Answers on scope, platforms, and pricing
- A pilot proposal on your own data, when you’re ready
Demo request sent.
Thanks. We’ll reply to your work email to arrange it.