InverseOne

What It Is

InverseOne is a strategy validation service that checks trading rules in reverse. It applies each rule to all 60 Korean large caps in the same way and asks whether the timing of the trades still mattered after costs. Only rules that pass are called "surviving rules", and for each stock the service shows that rule's history and its current state.

The Problem

A backtest can look good simply because the stock happened to be rising during that period. A result from fitting a rule to one stock cannot tell you whether that rule will work in the next period.

InverseOne started from reducing that illusion. The core questions are whether a rule worked regardless of which stock it ran on, and whether the money made from timing can be separated from the money made by simply holding longer.

Who It Is For

• Retail investors who want to check their own rules: people who want numbers on whether a moving-average breakout or an oversold rebound rule is actually worth using
• People who never trusted a backtest: people who saw a good return curve but never traded on it because nothing explained why it was good
• People who keep running a strategy: people who want a validated rule recalculated every day so they can see today's signal

Core Flow

1. Search a stock: picking a stock opens the analysis window, where you set the period, capital and costs.
2. Read the validation: see how the rules that passed the large-cap test behaved on this stock and whether their conditions are on right now.
3. Edit the strategy: change entry conditions or position size in the strategy editor and run it again.
4. Compare with real fills: enter your actual trades to see the gap caused by fill prices, delays and trades that broke the rule.
5. Keep it running: install the automation bot to recalculate the chosen strategy every time a new daily bar arrives.

Key Features

• Large-cap-wide verdict: 118 trading rules in 18 families run across 60 Korean large caps. A rule survives only if it beat holding the same average weight on all three tests: a monthly t-value of 2 or more, positive results in both halves of the period, and positive results on a majority of the stocks checked.
• Per-stock record: net return, drawdown and trades for a surviving rule on your stock, side by side with buy-and-hold and the opposite strategy.
• Today's signal ranking: the home page lists five stocks where surviving rules recently switched on.
• Strategy editor and direct runs: start from a template or a draft written from a sentence, edit the conditions, then pin the strategy version and start date and run it again.
• Automation bot: without checking signals by hand every day, your chosen strategy is recalculated each time a new daily bar arrives. You provide the server it runs on. At this stage it only checks a read-only connection to Toss Securities and runs paper trading; it sends no real orders.
• Public results: if a bot owner opts in, only a pseudonym and ratios such as return and drawdown appear on the public page. Accounts and amounts stay on the owner's screen.
• Market pages: separate pages narrow candidates from public prices, track stock lending balances and share count changes, and show before-and-after edits in filings for 60 Korean stocks.

Strategy validation and the automation bot run in Korean. The market pages are available in Korean, English and Japanese.

Product Decisions

• Don't trust a rule picked on one stock: while refining the validation, I found that rules chosen and tested on a single stock did not predict the next period. So the unit of judgment moved from one stock to all large caps.
• Separate holding from timing: results are compared against holding the same average weight, so gains from a rising market are not credited to the rule.
• Charge for results that fail too: an analysis costs the same even if no rule survives. What was tested and why it failed is also useful for an investment decision. Credits are refunded only when an analysis cannot finish because of missing data or an error.
• No recommendations: every value on screen is a calculation on past data. Each page states that it does not imply future returns and is not a recommendation to buy or sell.
• Real orders come later: the automation bot started with connection checks and paper trading. The service does not store your server password or brokerage keys.

Revenue Model

• Analysis pass: one analysis costs 50 credits ($5). For 7 days after you receive the result, revisiting it or running it directly costs nothing more.
• Korean market day pass: 1 credit unlocks the names of the top three stocks in the day's signal ranking and ranks 11 to 30 of the supply radar. Ranks 4 and 5 and radar ranks 1 to 10 are open to everyone.
• What stays free: strategies the operator publishes, the automation bot, and each day's public indicators on the market pages.

Credits are a balance shared across the Mr. Latte family of services, so credits topped up elsewhere work here too.

Next Hypotheses

• Whether a 7-day pass is long enough to edit a strategy and run it again.
• Whether other users rely on the paper-trading records on the public results page when choosing a rule.
• Whether the free slots in the signal ranking (ranks 4 and 5) work as an entry point to buying a day pass.