hyperc
tabular profit prediction without the optimism bias
Last verified:
What is hyperc?
HyperC P34 is a tabular model for profit targets that turns available deal options into profitable portfolios. It is positioned as a Profit-as-Regression model, meaning it predicts which quantity to choose for each available option and estimates the profit score for that choice. The site presents it as a replacement for naive tabular fitting that can look good on history but be too optimistic in real markets.
The main workflow is built around two actions: predicting trades and fitting on historical trade data. For prediction, you pass feature rows and unique keys, and the model returns the selected quantity and profit estimate per key. For training, it accepts historical features, historical profit, keys, dates, menu identifiers, optional availability flags, and optional exact historical choices.
HyperC describes the product as useful for computable markets where the business must choose among many options under constraints. The examples and markets mentioned include Amazon wholesale reselling, small loans, startup investing, electricity, collectibles, virtual goods, betting markets, used cars, industrial chemicals, construction materials, and retail real estate.
The product is aimed at teams that work with partially observed markets and want portfolio-level decision support rather than simple prediction. The site also emphasizes that it is still early-access software, available upon approval, and intended for technical research, engineering evaluation, and discussion rather than guaranteed live-market performance.
hyperc pricing
Pricing model: Freemium
The website does not list public pricing tiers, a free tier, or published paid plans. It says
hyperc pros
- Predicts best quantity per key
- Returns profit score per option
- Handles portfolio selection
- Supports historical fitting
- Accepts multiple market types
- Designed for partially observed markets
- Works with menu-based choice data
- Uses key-based option grouping
- Supports availability flags per quantity
- Supports exact historical choices
- Targets real-world trade inefficiency
- Focuses on portfolio-level outcomes
- Provides synthetic benchmark guidance
- Offers pre-configured markets
- Covers many computable markets
- Aims to reduce over-optimistic fitting
- Built for decision-time choice sets
- Can model custom grounded markets
hyperc cons
- Requires full historical menu data
- Does not infer missed opportunities
- Designed to be slow
- Not meant for high-frequency use
- Not useful for fully materialized markets
- Not suited to regulated order-book markets
- Available only upon approval
- Limited public documentation
- Quick examples are still TBD
- Advanced mode is still TBD
- Pricing is not published
- No self-serve checkout
- Performance needs independent validation
- Not presented as a live trading system
- Safety and limitations still being assessed
Frequently asked questions about hyperc
What is HyperC P34?
P34 is HyperC’s tabular Profit-as-Regression model for profit targets. It takes available deal options and tries to select the best quantity for each key while estimating the profit associated with that choice.
What does P34 predict?
P34 predicts the best quantity to trade for each key and provides a profit score for that decision. The output is described as a table of items to purchase, with zero used when the model decides not to trade an option.
How do you use P34 for prediction?
For prediction, the site says you feed feature data to predict along with x_keys that identify the separate options. The model then returns a selected dataframe containing the chosen quantity and the predicted profit for each key.
What data is needed to fit the model?
The fitting flow expects historical features, historical profit outcomes, historical keys, dates, menu identifiers, and optionally availability indicators or exact historical choices. HyperC describes the historical menu as the full set of options that were available at decision time.
What is a menu in HyperC’s terminology?
A menu is the list of options available at a specific business decision moment. The site uses menu identifiers to group the options that belonged to the same choice set.
Which markets does HyperC mention?
The preview page mentions synthetic inventory, and says pre-configured markets like amazon will be available. It also lists examples such as electricity, collectibles, virtual goods, betting markets, used cars, industrial chemicals, construction materials, and retail real estate.
What are the main limitations?
HyperC says the full historical menu must be provided and that the model does not speculate about missed opportunities. It also says P34 is slow, is intended for partially observed markets, and has no value in regulated, fully materialized, high-frequency environments.
Is HyperC meant for live trading advice?
No. The terms state that the service is for technical research, engineering evaluation, and discussion only, and explicitly say it is not investment advice, financial advice, trading advice, or an offer of investment-management services.
Is P34 generally available?
The page says P34 is available upon approval and that HyperC is working with select partners. It also points users to request early access, which suggests restricted access rather than open public availability.
How does HyperC position itself versus naive tabular fitting?
HyperC says P34 is intended as a replacement for naive tabular data fitting that can be too optimistic on real-world markets. The explanation says it searches over engineering choices and trains against a survival-style criterion instead of just fitting the past best.