ROI, yield and hit rate are not the same result
Three familiar numbers can describe very different systems. Here is how price, sample size and a complete ledger change their meaning.
Hit rate ignores the price paid
Hit rate is wins divided by settled decisions. Winning 70% sounds stronger than winning 55%, but it says nothing about whether the average price was 1.20 or 2.10.
A high hit rate can lose money when prices are too short. A lower hit rate can be profitable at longer prices. That arithmetic is why probability and price cannot be separated.
ROI needs a declared denominator
For a flat one-unit ledger, ROI is net unit profit divided by total units staked. Some products use yield for the same concept; others use ROI against starting bankroll. A serious report states the exact formula rather than relying on the label.
Void and withdrawn decisions need explicit treatment. Excluding losses, stale-price failures or unavailable selections after the fact turns a clean-looking metric into selection bias.
Small samples produce loud numbers
A few long-priced wins can dominate early ROI. Publish settled count, average odds, drawdown, probability calibration and closing-price coverage beside the headline number.
- Use one immutable ledger.
- Show the staking assumption.
- Do not annualise a short sample.
- Separate research and verified cohorts.
- Retain corrections and withdrawals.
Sources
Direct links are preserved so the editorial reasoning can be checked independently.