Oddup market-intelligence platform
1–100
One comparable score, from many incomparable signals
34
Cities covered, by Oddup's published figures
$19.9M
Raised to build the data business
A rating is only worth anything if it means the same thing this quarter that it meant last quarter.
Oddup made its name doing for startups what rating agencies do for bonds: a 1-to-100 'Oddup Score' assessing the investability of a company, later expanding into indexes and data products across 34 cities worldwide. TrivialWorks engineers worked inside the team as an offshore engineering partner.
The problem domain
A ratings business sells a single number, which makes it look simple and is the reason it is not. Behind the number sits everything that had to happen for it to be defensible: signals gathered from sources that disagree with each other, entities resolved so that two spellings of one company do not become two companies, gaps filled or honestly left empty, and a methodology applied consistently enough that a score can be compared with the score of a different company, in a different city, computed six months earlier.
The hard constraint is that the customer is an investor. An investor will act on the number, and then — if it is wrong — will ask exactly how it was produced. So the system has to be able to explain itself: which inputs moved a score, when, and why. That requirement reaches all the way back into the ingestion layer, because a pipeline that overwrites yesterday's view of the world cannot answer the question at all.
What that demands of the engineering
Ingestion that assumes the sources are wrong
Market data arrives late, incomplete, duplicated and occasionally contradictory. The engineering problem is less 'fetch it' than deciding what to believe: reconciling entities across sources, tracking provenance so a figure can be traced to where it came from, and treating a missing value as a fact about the world rather than a zero.
Scoring that stays stable while the inputs move
- Methodology changes have to be versioned, or history stops being comparable with the present.
- Recomputation has to be cheap enough to run often, because a stale rating is a wrong rating.
- Every published score needs an audit trail back to the inputs that produced it.
- Analyst judgement has to be expressible in the system, not applied on top of it in a spreadsheet.
Presentation that respects the reader's time
Indexes tracking real-time trends, dashboards and reports across 34 cities — the surface where the whole pipeline either earns trust or loses it. A data product is judged on whether the number is there when the reader looks, loads quickly, and matches what the same reader saw yesterday.
Our role
Embedded senior engineering inside Oddup's team — working their process and their repositories, delivering on a data-heavy product where ingestion, scoring and presentation all have to keep pace with markets that move daily.
Most of what we build is under NDA — yours would be too, unless you told us otherwise. Send the requirement and you get back a functional specification, at no charge.
Student transport · United States
Student transport development: two apps, two portals and an automated trip builder
Parent and driver apps, school-district and franchise portals, and an automated trip builder that routes and prices shared trips across several school districts — apportioned to the cent.
Read
Language services · Canada
Language access development: an interpreter on demand, in 230+ languages
The application suite and client portals behind an on-demand phone and video interpretation network — 230+ languages, 24/7, run by the non-profit sector.
Read